Vous êtes sur la page 1sur 8

Intermetallics 95 (2018) 94–101

Contents lists available at ScienceDirect

Intermetallics
journal homepage: www.elsevier.com/locate/intermet

Heat treatment of laser metal deposited TiAl TNM alloy T


a,∗ b c a
Silja-Katharina Rittinghaus , Ulrike Hecht , Valérie Werner , Andreas Weisheit
a
Fraunhofer Institute for Laser Technology ILT, Steinbachstr. 15, 52074 Aachen, Germany
b
Access e.V., Intzestr. 5, 52072 Aachen, Germany
c
Chair for Laser Technology, RWTH Aachen University, Steinbachstr. 15, 52074 Aachen, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: For the manufacturing of low pressure turbine blades, titanium aluminides are used as structural materials,
A. Intermetallics among them the alloy Ti-43.5Al-4Nb-1Mo-0.1B (at.-%) known as TNM™ alloy. In this study, laser metal de-
B. Annealing position (LMD) was investigated as a potential technology to repair defective, damaged or worn blades, and
Phase transformation process conditions were successfully optimized. However, post-process heat treatment is required to optimize the
C. Casting
microstructure. In this work, different annealing treatments and the resulting microstructures were characterized
Heat treatment
for bulk LMD samples as well as for investment cast reference samples. Annealing temperatures were selected
Laser processing and cladding
D. Microstructure from 1245 to 1290 °C. The results show that annealing at 1290 °C for a duration of 60 min with subsequent
G. Aero-engine components furnace cooling can be used so that the LMD and cast micro-hardness nearly match, albeit with significant
differences in microstructure. Nonetheless, yield strength mismatch can be mitigated and LMD can qualify as
prospective additive repair technology, but also as a manufacturing technology for functional TiAl parts.

1. Introduction similar heat treatment conditions while creep strength is in the same
range. Annealing is used to coarsen the excessively fine-grained struc-
Given their low density, high strength and creep resistance com- ture of the LMD material obtained in the as-built condition. A nearly
bined with good oxidation resistance, intermetallic TiAl-based alloys lamellar microstructure with a small amount of globular γ-phase is
have been qualified for aircraft applications, e.g. for low pressure tur- targeted to achieve a good compromise between strength, ductility,
bine blades [1–6] which operate at temperatures below 800 °C. Con- fracture toughness and creep resistance [20]. The amount and size of
ventional manufacturing but also additive manufacturing (AM) have globular γ-grains and of lamellar grains, respectively, are strongly de-
been considered for blade production [7] and compete with respect to pendent on alloy composition, annealing temperature and holding time.
materials quality, fly-to-buy ratio and production costs. Additive repair For the TNM™ alloy, the fraction and size distribution of BCC-B2 (or β0-
technologies may contribute to increasing cost effectiveness by offering phase) sensitively also depend on the heat treatment conditions [14].
a means to repair defective parts and thus lower the scrap rate, but they As a general rule, lower annealing temperatures favor the formation of
may also be envisaged for repairing damaged or worn blades within the globular γ-grains while longer annealing times increase the amount of
Maintenance, Repair and Overhaul business (MRO). Laser metal de- α-phase [21,22]. Measured β0-fractions are decreased with both faster
position (LMD) is an established repair technology for e.g. parts made cooling and increased annealing temperature [11]. Finally, the lamellar
from Ni-based and titanium alloys [8]: a corresponding technology for spacing is mainly influenced by the cooling rate applied after annealing.
repair of TiAl turbine blades is currently being investigated [9]. Crack It decreases with increasing cooling rate [17].
free samples with low porosity can be built. Among others, a key With respect to samples processed with LMD, additional aspects
challenge is to devise adequate heat treatment steps which would allow have to be considered: first, the resulting microstructure is expected to
the industry to match the microstructure and properties between the be very fine, which may lead to accelerated phase transformation ki-
substrate and the newly and locally deposited alloy. netics during annealing treatments. Secondly, oxygen pick-up is likely
In this work we selected the β-solidifying TNM™ alloy with nominal to occur, even if processing is conducted under a protective argon at-
composition Ti-43.5Al-4Nb-1Mo-0.1B (at.-%) [10–17] for investigating mosphere: the increased oxygen content may change phase transfor-
the response of LMD and cast reference material to different annealing mation temperatures and stabilize the α2 and α-phases [23,24] while γ-
treatments between 1240 and 1290 °C, as according to [18,19] micro- and β0-phase fractions might decrease. Finally, evaporation of alloying
structures of cast and forged TNM™ do not differ very strongly under elements during LMD may cause global or local composition changes in


