Mechanical behavior and related microstructural aspects of a nano-lamellar TiAl alloy at elevated temperatures

Accepted Manuscript Mechanical behavior and related microstructural aspects of a nano-lamellar TiAl alloy at elevated temperatures T. Klein, L. Usategui, B. Rashkova, M.L. Nó, J. San Juan, H. Clemens, S. Mayer PII:

S1359-6454(17)30151-9

DOI:

10.1016/j.actamat.2017.02.050

Reference:

AM 13578

To appear in:

Acta Materialia

Received Date: 9 February 2017 Revised Date:

17 February 2017

Accepted Date: 18 February 2017

Please cite this article as: T. Klein, L. Usategui, B. Rashkova, M.L. Nó, J. San Juan, H. Clemens, S. Mayer, Mechanical behavior and related microstructural aspects of a nano-lamellar TiAl alloy at elevated temperatures, Acta Materialia (2017), doi: 10.1016/j.actamat.2017.02.050. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

RI PT

Nano-lamellar TiAl Mechanical properties

Microstructural changes

M AN U

SC

• Tensile creep testing

→ Creep mechanisms

50 nm

AC C

EP

50 nm

TE D

• Mechanical spectroscopy

• Scanning and transmission electron microscopy • Atom probe tomography → Precipitation mechanisms

ACCEPTED MANUSCRIPT

Mechanical behavior and related microstructural aspects of a nanolamellar TiAl alloy at elevated temperatures T. Klein,a,* L. Usategui,b B. Rashkova,a M.L. Nó,c J. San Juan,b H. Clemens,a S. Mayera a

Department of Physical Metallurgy and Materials Testing, Montanuniversität Leoben, Roseggerstr. 12, 8700 Leoben, Austria

b

RI PT

Departamento Física Materia Condensada, Facultad de Ciencia y Tecnología, Univ. del País Vasco, Aptdo. 644, 48080 Bilbao, Spain c

Departamento Física Aplicada II, Facultad de Ciencia y Tecnología, Univ. del País Vasco, Aptdo. 644, 48080 Bilbao, Spain *

SC

Corresponding author: Thomas Klein, Tel: 0043 3842 402 4204, Fax: 0043 3842 402 4202, e-mail address: [email protected]

Abstract

M AN U

Advanced intermetallic γ-TiAl based alloys, which solidify via the disordered β phase, such as the TNM+ alloy, are considered as most promising candidates for structural applications at high temperatures in aero and automotive industries, where they are applied increasingly. Particularly creep resistant microstructures required for high-temperature

TE D

application, i.e. fine fully lamellar microstructures, can be attained via two-step heattreatments. Thereby, an increasing creep resistance is observed with decreasing lamellar interface spacing. Once lamellar structures reach nano-scaled dimensions, deformation

EP

mechanisms are altered dramatically. Hence, this study deals with a detailed

AC C

characterization of the elevated temperature deformation phenomena prevailing in nanolamellar TiAl alloys by the use of tensile creep experiments and mechanical spectroscopy. Upon creep exposure, microstructural changes occur in the lamellar structure, which are analyzed by the comparative utilization of X-ray diffraction, scanning and transmission electron microscopy as well as atom probe tomography. Creep activation parameters determined by mechanical characterization suggest the dominance of dislocation climb by a jog-pair formation process. The dislocations involved in deformation are, in nanolamellar TiAl alloys, situated at the lamellar interfaces. During creep exposure the

-1-

ACCEPTED MANUSCRIPT precipitation of βo phase and ζ-silicide particles is observed emanating from the α2 phase, which is due to the accumulation of Mo and Si at lamellar interfaces. Keywords: Titanium aluminides; Atom probe tomography (APT); Creep; Nanostructure; Precipitation

RI PT

1. Introduction

Intermetallic γ-TiAl based alloys belong to the most promising candidates to meet todays’ demand for high-temperature structural materials of high load-bearing capacity

SC

combined with low density [1,2]. These materials are nowadays incorporated increasingly into latest generation propulsion systems in both aero and automotive industries and,

M AN U

thereby, effectively increase fuel efficiency and simultaneously reduce the emission of environmentally detrimental pollutants [3,4]. Particularly, β-solidifying alloys, such as TNM alloys with the major alloying element additions Nb and Mo, have emerged in recent years due to their versatility for applications and rather good processability [2,5,6].

TE D

Additionally, augmented creep resistance of these alloys can be attained via microalloying with C and Si [7,8], while allowing for the use of approved manufacturing routes [9]. Thus, these alloying elements were incorporated into a refined TNM alloying concept, denoted as

EP

TNM+ alloy. At service temperature, which is in the range of 700 °C to 800 °C, the microstructure comprises of the three major intermetallic phases γ-TiAl, α2-Ti3Al and βo-

AC C

TiAl, next to small and varying volume fractions of ωo-Ti4Al3Nb, ζ-Ti5Si3 silicides and pTi3AlC or h-Ti2AlC carbides, depending on the exact alloy composition and its thermal history [7,9,10].

Typical microstructures of TiAl alloys that are optimized for high-temperature applications are composed of aligned lamellar arrangements of γ and α2 phases, so-called α2/γ colonies, which form upon decomposition of α/α2 grains below the γ-solvus temperature [11–13]. Changeable amounts of globular γ and βo constituents mainly decorate the colony boundaries [2]. In this respect, mechanical properties of TiAl alloys -2-

ACCEPTED MANUSCRIPT have been reported to be related closely to the morphological appearance of the microstructural constituents, which can be adjusted via critical selection of heat-treatment parameters [14,15]. In particular the fully lamellar microstructure exhibits a superior

RI PT

combination of strength, creep resistance, fracture toughness and ductility. Thereby, an increase in creep resistance has been correlated to a reduction of lamellar interface spacing as lamellar interfaces effectively inhibit dislocation motion [16,17].

