Materials Science and Engineering A 402 (2005) 294–306
Phase formation and microstructure evolution in laser rapid forming of graded SS316L/Rene88DT alloy X. Lin, T.M. Yue ∗ The Advanced Manufacturing Technology Research Centre, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hung Hom, Hong Kong Received in revised form 18 April 2005; accepted 11 May 2005
Abstract The graded material of the SS316L stainless steel/Rene88DT superalloy was fabricated using laser rapid forming. A linear compositional gradient, from 100% SS316L stainless steel to 100% Rene88DT superalloy was achieved. The dendritic growth of the ␥ phase, the precipitation of the ␥ precipitate, as well as the formation of the phase and the carbides along the compositional gradient zone were investigated. The growth morphology of primary ␥-dendrites can be predicted by the microstructure selection models based on the maximum interface temperature criterion. The precipitation of the ␥ phase and phase was to a large extent governed by the concentration of Ti and Al elements as well as by the Ti/Al ratio in the gradient zone. The size of the ␥ phase along the gradient zone can be explained by the LSW theory. In the deposit, apart from MC carbides, some M6 C carbides were obtained, and there was no evidence of M23 C6 carbides. This differed from the predictions of the phase diagram. It was interesting to find that the MC carbides exhibited a cored structure. The formation of the cored structure as well as the M6 C carbides was discussed. © 2005 Elsevier B.V. All rights reserved. Keywords: Laser rapid forming; Functionally graded materials; Stainless steel; Superalloy; Precipitates
1. Introduction Over the past years, significant advances have been made in the development of new superalloys that are capable of operating at higher service temperatures, with the aim of achieving better engine efficiency. To cope with a more severe service environment, superalloys must possess excellent strength at elevated temperatures, outstanding creep, fatigue and impact resistance, as well as adequate resistance to hot corrosion [1]. To this end, perhaps, it should be mentioned that for most hot-end components, only some regions of these components will encounter extreme hightemperature environments, and they therefore need not be made of mono-composition superalloys. Rather, the use of functionally graded materials (FGMs) could be more appropriate. In so doing, the cost of materials and the final density ∗
Corresponding author. Tel.: +852 27666601; fax: +852 23625267. E-mail address:
[email protected] (T.M. Yue).
0921-5093/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2005.05.024
of the component could be reduced. For example, if Ni-based mono-superalloy components can be fabricated by FGMs of a Fe-based/Ni-superalloy or a Ti-based/Ni-superalloy, according to the distribution of the working temperature, then considerable savings in material costs and weight can be achieved. Traditional processing methods for making FGMs, such as powder metallurgy and thermal spraying [2,3] can only produce objects with simple geometry, and both the compositional and functional gradients cannot be easily tailored. Although powder consolidation processes can produce dense FGMs, the product normally requires subsequent treatment, such as hot isostatic pressing (HIP) to improve its mechanical properties. To prevail over these deficiencies, a new solid freeform fabrication technology, namely, laser rapid forming (LRF) has been developed to directly fabricate bulk near-netshape metallic based components. LRF works on the principle that after the CAD model of the component is constructed and sliced electronically into a sequence of layers that define
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
the regions that compose the component, a metal component is then fabricated directly by laser multilayer cladding. Since this technique has many outstanding advantages, e.g. a metal component can be fabricated rapidly without using a mold, many methods using a similar principle of fabrication have been developed [4–9]. It should be noted that most of the present research on LRF concentrates on homogeneous materials [4–9]. In fact, the forming characteristics of LRF demonstrate that if the information on local elemental constituents is coupled with the geometry of the layers, which drives the laser-forming path, then fully dense, metallurgically bonded freeform functionally graded materials can be produced. The freedom of selectively cladding different elemental powders or premixed blends at discrete locations and the employment of multiple powder feeder systems make the fabrication of functionally graded materials possible. In fact, a number of studies have been conducted on the laser rapid forming of functionally graded materials [10–22]; however, only a few studies concerning superalloys can be cited in the open literature [19–22]. Kahlen et al. [19] employed the laser deposition of metal layers to create graded materials by varying the composition of the parts from 100% SS304 stainless steel to a 100% nickel-based superalloy. He briefly studied the effects of the solidification rate on the mechanical properties of the graded materials. As for process development, Griffith et al. [20,21] developed precise multiple-powder feeding capabilities for the laser engineered net shaping process (LENSTM ). The process has been used to fabricate graded or layered parts of stainless steel SS316/Inconel 690 and Ti6Al4V/Inconel 718 materials. Notwithstanding the contributions from these studies, our understanding of the solidification behavior of Ni-based FGMs is still far from satisfactory. Our previous paper [22] reported that graded material of stainless steelSS316L/superalloy-Rene88DT was successfully fabricated using laser rapid forming. A linear compositional gradient, from 100% SS316L stainless steel to 100% Rene88DT superalloy, was achieved within a thickness of 40 mm of the laser multilayer deposition. The phenomena of columnarto-equiaxed transition and continuous epitaxial columnar growth were also investigated. To further the investigation, the dendritic growth of the ␥ phase, the development of the ␥ precipitate, as well as the formation of the phase and carbides along the gradient zone was studied, and the results are presented in this paper. This study is considered to be valuable because most age-hardenable Ni-based superalloys derive their good mechanical properties from the dispersion of the ordered face-centered cubic ␥ precipitates in the ␥ matrix [23,24].