Corresponding author.
E-mail address: siljakatharina.rittinghaus@ilt.fraunhofer.de (S.-K. Rittinghaus).

https://doi.org/10.1016/j.intermet.2018.02.002
Received 21 September 2017; Received in revised form 23 January 2018; Accepted 2 February 2018
0966-9795/ © 2018 Published by Elsevier Ltd.
S.-K. Rittinghaus et al. Intermetallics 95 (2018) 94–101

addition to meso-scale segregations inherent to the layer-by-layer pro- Table 2


cessing. All these issues will have to be accounted for, when the po- Post-processing annealing treatments: list of parameters.
tential of LMD as additive repair technology is developed and assessed
Parameters, step 1 Parameters, step 2
for TiAl-based alloys in comparison with other technologies, including
multipath welding [25], diffusion bonding [26,27] or diffusion brazing Temperature °C Holding Cooling- Temperature°C Holding Cooling-
[28]. time min time min
The present paper is set out to compare microstructure and micro-
1245 240 Air cooling 800 360 FC
hardness evolution in bulk LMD specimens and cast reference speci- (AC)
mens as function of the annealing temperature, duration and cooling 1265 60 Furnace 1085
conditions for the TNM™ alloy with nominal composition Ti-43.5Al- cooling
4Nb-1Mo-0.1B (at.-%). As a guideline for experimental parameter se- (FC)
1290 20 Water
lection and the interpretation of experiment results, we used the phase quenching
diagram for Ti-xAl-4Nb-1Mo-0.1B proposed by Ref. [29]. (WQ)

2. Experimental
2.3. Analysis methods

2.1. Material and LMD


Chemical composition: the oxygen content of as-cast and as-built
material was analyzed by means of hot gas extraction, while metallic
Cast reference samples were produced by centrifugal investment
elements were analyzed with Energy Dispersive X-ray Spectroscopy
casting with the nominal composition Ti-43.5Al-4Nb-1Mo-0.1B (at.-%).
(EDX). An independent standard was used for TNM™, previously
LMD samples 4 × 10 × 10 mm3 in size were built out of powder with
characterized by Inductively Coupled Plasma Optical Emission
the same nominal composition. The powder was produced by TLS
Spectrometry (ICP-OES). Occasionally and for control reasons selected
Technik GmbH & Co Spezialpulver KG, which used the EIGA atomiza-
samples were additionally analyzed by ICP-OES.
tion process with a delivered particle size distribution ranging from 20
Microstructure analysis was performed using an SEM type Gemini
to 90 μm. The resulting actual composition was Ti-44.7Al-4.5-Nb-
1550 (Zeiss) equipped with an energy dispersive X-ray system (Oxford
1.0Mo-0.15B (at-%) and 600–820 ppm oxygen and is further mentioned
INCA). The samples for microstructure analysis were cut parallel to the
in chapter 3.1. The LMD process was performed in an argon purged
build direction “z”. The SEM was operated in the back scatter electron
inert gas chamber (O2 < 50 ppm) [9] with inductive preheating of the
mode (BSE) to take advantage of the Z-contrast during image analysis.
processing plane to approx. 850 °C, implemented by sequential eleva-
Two crystal orientation maps were acquired by Electron Backscatter
tion of the induction coil with each layer built. The LMD parameters
Diffraction (EBSD) using the Oxford INCA-Crystal software. The aim
applied are listed in Table 1. An alternating hatching strategy was used
was to compare the as-cast and LMD microstructure obtained after an
for adjacent tracks and layers.
annealing treatment of 1290 °C/60 min/FC, specifically to verify the
The laser-beam source was a 3 kW Nd:YAG laser. Powder was fed
phase fractions obtained by simple image analysis. For simplicity's sake,
into the processing zone through a continuous coaxial powder nozzle
a cubic lattice was used for the γ-phase instead of the tetragonal L10-
with an additional protective inert gas stream of 15 l/min.
lattice.
Image analysis: the lamellar colony size and phase fractions were
2.2. Annealing treatment determined through image analysis of SEM micrographs using the
Stream Motion Software by Olympus. Lamellar (α2+γ) colonies were
Post-processing annealing treatment was performed in a THERM- measured both along the longest lamella (referred to as length) and
CONCEPT KLS 05/13 furnace with 300 l/h argon inert gas flow. Table 2 across the respective colony in direction perpendicular to the lamellae
lists the different heat treatment parameters simultaneously applied to (referred to as width). For each sample, a minimum of three micro-
as-cast and as-built LMD samples. graphs was used to obtain representative values. Distorting projection
Air cooling (AC) refers to cooling in air immediately after the re- effects caused by lamellae not being cut perpendicular to the respective
spective heat treatment cycle ends, while furnace cooling (FC) refers to growth direction were neglected. For phase fraction measurements,
passive cooling in the furnace through controlled temperature decrease regions of interests (ROI) in the SEM micrographs were defined. In cases
down to 300 °C with a maximum rate of 5 K/min followed by shutdown of unevenly distributed phases, multiple ROIs were used to detect local
of the furnace. After the samples reached 300 °C, they were air cooled to differences. The phases were differentiated corresponding to their
room temperature. For water quenching (WQ) the samples were re- measurable shading (grey scale) in the micrographs. Through manual
moved from the furnace immediately after the heat treatment cycle definition of limit values, the phase fractions were determined as area
ended and quenched in a bucket with approx. 5 l of water at 21 °C. The segments. The detected areas were accumulated and expressed as phase
time span between opening the furnace and the samples hitting the fraction relative to the selected ROI area. Since it is manual, this
water surface was less than 10 s. All second heat treatment steps ended measurement procedure is, however, subjective. Therefore, the values
with FC. While the cooling rate during FC ranged from 5 to 1 K/min, AC presented here can be considered best estimates rather than hard data.
leads to cooling rates in the range of 103 K/min. WQ cooling rates were Changes in grey scale settings by two to three units can increase or
not determined. decrease the measured phase fraction by up to ± 4.5%. The phases
determined were β0 (bright), α2 (grey) and γ (dark grey).
Table 1 Micro hardness: the micro-hardness was measured as HV0.3 using a
LMD parameters used in this study. Leco semi-automatic Vickers tester with a test load of 300 g and a
testing time of 20 s. The test indentation distance was 0.5 mm.
Parameter Symbol Unit Value
Measurements were performed in the middle section of each sample
Laser beam power P W 60 and parallel to the build direction.
Laser beam diameter D mm 0.6
Velocity V mm/min 500
Layer thickness d mm 0.2
Powder mass flow M g/min 1.5