SC

Specifically, in terms of mechanisms governing creep in γ-TiAl based alloys, a multitude of investigations claiming different underlying phenomena responsible,

M AN U

depending on microstructure, chemical composition and regime of thermal exposure, have been forwarded in the past years [18–26]. However, studies of creep behavior of nanolamellar structures are documented scarcely in literature [27]. Hence, this study aims at characterizing creep phenomena and related microstructural changes occurring in advanced

TE D

TiAl alloys with a nano-lamellar microstructure. Additionally, the analyses forwarded here may serve as model for other nano-layered materials, e.g. cold drawn pearlite [28] or lamellar eutectics [29,30]. In the context of this study we aim at characterizing the creep

EP

behavior of a TNM+ alloy using tensile creep experiments and mechanical spectroscopy. Special emphasis is, moreover, placed on the characterization of decomposition reactions

AC C

occurring in the lamellar structure. By means of scanning and transmission electron microscopy, X-ray diffraction and atom probe tomography, a detailed analysis of the precipitation phenomena prevailing during creep exposure is conducted. Based on these data a model for the governing creep mechanisms and the concomitant precipitation reactions present in an advanced TiAl alloy exhibiting a nano-lamellar microstructure is proposed. 2. Materials and methods 2.1 Alloy composition and processing -3-

ACCEPTED MANUSCRIPT The material investigated is based on the TNM alloying concept [2] refined with nominal additions of 0.3 at.% C and 0.3 at.% Si, which is further denominated as TNM+ alloy. The actual chemical composition is Ti-43.3Al-4.02Nb-0.96Mo-0.12B-0.34C-0.31Si

RI PT

(in at.%). The alloy, processed by plasma arc melting (PAM), was supplied by Hanseatische Waren Handelsgesellschaft mbH & Co KG, Bremen. Subsequent to casting, the material was subjected to a hot isostatic pressing (HIP) procedure at 1200 °C and 200

SC

MPa for 4 h in order to homogenize the microstructure and to close residual casting porosity.

M AN U

The material’s microstructure investigated in the present study was adjusted under atmospheric condition in a high-temperature furnace Carbolite RHF 1600 equipped with three type-S thermocouples. The first heat-treatment step (HT#1) corresponds to a solution heat-treatment and was conducted slightly above the γ-solvus temperature at 1280 °C for

TE D

30 mins followed by air cooling. Subsequently, the material was reheated to a temperature slightly above the envisaged service temperature (HT#2), i.e. in this case to 900 °C for 3 h followed by furnace cooling [31]. This heat-treatment step allows for the precipitation of

EP

fine-scaled γ lamellae from the supersaturated α2 grains, forming α2/γ colonies, which represent the nano-lamellar microstructure subjected to the detailed analyses reported here.

AC C

2.2 Mechanical properties characterization Hardness measurements according to Vickers HV10 were conducted using a universal

testing machine M4C 025 G3M from Emco-Test. All values given correspond to the arithmetic mean of five different indents and the error given is the standard deviation of the mean. Creep properties were assessed by load-controlled tensile creep experiments in air using a Mark II TC 30, Denison Mayes Group, whereby accurate temperature control was guaranteed in the range of ± 2 °C. Specimens of 30 mm gauge length and 6 mm diameter -4-

ACCEPTED MANUSCRIPT were mounted by metric threads (M10) and the elongation was determined by inductive displacement sensors in a differential circuit. Internal friction (IF) was measured using non-destructive mechanical spectroscopy [32]

RI PT

in a high-temperature subresonant torsion pendulum under high vacuum (< 10-5 mbar) described elsewhere [33,34]. The samples of 50x5x1 mm3 were cut by spark erosion and subsequent removal of the surface layer by mechanical grinding. A detailed description of

SC

the different working modes and the analysis procedures can be found in Refs. [35,36]. As the applied stress during IF measurements is two orders of magnitude smaller than in creep

M AN U

measurements (oscillating amplitude ε = 10-5), the temperature required to activate a deformation mechanism is always slightly higher under IF than under creep conditions, and consequently, the IF measurements must be performed up to a higher temperature range. However, the present material was aged at 900 ºC (see subsection 2.1) and, thus, the

TE D

maximum temperature inside the pendulum should not exceed that temperature to avoid drastic changes in microstructure. Hence, the data collected by measuring as a function of temperature, provided insufficient data to study the high-temperature background (HTB)

EP

and a novel analysis method was, therefore, developed recently, which involves measuring the IF as a function of frequency at different constant temperatures [37]. To this end, IF

AC C

data were collected at five isotherms, 840 ºC, 850 ºC, 860 ºC, 870 ºC and 880 ºC, in a frequency range between 1 Hz and 0.005 Hz. 2.3 Microstructural characterization Scanning electron microscopy (SEM) was conducted using an EVO 50 from Zeiss. All

images were taken in back-scattered electron (BSE) mode at an acceleration voltage of 15 kV. Specimens were ground and electrolytically polished using the A3-electrolyte from Struers. Evaluation of colony size, fraction of discontinuous precipitation and βo phase

-5-

ACCEPTED MANUSCRIPT fraction was conducted by quantitative image analysis using the Olympus Stream Motion 1.9 software. Quantification of phase fractions was conducted by X-ray diffraction (XRD) using a

RI PT

D8 Advance from Bruker AXS with a Cu-Kα-radiation source. The software package TOPAS from Bruker AXS was utilized for Rietveld refinement of the diffraction pattern [38].

SC

Transmission electron microscopy (TEM) was performed on a Philips CM 12 at 120 kV and on a Tecnai F20 at 200 kV. The Philips CM 12 was employed for bright field (BF)

M AN U

imaging. For optimal contrast of the lamellar structure in BF mode, specimens were tilted into Blackburn orientation relationship and subsequently tilted 1-2 ° along the α2/γ interface. High-angle annular dark field (HAADF) imaging was conducted using the Tecnai F20 in scanning transmission electron mode (STEM). TEM specimens were

TE D

prepared by precision cutting and subsequent grinding to a thickness of ≈ 100 µm, followed by electrolytic etching at -12 °C with an A3-electrolyte using a LectroPol-5 (both from Struers).

EP

In this study atom probe tomography (APT) was carried out using a local electrode atom probe (LEAP™) 3000X HR by Cameca Instruments Inc. The experiments were

AC C

performed in laser pulsing mode at a temperature of 40 K, a pulse repetition rate of 250 kHz and a pulse energy of 0.4 nJ. Analysis of APT data was conducted using the Cameca IVAS™ 3.6.8 software package, whereby peak overlaps were accounted for by peak deconvolution utilizing knowledge about the natural abundance of the elements identified. Needle-shaped APT specimens were prepared by electrolytic etching of precision cut blanks using 5% perchloric acid in acetic acid to generate a rough tip shape and subsequently using 2% perchloric acid in butoxyethanol for final tip-shaping [39]. 3. Results -6-

ACCEPTED MANUSCRIPT 3.1 Microstructural analysis of the initial material condition In Fig. 1 the microstructure of the TNM+ alloy subjected to the heat-treatment detailed in subsection 2.1 is depicted taken by (a) and (b) SEM in BSE mode and (c) and (d) TEM

RI PT

in BF mode. The microstructure is dominated by α2/γ colonies with an average size of ≈ 40 µm, whereby individual lamellae cannot be resolved using SEM. Moreover, residual but very limited βo phase is visible. This observation is strengthened by quantitative evaluation of phase fractions given in Table 1. Clearly the preponderance of the γ phase followed by

SC

intermediate amounts of α2 phase and low amounts of βo phase is proved. Thus, the present

M AN U

microstructure can be denominated as a fully lamellar microstructure. At colony boundaries and triple points ζ-silicides, which can be identified by their bright appearance in the BSE contrast [10], are visible. Moreover, the colony boundaries are decorated by a zone of discontinuous precipitation (DP). Images in higher magnification (Fig. 1(b) and (c)) evidence the presence of globularized microstructural constituents dominating the DP

TE D

regions. This reaction is mainly triggered by the interface energy stored in the large interfacial area of the lamellar microstructure and the concomitant coherency stresses [40].