295
2. Experimental procedures The compositional graded material was fabricated using a laser rapid forming system consisting of a 5 kW continuous wave CO2 laser from Rofin Sinar, a four-axis numerical control working table and a powder feeder with a lateral nozzle. The experiment was conducted inside a glove box, whose atmosphere was controlled. The laser was mounted on an overhead carriage and the beam was directed into the glove box through a window on top of the chamber. The controlled atmosphere glove box was filled with argon gas, and argon gas was also used to deliver the metal powders to prevent the melt pool from oxidizing and oxide contamination from occurring during processing. The laser beam was directed onto the substrate to create a molten pool into which the premixed powders were injected through the laser nozzle. The metal powders were melted and subsequently re-solidified to form the clad layer. A solid structure with a rectangular profile was fabricated, with its first 25 layers (∼10 mm) composed of 100% 316L stainless steel. The composition of the deposition was then changed linearly from 100 316L to 100% Rene88DT over the next 100 layers (∼40 mm). Finally, an additional 10 layers (∼4 mm) of 100% Rene88DT superalloy were deposited. The variation in composition along the height of the solid structure was achieved by the in situ adjustment of the ratio of the volume of 316L stainless steel to the Rene88DT superalloy of the premixed powder according to the predetermined graded structure. The nominal compositions of SS316L stainless steel and Rene88DT superalloy powders are listed in Table 1. The processing parameters are presented in Table 2. The substrate material used for the experiment was cold rolled 316L stainless steel sheet. The surface of the substrate was cleaned by sandblasting prior to laser cladding. In order to eliminate any water that was trapped in the powders, the powders were dried in a vacuum oven for 24 h. The composition and the microstructure along the gradient direction were characterized using a Leica Stereoscan 440 scanning electron microscope (SEM) equipped with the facility of energy disperse X-ray analysis (EDS). The microstructure was also studied by the JEOL JEM-2010 transmission Table 2 The laser processing parameters Laser power (kW) Scanning velocity (mm/s) Spot diameter (mm) Powder feeding rate (g/min) Flow of shielding gas (l/min)
2.0–3.3 5–10 4 8–12 4–8
Table 1 The chemical composition of the powders (wt%)
316L Rene88DT
Cr
Co
W
Ti
Al
Nb
Mo
Si
Mn
Fe
Ni
C
16.8 16.5
0.3
Bal
3.9
3.7
2.3
0.7
2.2 4.2
0.75
13.3
13.8 Bal
0.03 0.04
296
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
electron microscopy (TEM). Phase identification was performed using the Philips Xpert X-ray diffractometer (XRD) system.
3. Results and discussion 3.1. Composition gradient Fig. 1 shows the results of the EDS analysis along the vertical direction of the graded deposit. The main elements of SS316L and Rene88DT alloys, i.e. Fe, Ni, Cr, Co, exhibited a good linear gradient along the deposition direction in the graded zone. The other elements, such as Mo, Ti, Al, W and Nb, also showed a reasonably good linear relationship. Based on the measured composition, the equilibrium liquidus temperature, the solidus temperature and the ␥ phase solvus temperature along the compositional gradient were calculated using Thermo-Calc software with the aid of the TTNI superalloy database [25]. The results (Fig. 2) show that at a
Fig. 2. The calculated equilibrium liquidus, solidus and ␥ phase solvus temperatures.
distance of about 35 mm from the substrate, where the composition of the alloy was 40% SS316L + 60% Rene88DT, the equilibrium freezing range (T0 ) of the alloy reached a maximum value of 103 K. In addition, the liquidus temperature of the alloy decreased as the amount of Rene88DT increased, while the solvus temperature of the ␥ phase increased as the amount of Rene88DT increased. Fig. 3 shows the calculated fraction of the equilibrium ␥ phase present in the direction of the compositional gradient at 600 ◦ C. The amount of the ␥ phase increased as the amount of Rene88DT increased. The increase in the solvus temperature of the ␥ phase and the amount of ␥ phase along the compositional gradient means that a higher serviceable temperature can be tolerated. 3.2. Dendritic growth of the γ phase By employing the Thermo-Calc software, the equilibrium phase diagrams of SS316L and Rene88DT are calculated and are presented in Fig. 4a and b, respectively. According to these
Fig. 1. The measured compositional gradient (a and b).