95
S.-K. Rittinghaus et al. Intermetallics 95 (2018) 94–101

Fig. 1. SEM micrographs of TNM alloy in the LMD as-built (a) and as-cast (b) condition, magnification 1000x.

3. Results Table 3
Chemical composition of as-built TNM material (integral), quantified by EDX measure-
3.1. Microstructure and chemical composition in the as-built condition ments (average of a minimum of three measurements).

Material Ti, at.% Al, at.% Nb, at.% Mo, at.% B, at.% O, wt-ppmb
3.1.1. Microstructure
LMD leads to fine-grained microstructures caused by high cooling Powder (EDX) 51.2 43.9 4.0 0.9 a
600–820
rates up to 105 K/min. Microstructure inhomogeneities are associated Powder (ICP- 49.4 44.7 4.5 1.0 0.15
OES)
with local differences in composition and cooling rates in the deposited a
LMD-sample 49.0 45.7 4.4 0.9 1500–2000
tracks/layers compared to interlayer boundary regions. The differences Cast reference 51.0 44.1 4.0 1.0 a
400–600
will be presented and discussed in a dedicated paper along with process
a
and microstructure simulation results. Here, we only state that com- Not traceable.
b
position differences at the level of 0.5 at.% Al are found between layers Analysis by carrier hot gas extraction.