EP

During the reaction phase fractions shift toward thermodynamic equilibrium. Quantification of the amount of DP from SEM micrographs yields a fraction of ≈ 2 vol.%

AC C

prior to creep exposure. The lamellar structure of the colonies’ interior (Fig. 1(d)) is dominated by even interfaces (see inset of Fig. 1(d)) with an average interface spacing of ≈ 9 nm, which is comparable to the microstructure reported in Ref. [41] for a similar alloy and heat-treatment. This narrow interface spacing is also responsible for the material’s remarkably high hardness (Table 1) in comparison to the hardness of the initial cast and HIPed condition of 379 ± 2 HV10. 3.2 Mechanical properties at elevated temperature 3.2.1 Tensile creep experiments

-7-

ACCEPTED MANUSCRIPT The mechanical properties of the fully lamellar TNM+ alloy were assessed by tensile creep experiments as summarized in Fig. 2. In Fig. 2(a) creep curves registered with applied stresses of 150 MPa (blue), 200 MPa (green) and 250 MPa (red) at a temperature

RI PT

of 800 °C and in Fig. 2(b) creep curves registered at temperatures of 800 °C (blue), 825 °C (green) and 850 °C (red) with an applied stress of 200 MPa are depicted. The minimal creep rates are indicated next to each respective creep curve. The relationship of minimal

SC

creep rates ε
m , applied stresses σ and temperatures T may be expressed using the constitutive equation given by Eq. (1):

(1)

M AN U

⁄kT ) ε
m = Aσn exp(- Ecreep a

where A is a microstructure dependent material constant, n is the stress exponent, Ecreep is a the apparent activation energy and k is the Boltzmann factor. The slope of the linear regression of ln(ε
m ) as a function of ln(σ) provides the stress exponent n. In case of the

TE D

experiments presented in Fig. 2(a) this derivation yields n = 3.44 ± 0.05. Linear regression of an Arrhenius-plot, i.e. ln(ε
m ) vs. 1/T provides the apparent activation energy Ecreep . a Application of this approach to the data shown in Fig. 2(b) results in Ecreep = 4.63 ± 0.07 a

EP

eV.

A further useful analysis to identify the creep mechanisms involved can be attained via

AC C

determination of the activation volume Va (see works by Appel et al. [25] and Appel and Wagner [13]). The magnitude of Va is in the context of thermal activation representative for the number of atoms involved in the displacement of a dislocation to overcome local obstacles and, thus, may be used to characterize the elapsing micromechanisms of deformation. According to Evans and Rawlings [42], Va can be evaluated from stress jump tests according to Eq. (2): Va = kT

∆ lnε




(2)

∆σ T

-8-

ACCEPTED MANUSCRIPT where k corresponds to the Boltzmann factor, T to the absolute temperature and ∆ε
to the difference in strain rates after increasing the applied stress σ1 to σ2 (∆σ = σ2 - σ1). Evaluation of Va from stress jump experiments is particularly convenient when

RI PT

microstructural changes might occur in the material during creep. In this equation no Taylor factor has been included to account for the polycrystallinity as its definition is uncertain under creep conditions [21,43]. Eq. (2) was applied to the data from a stress jump

SC

experiment conducted at 800 °C, where the stress was raised from initially σ1 = 150 MPa to σ2 = 200 MPa after reaching ε
m . Under these conditions the corresponding creep rates are

M AN U

ε
1 = 5.37·10-9 s-1 and ε
2 = 2.94·10-8 s-1 (instantaneous creep rate after stress jump). Va is then evaluated as ≈ 22b3, where b corresponds to the magnitude of the burgers vector of the ordinary dislocation in the γ phase of 0.284 nm as evaluated from XRD data. 3.2.2 Mechanical spectroscopy

TE D

In order to underpin and extend our understanding of the prevailing high-temperature deformation processes, we tried to corroborate the measured activation energy by an alternative method. Indeed, the HTB of the IF spectra can be related to the material’s creep

EP

behavior, as was reported in previous works [22,44]. In order to describe the HTB a generalized Maxwell rheological model (Eq. (3)) was employed including the distribution

AC C

factor m (0 ≤ m ≤ 1) (here, m = 1, corresponds to ideal viscoelasticity), as proposed by Weller et al. [45]. The factor m physically resembles a distribution of the relaxation time τ associated with the microscopic mechanisms controlling the thermally activated process: tan φω ≡ Q = (ωτ)-m -1

(3)

φ denotes the lag angle between oscillating strain and stress with angular frequency ω. Moreover, as the processes controlling the HTB are thermally activated, an Arrhenius equation holds for the relaxation time τ. Hence, the HTB of the IF spectra follows Eq. (4): Q-1 = ωτ0 exp(EIF a /kT)

-m

(4) -9-

ACCEPTED MANUSCRIPT where τ0 is the pre-exponential factor of the relaxation time related to the atomic attempt frequency, k is the Boltzmann factor, T is the absolute temperature, and EIF a the apparent activation energy.

RI PT

In Fig. 3, IF spectra are plotted exemplarily as a function of temperature for two different frequencies, 0.03 Hz (magenta) and 0.1 Hz (blue). The measured IF spectra and the HTB (cyan and violet) determined according the method developed in [37] and

SC

subsequently analyzed by Eq. (4) are plotted for the nano-lamellar TNM+ alloy. Good accordance of the fitted HTB is discernible with each experimental curve, which is a

yields EIF a = 5.07 ± 0.05 eV.