Fig. 3. The calculated volume fraction of the equilibrium ␥ phase at a temperature of 600 ◦ C along the compositional gradient.
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
297
for planar front growth: Tp = Tm +
n−1
∗ mvi Cpi −
i=1
R g Tm V V0 Sf
(1)
and for cellular and dendritic growth: Tc/d = Tm +
n−1 i=1
Fig. 4. Phase diagrams of (a) SS316L and (b) Rene88DT simulated by Thermo-Calc.
phase diagrams, stainless steel SS316L would solidify as a duplex austenite (␥) + ferrite (␦) structure; while superalloy Rene88DT will solidify as a single-phase ␥-solid solution. It should be noted that the as-solidified structure may deviate appreciably from that predicted by the equilibrium phase diagram during rapid solidification. In fact, several studies have found that the high-temperature gradient, high cooling rate and high solidification growth rate that prevailed during high-power welding processing, such as electron beam or laser joining, can cause a change in the solidification mode of the 300 series stainless steel, from primary ␦ ferrite to primary ␥ austenite [26]. In order to determine the solidification mode of 316L in laser rapid forming, microstructure selection models [27,28] based on the maximum interface temperature criterion were applied to the present study. Using these models, the growth temperature for the different solid/liquid interface morphologies of each phase can be calculated, both
mvi Ci∗ −
¯ R g Tm 2Γ GD − V− , R V0 Sf V
(2)
where Tm is the melting temperature of the pure component, ∗ = C /k the composition of the liquid at the interface, Cpi 0i vi kvi is a velocity dependent distribution coefficient [29], mvi the velocity dependent liquidus slope defined by Boettinger et al. [30], Rg the gas constant, Sf the molar entropy of fusion, V0 the limit of crystallization, which, for pure components, has an upper limit in the order of the velocity of sound. At steady state, Tp corresponds to the solidus temperature of the composition C0i . Γ is the Gibbs–Thomson coefficient and R is the radius of the dendrite tip. Ci∗ can be evaluated using the Ivatsov solution, Ci∗ = C0i /1 − (1 − kvi )Iv(Pci ) [27]. Here, Pci is the solutal Peclet number. The subscript i indicates each elemental species. G is the temperature gradient, V the ¯ is the average diffusivity of the solidification velocity and D solutes in the liquid. The results of the above calculations can be presented in the form of diagrams shown in Fig. 5, where the solidification front temperature is plotted as a function of the growth rate for the competing of the austenite (␥) and ferrite (␦) phases. The thermal gradient for laser rapid forming was calculated by solving the heat diffusion equation with a modified Rosenthal approach [31–33]. The weight average temperature gradient calculated for the present case is 8 × 105 K/m. From the results, it is apparent that the solidification front temperature for the growth of ␥ is always higher than that for the ␦ phase, and that the growth of single-phase ␥ is favored in the laser rapid forming of SS316L. Meanwhile, Fig. 1a shows that during the laser rapid forming of the SS316L/Rene88DT
Fig. 5. Calculated liquid–solid interface temperature vs. the growth velocity for different growth morphologies of various phases in 316L stainless steel.
298
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
Fig. 6. XRD patterns of sections perpendicular to the gradient direction of different amounts of Rene88DT.
graded material, increasing the amount of Rene88DT caused the value of [Cr/Ni]equivalent to gradually decrease. As a result, the tendency for ␥ to form was further increased and singlephase ␥ dendritic growth could occur within the entire FGM of SS316L/Rene88DT. Fig. 6 shows the results of the XRD analyses conducted on the cross-sections perpendicular to the gradient direction of the LRF FGM deposit. In the SS316L substrate, there exist two phases, ␥ and ␦, and their directions of growth are in the 1 0 0 and 1 1 1 crystallographic orientations, respectively. On the other hand, the ␥ phase with a strong directional growth in the 1 0 0 crystallographic direction was obtained in the whole FGM deposit, including the SS316L layers deposited by LRF. This shows good agreement with the predictions, as shown in Fig. 5. Fig. 7 shows the typical microstructure in cross-sections parallel to the gradient direction. It is well known that the 1 0 0 crystallographic directions are preferred growth directions for cubic metals. In most cases, the dendrite stalks lie between the direction of the temperature gradient and the 1 0 0 directions. During laser rapid forming, the highest temperature gradient is at the bottom of the molten pool, and is perpendicular to the laser beam scanning direction, i.e. along the compositional gradient direction. It is at the bottom of the molten pool where solidification is first initiated, so most of the dendrites will grow in the 1 0 0 crystallographic direction along the compositional gradient during LRF. This also ensures continuity in the growth of the microstructure in the entire FGM deposit. 3.3. The precipitation of the γ phase
Fig. 7. Microstructure of parallel sections to the gradient direction of composition (a) SS316L, (b) 50% SS316L + 50% Rene88DT and (c) Rene88DT.