and interlayer boundaries, being the origin of minor microstructure


differences. The following results are mainly representative for layers. considered, Al still increases significantly during the LMD process. A
The thin interlayer boundary regions have been excluded as far as rise from 44.7 at.% Al to 45.7 at.% Al from powder to the solid LMD
possible. samples, respectively, is confirmed by several control measurements,
Fig. 1 presents SEM-BSE images from longitudinal sections of the as- both EDX and ICP-OES. Any sources for additional aluminum during
built (a) and as-cast (b) samples. Both microstructures consist of la- storage or LMD processing have been considered and can be excluded.
mellar grains (α2+γ colonies) and intergranular γ -grains as well as The powder particles homogenously consist of the listed elements.
BCC-B2 (β0) grains. The phases can be distinguished in Fig. 1 as follows: Thus, theoretical considerations of flight characteristics of Al-rich par-
α2 (grey), γ (dark grey) and β0 (light grey/white). In the LMD-sample ticles differing from the ones of Ti-rich particles are doubtful. The
the lamellar grain size appears to be one order of magnitude smaller melting process and, therefore, evaporation of titanium over aluminum
than the as-cast reference microstructure. under special circumstances could provide an explanation. Examination
Lamellar (α2+γ) colonies and nearly globular γ-phase with β0 at of this possibility requires information about melt pool temperature and
colony boundaries are characteristic for as-cast samples. The size of the dynamics, which is currently not available. Hence, the reason why Al
lamellar (α2+γ) colonies, measured as the mean from their length and increases or Ti decreases remains unexplained for now.
width, ranges from about 15 μm up to 100 μm with an average of ap- Oxygen pick-up occurs both during powder production and the LMD
prox. 20 μm ± 7 μm. Lamellar spacings are measured between < 1 process. Control experiments without inert argon atmosphere result
and 5 μm with 0.7 ± 0.5 μm in average. The γ- and β0 phase at colony in > 20,000 wt-ppm oxygen in the LMD material, which is prohibitive
boundaries appear in coarse clusters [16]. In the as-built sample (LMD), for any application. The other alloying elements are detected in the
similar phases are visible: however, all microstructure components are expected range according to the nominal composition. Because of
smaller: the (α2+γ) colony size ranges from few μm up to 10 μm with complementary effects Al and oxygen might have on microstructure
an average value of 6–8 μm, being up to one order of magnitude smaller transformation, experiments with powder material with decreased Al
than in the as-cast condition. The β0-and γ-grains are much smaller, content are in progress as well as attempts to further decrease oxygen
blocky and rather uniformly distributed. The average size of γ-grains is pick-up during processing. This will allow to examine the effects of Al
3 μm, while lamellar spacings are well below 1 μm with 0.3 ± 0.1 μm and oxygen increase separately.
in average. These features primary result from rapid cooling. As such,
the differences between the as-built and as-cast microstructures may be 3.2. Microstructure after heat treatment – the effects of annealing
judged as significant, amounting to about one order of magnitude for all temperature and cooling rate
characteristic length scales. The purpose of post-processing annealing
heat treatments is to mitigate these differences through coarsening at Samples with a size of 4 × 4 × 10 mm3 were annealed at 1245 °C,
temperatures selected within two-phase fields, either (α+γ) or (α+β) 1265 °C and 1290 °C for 1 h in argon atmosphere and air cooled in order
which, according to [27], correspond to temperatures in the range to determine the phase region reached at the annealing temperature;
1230 °C–1300 °C depending on the alloy composition, i.e. the aluminum see Fig. 2. While in the cast samples, lamellae start to disintegrate at
content. 1245 °C (Fig. 2b) and nearly completely dissolve at 1265 °C (Fig. 2d),
the aluminum-rich LMD samples keep significant amounts of γ-phase
3.1.2. Chemical composition even up to 1290 °C (Fig. 2a, c, e). Moreover, the β0 phase is nearly
In contrast to reported values in literature [30] for e.g. EBM, no loss absent in the LMD samples, again in agreement with their increased
but rather gain in Al is measured for LMD samples; see Table 3. When aluminum content. At T = 1290 °C the LMD sample is thus in the
measurement errors - up to 0.2 at.% calculated standard deviation – are (α + γ) phase field while the cast sample is in the (α + β) phase field.

96
S.-K. Rittinghaus et al. Intermetallics 95 (2018) 94–101

Fig. 2. SEM micrographs of LMD (top) and cast (bottom) samples after annealing at 1245 °C (a, b), 1265 °C (c, d) and 1290 °C (e, f), for duration of 1 h, followed by air cooling,
magnification 1000x.

One can observe in Fig. 2e that interboundary layers possibly consist of equiaxed γ-grains and also coarse lamellae are visible. The latter are
pure α-phase. ripened primary lamellae which have been stabilized by comparatively
Fig. 3 shows samples annealed at 1290 °C for 1 h and subsequently higher Al- and/or oxygen contents and therefore not been dissolved
cooled with different cooling methods: The effect of furnace cooling during annealing. Above mentioned microstructure evolution trends
(Fig. 3a and b) is distinct, leading to coarser γ-lamellae as well as are confirmed by image analysis data, displayed in Fig. 4 for the la-
coarser γ-grains at colony boundaries. Fig. 3a shows the small amount mellar spacing and the fraction of β0.
of β0 grains formed upon cooling at γ-grain boundaries in the LMD The distribution of lamellar spacings for all conditions is broader in
sample. Lenticular γ inside the β0 regions formed during air cooling is LMD compared to that of cast material (Fig. 4, left). When the types of
visible in the cast material. Water quenched samples (Fig. 3e and f) are lamellae are separated into coarse and fine, the similarity of fine la-
roughly similar to air cooled samples (Fig. 3c and d) for LMD and cast mellae in LMD and cast lamellae becomes apparent, which is not sur-
material, both of which show supersaturated α2 and ultrafine γ-la- prising since they stem from the same solid state transformation event
mellae. and exhibit similar phase transformation kinetics.
Both cast and LMD furnace cooled microstructures (Fig. 3a and b) From the diagram in Fig. 4 (right), it follows that the β0-phase is
contain typical, fine (α2+γ) secondary lamellar colonies formed out of completely or nearly absent in the LMD samples after fast cooling. The
former α-grains. However, in the LMD samples higher amounts of significantly smaller amount of β0 in LMD compared to that of cast