M AN U

prerequisite for an accurate evaluation of the activation energy. Application of Eq. (4) then

The relaxation peak visible in Fig. 3, denoted as P1, was previously attributed to atomic relaxation phenomena in the α2 phase in a TNM alloy [46]. In case of the TNM+ alloy,

TE D

however, it should be noted that the maximum of the relaxation peak is shifted to slightly higher temperatures in comparison to the TNM alloy, which indicates a higher activation energy for diffusion.

EP

3.3 Comparative assessment of microstructural changes For the purpose of assessing the microstructural changes associated with creep, the

AC C

creep specimen, which was subjected to the longest exposure, i.e. 800 °C and 150 MPa for ≈ 375 h, was analyzed in detail, which will be referred to as the crept microstructural condition in the following sections. During creep exposure hardly any changes in microstructure are traceable by SEM except for a slight progression of the DP reaction fronts. When comparing the phase fractions from XRD data before and after creep exposure (Table 1), small alterations are evident. This data suggests that an α2 → γ phase transformation is elapsing during exposure, which is, however, very limited in comparison to other TiAl systems with a lamellar morphology, where the α2 phase fraction diminishes - 10 -

ACCEPTED MANUSCRIPT strongly during creep [47,48]. Upon creep, a slight reduction in hardness occurs in the fully lamellar TNM+ alloy (∆HV10 ≈ 10), which might be due to softening by the progression of DP and lamellar coarsening [31].

RI PT

In Figs. 4(a) and (b) TEM BF micrographs of the crept condition are depicted in different magnifications. Slight degradation of the lamellar structure in comparison to the initial material condition (Fig. 1) is discernible in Fig. 4(a) and marked exemplarily by

SC

arrows in Fig. 4(b). Clearly the interfacial roughness has increased during creep (compare Fig. 4(a) and (b) with Fig. 1(d)), whereby particularly macro-ledges (1), protrusions (2)

M AN U

and terminating lamellae (3) can be identified. In Fig. 4(c) a detail of the lamellar microstructure is visualized. Micro-ledges prevailing at lamellar interfaces are evident. For clarity one ledge is marked by a red circle exemplarily. The course of the interfacial plane is illustrated by a shifted staggered red line illustrating the ledged nature of the interface.

subsection 4.1.

TE D

This observation is distinctive for the governing creep mechanism as will be discussed in

In Fig. 4(d) the existence of precipitates in the lamellar microstructure after creep is

EP

visible. For this image the TEM specimen has been tilted for suitable contrast of the precipitate structure, thus, the lamellar microstructure cannot be visualized simultaneously.

AC C

The particles formed show a rod-shaped morphology and appear in regular angles with the respective parent lamellae, whereby different orientation variants can be distinguished. In order to determine their preferred location and possible differences between precipitates, HAADF imaging was conducted as represented in Fig. 4(e). In STEM mode a clear distinction between the phases in the lamellar structure is obtainable due to the pronounced Z-contrast. Particles appear bright, α2 lamellae are intermediate gray and the γ lamellae are dark gray. Apparently, the precipitates form within the α2 phase. Moreover, the enrichment of heavy elements in these particles is deductible from their bright appearance in - 11 -

ACCEPTED MANUSCRIPT comparison to the other phases. In order to characterize particles and the crept material condition in more detail, APT was conducted as presented in the following subsection. 3.4 Characterization of the lamellar structure by APT

RI PT

The chemical compositions of the lamellar structure prior and after creep were analyzed by APT. Reconstructions of 20 nm thick slices of the complete atom maps are depicted in Fig. 5. Here, only the strongly partitioning elements, Al (green), Mo (blue) and Si (red) have been indicated, which yield a reliable discrimination of the constituting

SC

phases. In Fig. 5(a) the lamellar arrangement of α2 and γ phases prior to creep is obvious.

M AN U

No precipitates and no severe chemical inhomogeneities are detectible. In Fig. 5(b) the corresponding reconstruction of the interior of a lamellar colony after creep is depicted. Different types of decomposition products are present, which were separated by isoconcentration surfaces at 2 at.% Mo (blue) and 10 at.% Si (red). These correspond to the particles detected by TEM displayed in Fig. 4(d) and (e). Particles preferentially inhabiting

TE D

the lamellar structure will, subsequently, be indicated by the index L for clarity. The Mo isoconcentration surface delineates two Mo-rich particles and the Si isoconcentration

EP

surface delineates a Si-rich phase. The cross-section through the elongated Si-rich particle suggests a hexagonal lattice (≈ 120° angles) and habit planes are visible, which appear

AC C

parallel to the α2/γ interface. Interestingly, a very fine γ lamella terminates at this particle. The absence of p-carbides observed in all APT measurements of the crept specimen condition is most likely due to their relatively large separation resulting from the low C concentration of the TNM+ alloy and the very small sampling volume of a typical APT analysis [7,49]. The chemical compositions of the phases identified have been evaluated quantitatively and are summarized in Table 2. Standard errors based on counting statistics are all below ± 0.03 at.%. Due to the high accuracy of the chemical analysis by APT, this information can - 12 -

ACCEPTED MANUSCRIPT be used to deduce the particles identity. The Ti and Al contents of the γ and α2 lamellae deviate little comparing the conditions prior and after creep. This is due to the fact that the chemical composition of newly formed γ lamellae is already close to thermodynamic

RI PT

equilibrium and, thus, only little changes in phase fractions during creep prevail [10]. In case of Nb, the partitioning ratio remains similar during exposure with a weak tendency toward the γ phase. Hence, different compositional values between the conditions prior and

SC

after creep are assumed to be due to slight chemical fluctuations within the sample material. Mo, on the other hand, shows a weak trend to accumulate in the α2 phase of the

M AN U

lamellar structure. Upon exposure, the concentration of Mo in the α2 phase is reduced slightly. The interstitial element C is preferentially accommodated in the α2 phase in agreement with Refs. [10,50]. In case of Si, a clear phase preference for the α2 phase is discernible, whereby also a slight reduction of Si content during creep is evident.

TE D

Fig. 6 depicts proximity histograms of the elements Mo and Si calculated across α2/γ interfaces of an APT specimen in the crept specimen condition in a precipitate-free region (reconstruction not depicted in this work). The aforementioned phase preference of Mo and

EP

Si is discernible. In case of Mo a weak pile-up and in case of Si a pronounced pile-up at the α2 side of the lamellar interface is visible. During lamellae growth, which proceeds during

AC C

creep as the γ phase fraction increases (see Table 1), these elements are redistributed by diffusion into the energetically preferred phase as will be analyzed in subsection 4.2. The composition of the Mo-rich precipitates (Table 2) is marked by significant

amounts of Nb and Mo, which are both known for their β-stabilizing effect, with the latter showing the stronger stabilizing effect [51]. Based on the pronounced enrichment of these elements, the Mo-rich phase is identified as βo phase. Moreover, C concentrations are very low, which agrees well with reports in literature evidencing a very low solubility of the βo phase for interstitial elements due to the fact that no favored incorporation sites, i.e. Ti6- 13 -

ACCEPTED MANUSCRIPT octaheadral cavities, are intrinsically present in this phase [50]. The amount of Si is lower than the concentration in the α2 parent phase. Hence, diffusional redistribution is inferred to occur during growth of βo particles.