As a precipitation-strengthening superalloy, the Rene88DT obtains its strength primarily from the presence of a large amount of Ni3 (AlTi)-type precipitates (␥ ) in the nickel-based fcc matrix (␥). In comparison with SS316L, Rene88DT has the addition of Co, W, Ti, Al and Nb. These elements favor further solid solution and the formation of
the ␥ phase. Fig. 8 shows the distribution of the ␥ phase along the composition gradient in the deposit; whereas, Fig. 9 shows some typical TEM photographs of the ␥ phase precipitates. It was observed that, the ␥ phase was found in the interdendritic regions when the content of
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
299
Fig. 8. Showing the distribution of the ␥ phase in (a) 40% Rene88DT, (b) 60% Rene88DT, (c) 80% Rene88DT, (d) Rene88DT (4 mm distance from the top) and (e) Rene88DT (1 mm distance from the top).
Rene88DT reached a level of about 40%. The formation of the precipitate was due to the eutectic reaction of the residual liquid. When the amount of Rene88DT reached about 60%, the precipitation of the ␥ phase become apparent within the dendrite arms. It appears that the decomposition of MC to M6 C by the reaction MC + ␥ → M6 C + ␥ also promotes the precipitation of the ␥ phase. Fig. 9a shows that relatively coarse ␥ precipitates were formed at the edge of a M6 C particle. Table 3 presents the results of the EDS analysis
of the MC carbides, which are present in the section of the deposit where the composition lies between 40 Rene88DT and 100% Rene88DT. The results show that the MC carbides were mainly TiC or (Ti, Nb)C with a high concentration of Ti, which is a ␥ -forming element. So, it is believed that the decomposition of MC will promote the formation of the ␥ phase. In addition, it was found that the amount and the size of the ␥ phase (Fig. 8a–d and Fig. 9a–c) both increased with an increase in Rene88DT. The variation in the size of the
300
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
Table 3 The EDS results of MC carbides (at%) Al 1 2 3 4
Periphery Core Periphery Core Periphery Core Periphery Core
Ti
Nb
99.11
0.89
98.89 64.78 91.76 21.73 92.63 52.19 61.96 97.24 86.95
1.11
34.23 72.09 44.92
0.73 0.19 1.21 27.23 0.62 6.61
␥ precipitate along the compositional gradient is shown in Fig. 10. Based on the measured composition of the alloy, the calculated equilibrium solute partition coefficients of Ti and Al (kTi , kAl ) along the composition gradient were obtained and are presented in Fig. 11a. It is apparent that kTi is less than unity in the entire gradient zone, and that kAl is greater than unity only when the amount of Rene88DT is beyond 95%. This means that a considerable amount of ␥ -forming elements (Al, Ti) will segregate to the interdendritic regions, and that the dendrite arms themselves will be depleted in these elements. Thus, the composition of the liquid at the interdendritic regions will first reach the composition for the precipitation of ␥ as the amount of Rene88DT is increased. Fig. 11b provides the calculated Gibbs free energy differ␥−␥ ence between the ␥ phase and ␥ phase (GV ) at the ␥ solvus temperature along the composition gradient. The ␥−␥ increases as the results show that the value of GV amount of Rene88DT increases. Based on the classical nucleation theory [34], the homogeneous nucleation rate can be expressed by: Q GC I = I0 exp − exp − , (3) kB T kB T where I0 is a constant, kB is Boltzmann’s constant, Q the activation energy for diffusion, GC the change in free energy associated with the formation of a critical nucleus, which can and be written as: 3 16πσ␥−␥ ␥−␥
3(GV
− G )
Cr
Ni
0.85 4.57 0.45 2.06 1.85 5.99 1.36 4.49
0.13 2.94 0.54
Fe
100
5 6 7
GC =
Co
,
(4)
where G is the strain energy change between the ␥ and ␥ states, and σ ␥ − ␥ is the interfacial free energy of the ␥ − ␥ interface, which can be divided into the chemical and elastic components. Since the ␥ precipitate is coherent with the ␥ matrix, and their lattice constant is also similar, thus making the value of σ ␥ − ␥ and G very small, the nucleation rate of ␥−␥
␥ phase therefore depends onGV
to a large extent. With ␥−␥
reference to Eq. (3), an increase in GV by increasing the amount of Rene88DT means that the nucleation rate of ␥
1.42
1.70 1.04 3.40 0.78 0.60
1.35