Fig. 3. SEM micrographs of LMD (top) and cast (bottom) samples after annealing at 1290 °C, 1 h, followed by furnace cooling (a, b), air cooling (c, d), and water quenching (e, f),
magnification 1000x.

97
S.-K. Rittinghaus et al. Intermetallics 95 (2018) 94–101

Fig. 4. Lamellar spacings and β0-phase fractions as functions of the cooling medium after annealing at 1290 °C, 60 min.

samples can be attributed to the higher Al-content and/or increased material type are similar regarding the size of lamellar colonies and
oxygen content. equiaxed γ-grains under consideration of the variance; however, the
The colony size of the LMD sample annealed at 1290 °C, 60 min and coarse LMD γ-lamellae (undissolved γ at 1290 °C) grow visibly after
furnace cooled increased from < 10 μm in the as-built condition to 240 min and reach widths of up to 5 μm with an average of
approx. 20 μm. In contrast, no increase in colony size is observed for 3.1 ± 1.2 μm compared to 2.3 ± 0.9 μm after 60 min (Fig. 4, coarse
annealing treatments at 1245 °C and 1260 °C. Here the lamellar colonies fraction), while the fine lamellae remain similar. In the cast samples no
remain at the same size (7–10 μm approximately). This can be ex- significant changes were detected when the annealing time was in-
plained by β- and γ-grains hindering the grain growth of α-grains below creased from 60 to 240 min. In both materials the phase fractions do not
Tγ, solv [3,15]. Above or close to Tγ, solv the γ-phase is at least partially significantly change with the holding time in the investigated range and
dissolved and less effective in grain boundary pinning. This means that thus correspond to those presented in Fig. 4 (FC samples).
annealing temperatures close to γ-solvus (e.g. 1290 °C) are required to According to Hall-Petch relationships, the colony growth fits well
initiate major changes in the LMD microstructure. with the decrease in measured values of micro-hardness for each ma-
terial separately (Fig. 6, left). Despite obvious differences in colony
sizes, the hardness for both cast and LMD samples is similar at the same
3.3. Microstructure after heat treatment – the effect of annealing time
level when heat treated. It seems that the smaller colony size is coun-
terbalanced by larger lamellar spacings [32] and, specifically, the
Microstructure coarsening was investigated at an annealing tem-
presence of coarse γ-lamellae undissolved γ at 1290 °C. In summary, the
perature of 1290 °C as function of holding time, followed by furnace
hardness of the LMD sample and the as cast reference sample is nearly
cooling. The microstructures obtained are depicted in Fig. 5.
the same after 60 min holding time, suggesting that some mechanical
Short-term heat treatment of 30 min above Tγ, solv leads to grain
properties like yield strength may be comparable, while others, like
coarsening as mentioned in Ref. [31]. In the present work this effect is
creep and fatigue, may still be very different. The higher oxygen con-
obtained after 20 min for both LMD and cast samples (Fig. 6, left). The
tent in LMD material might cause embrittlement and therefore
microstructures after 60 and 240 min of heat treatment duration of each

Fig. 5. SEM micrographs of LMD (top) and cast (bottom) samples after annealing at 1290 °C for holding times 20 min (a, b); 60 min (c, d), and 240 min (e, f), followed by furnace cooling,
magnification 1000x.