RI PT

In case of the Si-rich particle, the concentration of Si is ≈ 25 %. Moreover, its chemical composition is marked by large amounts of incorporated Nb and Al. The sum of transition metals (Ti, Nb, Mo) as well as the sum of Si and Al closely match the ideal stoichiometry

SC

of ζ-Ti5Si3 silicides, yielding evidence for this particle’s identification. Moreover, the preferred site substitution of Nb and Mo at Ti sites and Al at Si sites is deduced. Hence, the

M AN U

structure formula of these silicides should be adapted to (Ti,Nb,Mo)5(Si,Al)3. Furthermore, significant amounts of the interstitial element C are dissolved in these silicides. 4. Discussion

4.1 Analysis of dominant creep mechanisms based on mechanical characterization There is general consensus in literature [52] pointing toward the conclusion that creep

TE D

deformation involves the motion of dislocations, including gliding and climbing, with the latter mechanism corresponding to the rate controlling step at high temperature. In this

Eq. (5):

EP

case, the apparent activation energy can be described by a general expression according to

Ea = E0 - Va σk

AC C

(5)

where E0 is the energy of the thermally activated barrier (in the absence of any stresses), Va is the activation volume associated to the process and σk the resolved shear stress. k, in this context, refers to the different stress conditions during tensile creep and IF experiments. Thus, the term Vaσk represents the work done by the applied stress, which corresponds the responsible difference between the apparent activation energies due to the different experimental approaches (tensile creep and IF experiments). Indeed, in IF the applied stress is σIF = 10-5G (with G being the temperature compensated shear modulus at 800 °C - 14 -

ACCEPTED MANUSCRIPT taken as 55.26 GPa from Ref. [53]), while the tensile stresses applied during creep testing ranged from 150 MPa to 250 MPa. Consequently, due to this term Vaσk (using Va ≈ 22b3), a difference of about 0.3 eV to 0.4 eV is expected between tensile creep and IF

RI PT

measurements. This argumentation is clearly consistent with the results obtained being Ecreep ≈ 4.63 ± 0.07 eV and EIF a a ≈ 5.07 ± 0.05 eV. Generally, creep deformation involves long distance mobility of atomic defects contributing to the plasticity, while IF acts as a

SC

probe able to excite only short distance processes, which corresponds to the initial steps of the atomic defect motion during creep conditions. Based on these argumentation and the

M AN U

excellent agreement of tensile creep and IF experiments, we conclude that both methods supply complementary information toward the understanding of the micromechanisms involved in high-temperature deformation of TiAl alloys.

From the whole study developed by mechanical spectroscopy in the nano-lamellar

TE D

TNM+ alloy [37], we conclude that the presence of interstitially dissolved C retards the diffusion of Ti atoms, as was revealed by the shift of the P1 relaxation peak (compare P1 in Fig. 3 with analysis of Ref. [46]). This effect can be attributed to the C-vacancy-interaction

EP

and, consequently, it is expected that C will have a similar slowing effect on the diffusion of both Ti and Al atomic species in the different C-containing phases (see Table 2). This

AC C

phenomenon could explain the higher activation energy and creep resistance measured in the fully lamellar TNM+ in comparison with previous reports on the TNM system. Evaluation of the stress exponent and activation volume in the stress regime

investigated yields n = 3.44 and Va = 22b3. The former value is commonly associated with dislocation climb dominated creep [52], which was reported to govern creep of singlephase γ-TiAl [18] and two-phase γ-TiAl TiAl based alloys [19,21,26,54]. This would, however, require the latter value to be in the range of b3 ≤ Va ≤ 10b3 [42]. Nonetheless, activation volumes of the magnitude observed have been reported for different TiAl alloys - 15 -

ACCEPTED MANUSCRIPT under creep conditions [43]. Also, for classic climb processes the evaluated activation energy should correspond to that of self-diffusion of the slowest diffusing species. In this work we have obtained apparent activation energies of 4.63 eV (creep) and 5.07 eV (IF),

RI PT

which is substantially higher than reported values of Ti and Al self-diffusion by Mishin and Herzig in γ and α2 phases [55].

Castillo-Rodríguez et al. [22] suggested dislocation climb by a jog-pair formation

SC

process to govern creep in a Ti-46Al-1Mo-0.2Si (at.%) alloy in either cast (coarse lamellar) or extruded (fine globular) material conditions. This model still holds in case of

M AN U

nano-lamellar microstructures, which, in the following, will be utilized to explain the micromechanisms of interface displacement. The relevant energetic treatment can be expressed as given by Eq. (6): Ekj-p = 2Ej -

Gh2 b2 8π 1-ν xc

- hbxc σk

(6)

TE D

where Ekj-p is the jog-pair formation energy with the index k referring to the respective experimental setup and Ej corresponds to the self-energy of a single jog, which can be

Ghb2

j = 4π(1-ν) .

EP

formulated as given by Eq. (7):

(7)

AC C

The second term (in Eq. (6)) corresponds to the elastic interaction potential as originally introduced by Eshelby [56] and the third term corresponds to the work done by the applied stress σ. G is the temperature compensated shear modulus, h is the distance between {111}γ planes, b is the magnitude of the burgers vector and ν the temperature compensated Poisson ratio. The last term of Eq. (6) allows to evaluate the critical jog-spacing xc at the saddle point from the experimentally measured activation volume according to Eq. (8): V

xc = hba

(8)

- 16 -

ACCEPTED MANUSCRIPT evaluating these expressions with G = 55.26 GPa [53], h = 0.234 nm, b = 0.284 nm (h and b evaluated for the γ phase from XRD data), ν = 0.241 [53] and σcreep = 200 MPa and σIF/G = 10-5 yields xc = 7.58 nm and the middle term due to the jogs interaction is evaluated to be

creep

to amount to Ej-p

RI PT

equal to 0.01 eV. Finally, Ej is evaluated from Eq. (7) as 0.68 eV. Hence, Ej-p is calculated = 0.95 eV and EIF j-p = 1.35 eV. We are, thus, able to evaluate the overall

activation energy, which can be calculated according to Eq. (9) [57,58]: 1

Eka,i = Ea,s-d,i + 2 Ekj-p

SC

(9)

where the index i refers to the phase, where vacancy diffusion occurs. Using the activation