phase will increase and, subsequently, that more ␥ precipitates will form. With regard to the morphology of the ␥ precipitates, this evolves from the mechanisms of: (i) competitive coarsening, in order to reduce the specific area of the ␥ − ␥ interface, known as Ostwald ripening [35–37]; (ii) shape changes, in order to minimize the sum of the interfacial and elastic interaction energies. It has been observed that in general the precipitates grow at an almost constant volume fraction, following a (time)1/3 power law of diffusion-controlled particle coarsening. This is in agreement with the LSW theory of Lifshitz and Slyozov [38] and Wagner [39]. In fact the LSW theory gives the following expression: d 3 − d03 =
8σCe Vm2 D t, 9Rg T
(5)
where d0 and d are the average particle diameters of the precipitation phase at the onset of the coarsening process and at the time t after the onset, σ the matrix-precipitated phase interface specific free energy, Ce the matrix concentration of the precipitating elements in equilibrium with a flat surface, Vm the precipitate molar volume and D is an effective diffusion coefficient that mainly depends on the diffusion coefficients of the precipitating elements in the matrix. T is the absolute temperature. However, the LSW theory is applicable only when the precipitation volume fraction is small and approaches zero. In light of this, Voorhees and Glicksman [40] modified the LSW theory using a new statistical mean field theory that accurately reproduces the simulation data of the multi-particle diffusion problem and gives d 3 − d03 =
8σCe Vm2 D 1/3
9Rg T (1 − fv )
t,
(6)
where fv is the volume fraction of the precipitated phase. It is apparent that the size of the precipitate depends not only on the coarsening time, but also on the volume fraction and the matrix concentration of the precipitating elements. The rapid cooling rates of the LRF process will prevent the volume fraction of the ␥ phase from reaching the amount as indicated by the equilibrium diagram. Nonetheless, the already deposited layers will experience a reheating effect by the subsequent
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
301
Fig. 10. Size of the ␥ precipitate as a function of the distance from the substrate.
Fig. 9. TEM micrographs of the ␥ precipitate in (a) 60% Rene88DT, (b) 80% Rene88DT and (c) Rene88DT. ␥−␥
Fig. 11. The calculated values of (a) kTi , kAl and (b)GV position gradient.
along the com-
302
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
laser passes, which will result in the further precipitation and coarsening of the ␥ phase. But, towards the end of the deposition, the total reheating time will decrease; this means that the coarsening time for the ␥ precipitates will decrease. This could be the reason why the size of the ␥ phase did not increase in the pure Rene88DT zone as it grew towards the end of the deposit. On the other hand, the size of the ␥ phase increased in the gradient zone (Fig. 10), despite the fact that the coarsening time decreased towards the top of the deposit. This is believed to be due to the increase in the concentration of ␥ -forming elements (Al, Ti) in the matrix and to the increase in the volume fraction of the ␥ phase as the amount of Rene88DT increased. In Fig. 9a, it was observed that the size of the ␥ precipitates in the interdendritic regions was larger than those present within the dendrite arms. This difference in size is considered to be caused by solute segregation during dendritic solidification. Using a modified-Scheil model [41], the calculated solute concentration in the ␥-dendrites as a function of the solid fraction was calculated and is presented in Fig. 12. The results clearly show that dendrite segregation will cause the enrichment of the ␥ -forming elements (Al, Ti) to occur in interdendritic regions and, according to Eq. (5) a larger precipitation size is expected there. 3.4. The precipitation of the η phase It was reported that NiFe-based superalloys, which are strengthened by ␥ precipitates, are susceptible to the formation of the HCP phase (Ni3 Ti) [23]. In general, the phase can form during forging/heat treatment or from the exposure of the material to elevated temperatures for a prolonged period of time during service. Two forms of the phase may occur: (a) intragranular platelets that form by way of the ␥ – transformation, which sometimes appear in Widmanstatten form and (b) the grain boundary cellular form. To date, detailed research on the formation of the phase during laser rapid solidification is still lacking.
Fig. 12. Solute concentration in ␥-dendrites as a function of the solid fraction for 60% Rene88DT.
Fig. 13. XRD patterns of the sections perpendicular to the gradient direction.