98
S.-K. Rittinghaus et al. Intermetallics 95 (2018) 94–101

Fig. 6. Colony sizes and hardness as functions of the annealing time for annealing at 1290 °C, FC.

decreased ductility. One may further envisage heat treatments clearly 3.5. Comparison of cast and LMD in the selected heat treatment condition
above Tγ, solv to complete this investigation.
On basis of the observed microstructure evolution, an annealing
treatment at 1290 °C for 60 min followed by furnace cooling and
3.4. Microstructure after heat treatment – the effect of a second annealing eventually a second step at 800 °C for 6 h followed by furnace cooling
step for stress relaxation are recommended as a heat treatment for the LMD
material and potentially for hybrid material consisting of cast and LMD
When the area of application is taken into consideration, a second parts. Fig. 8 summarizes the main results in term of micro-hardness and
heat treatment step has to be applied to ensure a reliable long term high Fig. 9 summarizes the main microstructural features of LMD and cast
temperature performance. Thereby, internal stresses are reduced and materials, respectively.
the material is transferred into an equilibrium state. Two temperatures, Micro-hardness measurements (HV0.3) reveal that the hardness of
800 °C and 1080 °C, were tested and the resulting microstructures were both the LMD and the cast material decreases due to the annealing
compared with the respective initial state (Fig. 7). No visible or mea- treatment, and the values equalize around 325 to 350 HV0.3. It is no-
surable changes occurred in either the LMD or cast material (Fig. 7c and teworthy that the scattering of the micro-hardness is also reduced
d) when they were heat treated at 800 °C for 6 h followed by furnace during annealing for both materials, involving chemical homogeniza-
cooling. However, when the temperature of this second step was in- tion and microstructure re-formation. The remaining scatter in LMD
creased to 1085 °C, the formation of secondary β0-phase inside the α2- material can be attributed to systematic differences between layer and
lamellae can be observed. The effect is more pronounced in the cast interlayer regions, the latter showing lower γ-fractions. This is shown in
material with lower aluminum and oxygen content and less pronounced Fig. 8 (right) along with the associated length scales. The influence of
in the LMD material with higher aluminum and oxygen content. Most interboundary effects will be discussed elsewhere in detail.
likely, the solubility in the α2 phase changes as the temperature rises Finally the differences between LMD and the cast reference material
with α2 getting gradually richer in Al and poorer in Ti and Mo. after the annealing treatment 1290/60min/FC are summarized in

Fig. 7. SEM micrographs before the additional second annealing step (a, b) and after the additional second annealing step at 800 °C/6 h/FC (c, d) and 1085 °C/6 h/FC (e, f) for LMD (top)
and cast samples (bottom). The initial structure (a, b) corresponds to the first annealing step at 1290 °C/1 h/FC, magnification 1000x.

99
S.-K. Rittinghaus et al. Intermetallics 95 (2018) 94–101

Fig. 8. Hardness measurements of LMD and cast samples in the as-processed condition and the annealed condition (1290 °C/1 h/FC). The right hand image displays interlayer boundaries
which explain the systematic hardness variations in the LMD material.

Table 4 and Fig. 9 based on EBSD maps. Measured phase fractions are Table 4
listed in Table 4. Most likely, thin α2 –lamellae are not identified at the Phase fractions of heat treated TNM material (1290 °C/60 min/FC), quantified by EBSD
measurements.
magnification and mapping resolution used, such that the fraction of α2
is given as lower bound with an estimated +3% based on analytical Material α2-Ti3Al, area-% γ-TiAl, area-% β0, area- %
experience, while the fraction of γ as upper bound with an estimated
−3%. The most significant difference relates to the fraction of the β0- LMD-sample 20.2 (+3%) < 79.3 (−3%) 0.5
Cast reference 13.7 (+3%) < 73.8 (−3%) 12.5
phase, which is nearly absent in the aluminum and oxygen rich LMD
samples.
rich (approx. 1 at.-%) and titanium-lean, with the aluminum content
4. Summary and conclusions increased relative to the initial powder material. Additionally, higher
amounts of oxygen have been measured. As a consequence, the alloys
Microstructure evolution of LMD and cast reference material from go to distinct phase regions upon annealing at 1290 °C, i.e. LMD to
the TiAl alloy TNM™ with the nominal composition Ti-43.5Al-4Nb- (γ+α) and cast reference to (α+β), respectively. Nonetheless, this
1Mo-0.1B (at.-%) was investigated as function of annealing treatments annealing temperature is close enough to γ solvus to allow the ex-
performed in the temperature range 1245–1290 °C. Unexpectedly, and cessively fine microstructure obtained from LMD to coarsen and
as yet without explanation, the LMD material proved to be aluminum- achieve nearly the same hardness as in the cast reference material:

Fig. 9. EBSD mapping of LMD (left) and cast samples (right) after annealing at 1290 °C/1 h/FC showing the pattern quality maps (top) and the phase maps (bottom), respectively.
(yellow - γ; pink - α2, red - β0). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

100
S.-K. Rittinghaus et al. Intermetallics 95 (2018) 94–101

around 325 to 350 HV0.3. Microstructure differences, however, persist [9] Rittinghaus S.-K., Weisheit A., Mathes M., Garcia Vargas W.: Laser metal deposition
after this annealing as follows: of titanium aluminides – a future repair technology for jet engine blades?. In:
Proceedings of the 13th World Conference on Titanium.
[10] B.P. Bewlay, M. Weimer, T. Kelly, A.U. Suzuki, P.R. Subramanian, The science,
i. Due to differing chemical composition, the LMD material contains technology, and implementation of TiAl alloys in commercial aircraft engines, MRS
only minor fractions of β0, which lowers the hardness at room Proc. 1516 (2013).
[11] S. Bolz, M. Oehring, J. Lindemann, F. Pyczak, J. Paul, A. Stark, T. Lippmann,
temperature. S. Schrüfer, D. Roth-Fagaraseanu, A. Schreyer, S. Weiß, Microstructure and me-
ii. The LMD material displays smaller lamellar colonies (average 20 μm chanical properties of a forged β-solidifying γ TiAl alloy in different heat treatment
compared to 60 μm in cast reference) and smaller size of equiaxed γ- conditions, Intermetallics 58 (2015) 71–83.
[12] E. Schwaighofer, H. Clemens, S. Mayer, J. Lindemann, J. Klose, W. Smarsly,
grains, with grain size hardening being counter-balanced by larger V. Güther, Microstructural design and mechanical properties of a cast and heat-
lamellar spacings. treated intermetallic multi-phase γ-TiAl based alloy, Intermetallics 44 (2014)
iii. Lamellar spacings are partially larger in the LMD microstructure 128–140.
[13] P. Erdely, R. Werner, E. Schwaighofer, H. Clemens, S. Mayer, In-situ study of the
since annealing is done in the (γ+α) phase field with pronounced
time–temperature-transformation behaviour of a multi-phase intermetallic β-sta-
coarsening of the undissolved γ lamellae. This contributes to low- bilised TiAl alloy, Intermetallics 57 (2015) 17–24.
ering the hardness. Annealing durations at this temperature should, [14] E. Schwaighofer, M. Schloffer, T. Schmoelzer, S. Mayer, J. Lindemann, V. Guether,
therefore, not exceed 60 min. et al., Influence of heat treatments on the microstructure of a multi-phase titanium
aluminide alloy, Pract. Metallogr. 49 (2012).
[15] H. Clemens, W. Wallgram, S. Kremmer, V. Güther, A. Otto, A. Bartels, Design of
The present results imply that annealing at 1290 °C/60 min/FC al- novel β-solidifying TiAl alloys with adjustable β/B2-phase fraction and excellent
lows the LMD material to homogenize and coarsen such that it reaches hot-workability, Adv. Eng. Mater. 10 (2008) 707–713.
[16] W. Wessel, Mikrostrukturelle Untersuchungen der Rissinitiierung und -ausbreitung
hardness values comparable to cast reference material, despite per- in inter-metallischen TiAl-Legierungen unter zyklischer und quasistatischer
sisting microstructure differences. A second low-temperature annealing Belastung, Dissertation (2011) Kassel.
step at 800 °C for 6 h followed by furnace cooling can be applied [17] F. Appel, J.D.H. Paul, M. Oehring, Gamma Titanium Aluminide Alloys. Science and
Tech-nology, Wiley, Weinheim, 2012.
without noticeable changes of the microstructure. [18] M. Kastenhuber, B. Rashkova, H. Clemens, S. Mayer, Enhancement of creep prop-
Ongoing investigations are focusing on comparing other mechanical erties and microstructural stability of intermetallic β-solidifying γ-TiAl based alloys,
properties, specifically tensile, creep and fatigue properties in order to Intermetallics 63 (2015) 19–26.
[19] M. Kastenhuber, B. Rashkova, H. Clemens, S. Mayer, Effect of microstructural in-