M AN U

energies of self-diffusion Ea,s-d,i of the slowest diffusing species (Al) from Ref. [55] of 3.71 eV and 4.08 eV in γ and α2, respectively, we calculate Ecreep = 4.19 eV, Ecreep a,γ a,α2 = 4.56 eV, IF EIF a,γ = 4.39 eV and Ea,α2 = 4.76 eV. The impact of substitutional alloying elements on the

Al diffusivity is, thereby, assumed to be small as Nb and Mo preferentially occupy Ti sites

TE D

in γ as well as in α2 [59]. The values calculated are similar in magnitude in comparison to the experimentally determined activation energies of both creep experiments as well as IF experiments, provided that the activation energy for Al diffusion in the α2 phase is used.

EP

Deviations from the experimental apparent activation energy could be caused by

AC C

interstitially dissolved C, which is alloyed to the TNM+ system. C atoms may preferentially segregate to vacancies due to the lattice distortion in their close surrounding and, thereby, affect the vacancy diffusivity to a certain degree as deductible from the shift of P1 in the IF spectra, when comparing the TNM and TNM+ alloy. The energy of the P1 peak, linked to the Ti diffusion by exchange with a vacancy in the α2 phase [46], increases by about 0.1 eV when adding 0.3 at.% C [37], and it is expected that the binding energy, C-vacancy, would similarly modify the diffusion of all atomic species in both phases. Indeed, it is conceivable that this mechanism operates differently at the very initial stages of creep, i.e. corresponding to the IF experiments, in comparison to the tensile creep - 17 -

ACCEPTED MANUSCRIPT experiments, where a certain amount of C might already be ligated in precipitates (see Table 2), which may explain the more pronounced deviation between model and experiment in case of IF conditions.

RI PT

We may consequently ask, whether dislocation climb-based creep is a plausible mechanism in a nano-lamellar microstructure. In the interior of individual γ lamellae, dislocations have only been perceived in the few broad γ lamellae present in the condition

SC

investigated, which is consistent with the following argumentation: interfaces, in general, can act as Frank-Read-type dislocation sources [60]. Assuming dislocation segments to be

M AN U

fixed at a separation of 2l, dislocation emission is only possible at l < λ (γ lamella width), which, thus, can be calculated as given by Eq. (10) [61]: l = 2fσ Gb

(10)

where f corresponds to the Schmid factor. Using the same data as given above, f as 0.5 and

TE D

σ as 250 MPa (highest stress applied) yields l ≈ 63 nm. Since, most γ lamellae are by far narrower, dislocation motion is restricted and segments will attain a trapped configuration. Thus, it is assumed that creep deformation of the nano-lamellar-structured TNM+ alloy is

EP

dominated by the motion of interfacial dislocations as visualized in Fig. 7(a), where the motion of interfacial ledges and the concomitant climb of interfacial dislocations are

AC C

evident [62,63]. This mechanism is also congruous with Hazzledine’s [64] theoretical prediction of the loss of coherency at lamellar boundaries to occur at a γ lamella width of λ ≈ 8 nm, above which the introduction of misfit and misorientation dislocation occurs. Moreover, the progression of DP yields an additional contribution to the creep elongation, which is assumed to be of minor influence as the volume fraction of the DP zone is rather small. The fine-grained microstructural constituents of the DP zone would, however, presumably deform via grain boundary sliding. The, nonetheless, rather low value of n of 3.44 might, in comparison of classic dislocation climb models [52], be related to the fact - 18 -

ACCEPTED MANUSCRIPT that the concentration of interfacial dislocations results intrinsically from the formation process of γ lamellae [11,65]. In contrast, classic dislocation climb models assume that the mobile dislocation density ρm is a function of the applied stress [52]. As ε
= ρm bv , it is

RI PT

suggested that the stress dependence observed in nano-lamellar TiAl alloys stems from the dependence of the average dislocation velocity v on the stress in agreement with Ref. [62]. In summary, it is proposed that creep in nano-lamellar TiAl alloys occurs via the motion of

SC

interfacial dislocations, with the rate limiting step of climb by jog-pair formation. The forwarded model suggests that vacancy diffusion required for this process is predominantly

M AN U

limited by the α2 phase, explaining the high experimentally determined activation energies.

4.2 Segregation stimulated particle nucleation revealed by APT Combined stress and temperature exposure, i.e. creep conditions, allow the material’s

TE D

phase fractions to adjust toward thermodynamic equilibrium. In case of γ-TiAl based alloys an α2 → γ transformation occurs in many different systems with a lamellar microstructure including the TNM+ alloy as shown in subsection 3.3 (see also e.g. Refs. [47,48]). During

EP

the transformation sequence, the chemical compositions of the phases involved need to be adjusted via diffusion, in particular of strong partitioning elements, across the phase

AC C

boundary as indicated in Fig. 7(a). Especially, the enrichment of the alloying elements Mo and Si close to the interface is indicative for this process (see Fig. 6). It should be emphasized that the concentration profiles of Fig. 6 were evaluated as proximity histograms sampling over large interfacial areas. Indeed, it is conceivable that at the individual locations where the phase transformation proceeds, i.e. at interfacial ledges, the Mo and Si concentrations are locally significantly higher, which will eventually culminate into particle precipitation reactions. Recent results obtained by ab initio calculations suggest that the α2 phase is slightly destabilized by the incorporation of Mo in its crystal - 19 -

ACCEPTED MANUSCRIPT lattice, while the βo phase is significantly stabilized [59]. Comparing the changes in energies of formation calculated from Ref. [59] with and without Mo on its preferred β

α

o 2 lattice site yields ∆Ef,Mo→Ti = 7.1 meV/atom, whereas ∆Ef,Mo→Al = -20.5 meV/atom. Thus,

RI PT

it is likely that the local enrichment of Mo in the vicinity of ledges promotes the nucleation of βo,L particles by lowering the structural energy of the latter. This argumentation is consistent with an earlier analysis by Pettifor and Podloucky [66], who suggested based on

SC

a tight-binding model that in pd bond-compounds, such as transition metal aluminides (TM-Al), the nature of the pd bond determines the energetically stable crystal structure.