The results of the XRD patterns show that the phase was present in the 45–70% Rene88DT alloys (Fig. 13). On the other hand, the TEM analysis confirmed that phase precipitates also appeared in the 80% Rene88DT alloy; this was not revealed by the XRD results. This could be because the amount present was below the detection limit of the XRD technique. Figs. 1 and 14 show that the composition range of 45–80% Rene88DT is where the relatively large freezing range and the high ratio of Ti to Al lie. In fact, a high Ti/Al ratio will favor the nucleation and growth of the phase; the large freezing range will allow more time for phase precipitates to develop. Based on the results of the XRD and TEM analyses and the information of Fig. 12, the precipitation of the phase was found to have occurred when the Ti/Al ratio was greater than 2.5, which agreed with the published data [23]. Furthermore, with the increase in Rene88DT, the Ti/Al ratio further increased until the composition reached 65% Rene88DT, then decreased until pure Rene88DT was formed. It should be pointed out that although the Ti/Al ratio was
Fig. 14. The Ti/Al ratio along the compositional gradient direction.
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
still greater than 2.5 when the percentage of Rene88DT was beyond 80%, the content of aluminum was high. The enrichment of aluminum will hinder the nucleation and growth of the phase, since this phase has little solubility of Al, and for the phase to be developed, Al needs to be diffused away. Moreover, an excessive amount of Al will decrease the mismatch of the ␥ − ␥ lattice, hence reducing the driving force for the transformation of ␥ → [23]. Besides, the diminishing post-deposition heating effect occurring towards the end of LRF may bring to an end the development of the phase. Fig. 15 shows some typical TEM photographs and the SAD pattern of platelet precipitates. The phase has the crystal habit plane of the {1 1 1}␥ and the following crystalline orientation relationship with the ␥ matrix: {1 1 1}␥ //(0 0 1) ;
1 1 0 ␥ //[0 1 0] .
Such relationships make it easy for the intragranular platelets of phase to nucleate at the vicinity of the ␥ grain boundaries and to grow in the form of clusters. These micrographs show
Fig. 15. TEM micrographs showing cellular precipitates in (a) 70% Rene88DT and (b) 80% Rene88DT.
303
that some intragranular platelets of the phase grew with a high degree of parallelism (Fig. 15a), while others were initiated next to M6 C particles, and often intersected at different orientations (Fig. 15b). The nucleation of the phase next to the M6 C particles is believed to have been caused by the enrichment of Ti from the decomposition of MC when forming M6 C particles. Fig. 15 also shows that the interlamellar spacing of the phase is about 20–40 nm, which was much finer than the 500–100 nm normally obtained from conventional heat treatment conditions. The refinement of the phase should reduce the effect of its detrimental brittleness on the material. 3.5. The formation of carbides Carbides of the MC type are important strengthening phases in superalloy [42]. The phase diagram shown in Fig. 4 suggests that only two kinds of carbides, MC and M23 C6 , are possible phases in the whole FGM deposit; however, it is interesting to find that in the laser deposit, apart from MC, some M6 C carbides were obtained and there no evidence of the presence of M23 C6 carbides. Since the formation temperature of MC is higher than that of M23 C6 , it is possible that the nucleation of M23 C6 was completely suppressed due to the rapid cooling rates of the laser forming process. As for M6 C, this could be one of the decomposition products of MC carbides. The transformation could be brought about by reheating due to subsequent laser passes. Fig. 16 shows the morphology of some typical carbides and the SAD pattern of MC. The TEM examination revealed no crystallographic relationship between the MC phase and the matrix; that is to say, the MC carbides are highly likely to be nucleated directly from molten metal instead of involving solid state transformation. The TEM photos show that the MC carbides usually resemble blocks, and many of them exhibited a cored structure. The composition of the periphery and the core regions of the block carbides was determined by EDS, and the results are presented in Table 3. The calculated lattice constant of the MC carbides ranged from 0.424 to 0.445 nm, which is close to the 0.433 nm of TiC. This strongly suggests that the MC carbides in the FGM deposit are of the TiC type. Based on the EDS analysis and the SAD patterns, it is also considered that the core of the MC carbides contains carbonitrides, which have an fcc structure (Fig. 17). It was anticipated that the nitrogen gas that was dissolved in the melt could lead to the formation of carbonitrides, which have been found to be more stable than carbides [43]. The carbonitrides that form can act as nucleation sites for MC carbides to develop during solidification. It was also found that the core of MC has a crystallographic relationship of {1 0 0}core //{1 0 0}periphery ; 1 0 0 core // 1 0 0 periphery with the MC carbides at the periphery, except for the Type 1 carbides listed in Table 3. For the Type 1 carbides, which have a high concentration of Al in the core region, the relationship becomes {1 0 0}core //{1 1 0}periphery ; 1 0 0 core // 1 0 0 periphery .