verify the suitability of the LMD process and the proposed annealing stability on the creep resistance of an advanced intermetallic γ-TiAl based alloy,
treatment for repair applications. Both bulk and hybrid samples are Intermetallics 80 (2017) 1–9.
currently being tested. [20] Y.W. Kim, JOM 46 (1994) 30–39.
[21] H. Clemens, P. Schretter, W. Glatz, Prakt. Metallogr. 33 (1996) 19.
Future investigations shall be extended to include short duration
[22] S.A. Jones, M.J. Kaufman, Acta Metall. Mater. 41 (1993) 387–398.
annealing at temperatures in the range from 1300 to 1320 °C, aiming to [23] H. Zhong, Y. Yang, J. Li, J. Wang, T. Zhang, S. Li, J. Zhang, Influence of oxygen on
investigate the coarsening of LMD material in the (α + β) phase field. microstructure and phase transformation in high Nb containing TiAl alloys,
Materials Letters, vol. 83, 2012, pp. 198–201.
[24] Zollinger, J.; Lapin, J.; Daloz, D.; Combeau, H.: Influence of Oxygen on
Acknowledgements Solidification Behaviour of Cast TiAl-based Alloys.
[25] M.F. Arenas, V.L. Acoff, The effect of postweld heat treatment on gas tungsten arc
This work was supported by the German Federal Ministry for welded gamma titanium aluminide, Scripta Mater. 46 (3) (2002) 241–246.
[26] G. Çam, G. İpekoğlu, K.-H. Bohm, M. Koçak, Investigation into the microstructure
Economic Affairs and Energy BMWi [grant number 20T1311]. and mechanical properties of diffusion bonded TiAl alloys, J. Mater. Sci. 41 (16)
(2006) 5273–5282.
References [27] D. Herrmann, F. Appel, Diffusion bonding of γ(TiAl) alloys: influence of composi-
tion, microstructure, and mechanical properties, Metall. Mater. Trans. A 40 (8)
(2009) 1881–1902.
[1] K. Kothari, R. Radhakrishnan, N.M. Wereley, T.S. Sudarshan, Microstructure and [28] K. Hauschildt, A. Stark, U. Lorenz, N. Schell, T. Fischer, M. Blankenburg, et al.,
mechanical properties of consolidated gamma titanium aluminides, Powder Metall. Diffusion brazing of γ-TiAl-alloys: investigations of the joint by electron microscopy
(2007) 21–27 50/1. and high-energy X-ray diffraction, MRS Proc. 1516 (2013) 319.
[2] E.A. Loria, Quo vadis gamma titanium aluminide, Intermetallics 9 (2001) [29] H. Clemens, B. Boeck, W. Wallgram, T. Schmoelzer, L.M. Droessler, G.A. Zicker,
997–1001. H. Leitner, A. Otto, Experimental studies and thermodynamic simulations of phase
[3] D. Hu, X. Wu, M.H. Loretto, Advances in optimisation of mechanical properties in transformations in Ti-(41–45)Al-4Nb-1Mo-0.1B alloys, Mater. Res. Soc. Symp. Proc.
cast TiAl alloys, Intermetallics 13 (2005) 914–919. vol. 1128, 2009, p. U03.
[4] M.H. Loretto, D. Hu, Y.G. Li, Microstructural studies on some ordered Ti-based [30] J. Schwerdtfeger, C. Körner, Selective electron beam melting of Ti-48Al-2Nb-2Cr:
alloys, Intermetallics 8 (2000) 1243–1249. microstructure and aluminium loss, Intermetallics 49 (2014) 29–35.
[5] F. Appel, M. Oehring, R. Wagner, Novel design concepts for gamma-base titanium [31] M. Schloffer, F. Iqbal, H. Gabrisch, E. Schwaighofer, F.-P. Schimansky, S. Mayer,
aluminide alloys, Intermetallics 8 (2000) 1283–1312. A. Stark, T. Lippmann, M. Göken, F. Pyczak, H. Clemens, Microstructure develop-
[6] K. Kothari, R.U. Radhakrishnan, N.M. Wereley, Advances in gamma titanium alu- ment and hardness of a powder metallurgical multi-phase γ-TiAl based alloy,
minides and their manufacturing techniques, Prog. Aero. Sci. 55 (2012) 1–16. Intermetallics 22 (2012) 231–240.
[7] M. Varetti, Design and turbine architectures: AM enabling technologies. [32] G. Cao, L. Fu, J. Lin, Y. Zhang, C. Chen, The relationships of microstructure and
Presentation, ICTM, Aachen, 2015. properties of a fully lamellar TiAl alloy, Intermetallics 8 (5–6) (2000) 647–653.
[8] I. Kelbassa, Qualification of Laser Metal Deposition of BLISKs Out of Nickel and
Titanium Base Alloys, Ph.D. thesis RWTH Aachen University, 2006.

101

Vous aimerez peut-être aussi