M AN U

From their analysis it follows that an increasing band filling in a TM-Al, e.g. by an increasing Mo concentration at the expense of Ti, the B2 structure becomes energetically favored. Due to the local enrichment of Mo in the vicinity of the interface and the concomitant locally increased band filling, the B2 structure, i.e. the βo phase, eventually

TE D

reaches a lower structural energy and, thus, nucleation occurs. As visualized in Fig. 7(b) the nucleus is assumed to form heterogeneously presumably at an interfacial ledge, which in turn reduces the ledges mobility and contributes to the creep resistance of the material

EP

[67,68].

In case of Si, a similar precipitation sequence is conceivable. Comparison of

AC C

concentrations of Ti and Al in α2 and γ phases provides evidence that the equilibrium concentration of the ζL-silicide closely matches the composition of the interfacial region (see Table 2). Moreover, as evident from Fig. 6, Si is enriched in the vicinity of the interface, which increases the likeliness of silicide nucleation. Eventually, a silicide nucleus is formed heterogeneously, which is the preferred nucleation mode according to Refs. [67,69]. The presumption of the nucleation of both types of particles being stimulated by the presence of interfacial ledges is strengthened by the fact that βo,L and ζL precipitates were found to nucleate close to one another parallel to the interface as deductible from 3D - 20 -

ACCEPTED MANUSCRIPT reconstruction of APT data (not displayed in this work). Upon further exposure, precipitates are driven to grow into the adjacent lamellae as visualized by HR TEM in Ref. [8], whereby their morphology is determined by the minimization of interface energy and

RI PT

the spatial distribution of the diffusion fields of the diffusing species required for particle growth [70].

In Fig. 8 the concentrations of Mo and Si prior and after creep within the α2/γ lamellar

SC

structure are compiled. Clearly, both the Mo and Si concentrations diminish in the α2 phase, corresponding to the depletion triggered by the precipitation of βo,L and ζL particles.

M AN U

While βo,L particles are enriched in Mo, ζL-silicides are enriched in Si. The magnitude of Mo depletion is larger than of Si depletion, suggesting a significantly higher βo,L particle density, which is also supported by the fact that βo,L particles contain lower concentrations of Mo than ζL-silicides contain Si. Concomitantly, concentrations within the γ phase remain

TE D

similar during creep conditions, thus, evidencing that the chemical composition of the γ lamella persist, i.e. are not prone to decomposition and their chemical composition at the formation step is already close to equilibrium.

EP

5. Conclusions

The deformation behavior at elevated temperatures of γ-TiAl based alloys with a nano-

AC C

lamellar microstructure has been investigated in the present study. To this end, the established TNM alloying concept was refined by the addition of 0.3 at.% C and 0.3 at.% Si (TNM+ alloy) yielding improved creep properties attainable via existing manufacturing routes [7,9]. Creep activation parameters were evaluated comparatively by tensile creep and internal friction experiments. Especially, internal friction is a powerful and timesaving approach to identify the micromechanisms of the rate governing creep process and could be of technological interest [22]. Moreover, microstructural changes associated with creep exposure were analyzed using X-ray diffraction, scanning and transmission electron - 21 -

ACCEPTED MANUSCRIPT microscopy as well as atom probe tomography. In the light of the present investigations and analyses, the following conclusions can be drawn: * Tensile creep experiments and internal friction measurements indicate that creep

RI PT

deformation is accomplished by dislocation climb based on a jog-pair formation process. This mechanism allows to explain the magnitude of the stress exponent n, the activation volume Va and the activation energy Ea. The activation energies obtained from creep and

SC

internal friction experiments are in good agreement and can be described by the proposed process.

M AN U

* As the interface spacing of the nano-lamellar-structured TiAl is too narrow to allow for ordinary dislocation motion in the lamella interior, deformation is deduced to proceed via the motion of interfacial ledges, whereby associated interfacial dislocations climb via the aforementioned jog-pair formation process. This process is consistent with an increase of

TE D

interface roughness observed during creep as visualized by transmission electron microscopy.

* During creep exposure, two types of precipitates form exclusively within the α2 lamellae.

EP

These were identified by atom probe tomography as βo,L particles and ζL-silicides enriched in Mo and Si, respectively.

AC C

* The segregation of Mo and Si toward lamellar interfaces during creep exposure as quantified by atom probe tomography facilitates the particle precipitation reactions. Mo and Si accumulations eventually culminate in the precipitation of βo,L and ζL particles, which homogeneously decorate the nano-lamellar structure. In turn a decline of Mo and Si concentrations in the α2 phase accompanies the precipitation reaction. The precipitation of βo,L particles and ζL-silicides is expected to contribute to an increased creep resistance of TNM+ alloys as they potentially stabilize the microstructure and restrict interface sliding Acknowledgements - 22 -

ACCEPTED MANUSCRIPT Technical support by M. Brabetz, field office of the Montanuniversität Leoben at the Research Center Seibersdorf, is greatly acknowledged. T. K. would like to thank Dr. D. Holec for discussions regarding the atomistic concept of the precipitation reactions.