304
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
Fig. 17. The SAD patterns of the core of MC carbides (a) Type 1, a = 0.503 nm; (b) Type 2, a = 0.433 nm; (c) Type 3 a = 0.396 nm (refer to Table 3). Fig. 16. Morphology of typical carbides found in (a) 50% Rene88DT, (b) 70% Rene88DT and (c) Rene88DT.
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
4. Conclusions The stainless steel-SS316L/superalloy-Rene88DT graded material with a linear compositional gradient from 100% SS316L stainless steel to 100% Rene88DT superalloy can be fabricated using laser rapid forming (LRF). Primary ␥dendrites with strong directional growth in the 1 0 0 crystallographic direction were achieved in the entire FGM deposit of 40 mm in height; this is in agreement with the predictions of the microstructure selection models based on the maximum interface temperature criterion. When the amount of Rene88DT reached 40%, the ␥ phase started to form in the interdendritic regions, while ␥ precipitation was evident within the dendrites when the amount of Rene88DT reached 60%. Moreover, as the amount of Rene88DT increased, the amount and the size of the ␥ phase both increased. The size of the ␥ precipitates in the gradient zone was mainly governed by the concentration of the ␥ -forming elements. This is in agreement with the LSW theory. With regard to the precipitation of the phase, this was present in the gradient zone where the composition lay between 45 and 80% Rene88DT. Moreover, the precipitation of the phase occurred when the Ti/Al ratio was greater than 2.5; however, too high a concentration of Al could hinder the nucleation of the phase even when this condition is reached. The interlamellar spacing of the phase formed in LRF was much finer than that obtained by conventional heat treatments. Unlike the predictions of the phase diagram, apart from MC, some M6 C carbides formed, and there was no evidence of the presence of M23 C6 carbides. The absence of M23 C6 carbides could be due to the rapid cooling rates of the LRF process. Most of the MC carbides exhibited a cored structure; it is considered that the MC carbides were nucleated at some carbonitride phases that have some fixed crystallographic relationships with the adhered MC carbides.
Acknowledgments The work described in this paper was funded by The Hong Kong Polytechnic University under the Postdoctoral Fellowships Scheme (Project no. G-YX10). The authors would like to thank Dr. H.O. Yang and Prof. W.D. Huang of Northwestern Polytechnical University of P.R.C. for giving useful suggestions to the research work.
References [1] E. Nembach, G. Neite, Prog. Mater. Sci. 29 (1985) 177–319. [2] S. Suresh, A. Mortensen, Fundamentals of Functionally Graded Materials, The Institute of Materials, IOM Communications Ltd., London, 1998. [3] Y. Miyamoto, Functionally Graded Materials: Design, Processing, and Applications, Kluwer Academic Publishers, Boston, 1999.
305
[4] D.M. Keicher, J.E. Smugeresky, J.A. Romero, M.L. Griffith, L.D. Harwell, in: Proceedings of the 2293, 1997, pp. 91–97. [5] E. Schlienger, D. Dimos, M. Griffith, J. Michael, M. Oliver, T. Romero, J. Smugeresky, Proceedings of the Third Pacific Rim International Conference on Advanced Materials and Processing July 12–16, vol. I, Honolulu, Hawaii, 1998, pp. 1581–1586. [6] G.K. Lewis, R.B. Nemec, J.O. Milewski, D.L. Thoma, M.R. Barbe, D.A. Cremers, Proceedings of the ICALEO ’94, Laser Institute of America, Orlando, Florida, 1994, p. 17. [7] M. Gaumann, S. Henry, F. Cleton, J.-D. Wagniere, W. Kurz, Mater. Sci. Eng. A 271 (1999) 232–241. [8] Y.M. Li, H.O. Yang, X. Lin, W.D. Huang, J.G. Li, Y.H. Zhou, in Proceedings of the SPIE 4915, 2002, pp. 395–402. [9] J. Mazumder, A. Schifferer, J. Choi. Mater. Res. Innovat. 3 (1999) 118–131. [10] K.M. Jasim, R.D. Rawlings, D.R.F. West, J. Mater. Sci. 28 (1993) 2820–2826. [11] J.H. Abboud, D.R.F. West, R.D. Rawlings, Mater. Sci. Technol. 10 (1994) 414–420. [12] J.H. Abboud, D.R.F. West, R.D. Rawlings, J. Mater. Sci. 29 (1994) 3393–3398. [13] J.H. Abboud, R.D. Rawlings, D.R.F. West, J. Mater. Sci. 30 (1995) 5931–5938. [14] T. Seefeld, C. Theiler, E. Schubert, G. Sepold, Mater. Sci. Forum 308–311 (1999) 459–466. [15] G.K. Lewis, E. Schlienger, Mater. Des. 21 (2000) 417–423. [16] P.C. Collins, R. Banerjee, S. Banerjee, H.L. Fraser, Mater. Sci. Eng. A 352 (2003) 118–128. [17] R. Banerjee, P.C. Collins, D. Bhattacharyya, S. Banerjee, H.L. Fraser, Acta Mater. 51 (2003) 3277–3292. [18] Y.T. Pei, J.T.M. De Hosson, Acta Mater. 48 (2000) 2617–2624. [19] F.J. Kahlen, A. Von Klitzing, A. Kar, in Proceedings of the ICALEO’99, vol. 87, Laser Institute of America, San Diego, California, November 1999, pp. F129–F137. [20] M.L. Griffith, L.D. Harwell, J.A. Romero, E. Schlienger, C.L. Atwood, J.E. Smugeresky, Proceedings of the Solid Freeform Fabrication Symposium, Austin, TX, 1997, p. 387. [21] J.A. Brooks, C.V. Robino, M.L. Griffith, T.J. Headley, N.C.Y. Yang, S.H. Goods, D.F. Susan, J.D. Puskar, LDRD Final Report: Exploiting LENS Technology Through Novel Materials. Sandia Report SAND2001-8186, 2001. [22] X. Lin, T.M. Yue, H.O. Yang, W.D. Huang, Mater. Sci. Eng. A 391 (2005) 325–336. [23] C.T. Sims, N.S. Stoloff, W.C. Hagel, Superalloy II—High Temperature Materials for Aerospace and Industrial Power, John Wiley & Sons, New York, 1987. [24] E.F. Bradley, Superalloys: A Technical Guide, ASM International, Metals Park, OH, 1988. [25] N. Saunders, M. Fahrmann, C.J. Small, in: K.A. Green, T.M. Pollock, R.D. Kissinger (Eds.), Superalloys 2000, TMS, Warrendale, 2000, p. 803. [26] J.M. Vitek, S.A. David, in: J. Mazumder, K. Mukerjee, B.L. Mordike (Eds.), Prediction of Non-Equilibrium Solidification Modes in Austenitic Stainless Steel Laser Welds. Laser Materials Processing IV, The Minerals, Metals & Materials Society, Warrendale, PA, 1994, pp. 153–167. [27] W. Kurz, B. Giovanola, R. Trivedi, Acta Metall. 34 (1986) 823–830. [28] M. Rappaz, S.A. David, J.M. Vitek, L.A. Boatner, Metall. Trans. A 21 (1990) 1767–1782. [29] M.J. Aziz, J. Appl. Phys. 53 (1982) 1158–1168. [30] W.J. Boetinger, S.R. Corriel, R.F. Sekerka, Mater. Sci. Eng. 65 (1984) 27–36. [31] D. Rosenthal, Trans. ASME 11 (1946) 849–866. [32] H.E. Cline, T.R. Anothony, J. Appl. Phys. 48 (1977) 3895–3900. [33] M. Gaumann, C. Bezencqn, P. Canalis, W. Kurz, Acta Mater. 49 (2001) 1051–1062. [34] D. Turnbull, J.C. Fisher, J. Chem. Phys. 17 (1949) 71–73.
306
X. Lin, T.M. Yue / Materials Science and Engineering A 402 (2005) 294–306
[35] A.J. Aredll, Acta Metall. 20 (1972) 61–71. [36] C.K.L. Devies, P. Nash, R.N. Stevens, Acta Metall. 28 (1980) 179–189. [37] Y. Enemoto, M. Tokuyama, K. Kawazaki, Acta Metall. 34 (1986) 2119–2128. [38] M. Lifshitz, V.V. Slyozov, J. Phys. Chem. Solid. 19 (1961) 35–50. [39] C. Wagner, Z. Electrochem. 65 (1961) 581–591. [40] P.W. Voorhees, M.E. Glicksman, Metall. Trans. A 15 (1984) 1081–1088.
[41] U.R. Kattner, W.J. Borttinger, S.R. Coriell, Z. Metallkd. 87 (1996) 522–528. [42] R. Fernadez, J.C. Lecomte, T.Z. Kattamis, Metall. Trans. A 9 (1978) 1381–1386. [43] F. Beneduce, A. Mitchell, S.L. Cockcroft, A.J. Schmalz, in: R.D. Kissinger, D.J. Deye, D.L. Anton, A.D. Cetel, M.V. Nathal, T.M. Pollock, D.A. Woodford (Eds.), Superalloys, The Minerals, Metals and Materials Society, 1996, pp. 465–469.