RI PT

References

AC C

EP

TE D

M AN U

SC

[1] F. Appel, J.D.H. Paul, M. Oehring, Gamma Titanium Aluminide Alloys: Science and Technology, Wiley-VCH, Weinheim, 2011. [2] H. Clemens, S. Mayer, Design, Processing, Microstructure, Properties, and Applications of Advanced Intermetallic TiAl Alloys, Adv. Eng. Mater. 15 (4) (2013) 191–215. [3] B.P. Bewlay, S. Nag, A. Suzuki, M.J. Weimer, TiAl alloys in commercial aircraft engines, Mater. High Temp. 33 (4-5) (2016) 549–559. [4] U. Habel, F. Heutling, D. Helm, C. Kunze, W. Smarsly, G. Das, H. Clemens, Forged Intermetallic γ-TiAl Based Alloy Low Pressure Turbine Blade in the Geared Turbofan, in: V. Venkatesh, A.L. Pilchak, J.E. Allison, S. Ankem, R. Boyer, J. Christodoulou, Fraser, H.L. Ashraf Imam, M., Y. Kosaka, H.J. Rack, A. Chatterjee, A. Woodfield (Eds.), Proceedings of the 13th World Conference on Titanium, Wiley, 2016, pp. 1223–1227. [5] T. Tetsui, K. Shindo, S. Kobayashi, M. Takeyama, A newly developed hot worked TiAl alloy for blades and structural components, Scripta Mater. 47 (6) (2002) 399–403. [6] R.M. Imayev, V.M. Imayev, M. Oehring, F. Appel, Alloy design concepts for refined gamma titanium aluminide based alloys, Intermetallics 15 (4) (2007) 451–460. [7] M. Kastenhuber, B. Rashkova, H. Clemens, S. Mayer, Enhancement of creep properties and microstructural stability of intermetallic β-solidifying γ-TiAl based alloys, Intermetallics 63 (2015) 19–26. [8] Y.W. Kim, S.L. Kim, Effects of microstructure and C and Si additions on elevated temperature creep and fatigue of gamma TiAl alloys, Intermetallics 53 (2014) 92–101. [9] E. Schwaighofer, H. Clemens, J. Lindemann, A. Stark, S. Mayer, Hot-working behavior of an advanced intermetallic multi-phase γ-TiAl based alloy, Mater. Sci. Eng. A 614 (2014) 297– 310. [10] T. Klein, B. Rashkova, D. Holec, H. Clemens, S. Mayer, Silicon distribution and silicide precipitation during annealing in an advanced multi-phase γ-TiAl based alloy, Acta Mater. 110 (2016) 236–245. [11] R.C. Pond, P. Shang, T.T. Cheng, M. Aindow, Interfacial dislocation mechanism for diffusional phase transformations exhibiting martensitic crystallography: formation of TiAl + Ti3Al lamellae, Acta Mater. 48 (5) (2000) 1047–1053. [12] S. Zghal, M. Thomas, S. Naka, A. Finel, A. Couret, Phase transformations in TiAl based alloys, Acta Mater. 53 (9) (2005) 2653–2664. [13] F. Appel, R. Wagner, Microstructure and deformation of two-phase γ-titanium aluminides, Mater. Sci. Eng. R 22 (5) (1998) 187–268. [14] Y.-W. Kim, Strength and ductility in TiAl alloys, Intermetallics 6 (7-8) (1998) 623–628. [15] D.M. Dimiduk, P.M. Hazzledine, T.A. Parthasarathy, S. Seshagiri, M.G. Mendiratta, The role of grain size and selected microstructural parameters in strengthening fully lamellar TiAl alloys, Metall. Mater. Trans. A 29 (1998) 37–47. [16] K. Maruyama, R. Yamamoto, H. Nakakuki, N. Fujitsuna, Effects of lamellar spacing, volume fraction and grain size on creep strength of fully lamellar TiAl alloys, Mater. Sci. Eng. A 239240 (1997) 419–428. [17] T.A. Parthasarathy, M. Keller, M.G. Mendiratta, The Effect of Lamellar Lath Spacing on the Creep Behavior of Ti-47at% Al, Scripta Mater. 38 (7) (1998) 1025–1031. - 23 -

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

[18] B. Viguier, J. Bonneville, J.L. Martin, The mechanical properties of single phase Ti47Al51Mn2 polycrystals, Acta Mater. 44 (11) (1996) 4403–4415. [19] J. Beddoes, W. Wallace, L. Zhao, Current understanding of creep behaviour of near γ-titanium aluminides, Int. Mater. Rev. 40 (5) (1995) 197–217. [20] G. Viswanathan, V. Vasudevan, M. Mills, Modification of the jogged-screw model for creep of γ-TiAl, Acta Mater. 47 (5) (1999) 1399–1411. [21] J. Malaplate, D. Caillard, A. Couret, Interpretation of the stress dependence of creep by a mixed climb mechanism in TiAl, Philos. Mag. 84 (34) (2004) 3671–3687. [22] M. Castillo-Rodríguez, M.L. Nó, J.A. Jiménez, O.A. Ruano, J. San Juan, High temperature internal friction in a Ti–46Al–1Mo–0.2Si intermetallic, comparison with creep behaviour, Acta Mater. 103 (2016) 46–56. [23] M.A. Morris, M. Leboeuf, Analysis of thermal and athermal deformation mechanisms during creep of γ-TiAl alloys, Mater. Sci. Eng. A 239-240 (1997) 429–437. [24] F. Appel, U. Christoph, M. Oehring, Creep deformation in two-phase titanium aluminide alloys, Mater. Sci. Eng. A 329-331 (2002) 780–787. [25] F. Appel, U. Lorenz, M. Oehring, U. Sparka, R. Wagner, Thermally activated deformation mechanisms in micro-alloyed two-phase titanium aluminide alloys, Mat. Sci. Eng. A 233 (1997) 1–14. [26] F. Appel, Diffusion assisted dislocation climb in intermetallic gamma TiAl, Mater. Sci. Eng. A 317 (1-2) (2001) 115–127. [27] W. Schillinger, H. Clemens, G. Dehm, A. Bartels, Microstructural stability and creep behavior of a lamellar γ-TiAl based alloy with extremely fine lamellar spacing, Intermetallics 10 (5) (2002) 459–466. [28] X. Zhang, N. Hansen, A. Godfrey, X. Huang, Dislocation-based plasticity and strengthening mechanisms in sub-20 nm lamellar structures in pearlitic steel wire, Acta Mater. 114 (2016) 176–183. [29] T. Maity, B. Roy, J. Das, Mechanism of lamellae deformation and phase rearrangement in ultrafine β-Ti/FeTi eutectic composites, Acta Mater. 97 (2015) 170–179. [30] T. Maity, A. Singh, A. Dutta, J. Das, Microscopic mechanism on the evolution of plasticity in nanolamellar γ-Ni/Ni5Zr eutectic composites, Mater. Sci. Eng. A 666 (2016) 72–79. [31] L. Cha, H. Clemens, G. Dehm, Microstructure evolution and mechanical properties of an intermetallic Ti-43.5-4Nb-1Mo-0.1B alloy after ageing below the eutectoid temperature, Int. J. Mater. Res. 102 (6) (2011) 703–708. [32] J. San Juan, Mechanical spectroscopy, Mater. Sci. Forum 366-368 (2001) 32–73. [33] I. Gutiérrez-Urrutia, M.L. Nó, E. Carreño-Morelli, B. Guisolan, R. Schaller, J. San Juan, High performance very low frequency forced pendulum, Mater. Sci. Eng. A 370 (1-2) (2004) 435– 439. [34] P. Simas, J. San Juan, R. Schaller, M.L. Nó, High-Temperature Mechanical Spectrometer for Internal Friction Measurements, Key Eng. Mater. 423 (2009) 89–95. [35] G. Fantozzi, Phenomenology and Definitions, Mater. Sci. Forum 366-368 (2001) 3–31. [36] A.S. Nowick, Berry B.S., Anelastic relaxation in crystalline solids, Academic Press Inc., New York, 1972. [37] L. Usategui, T. Klein, M.L. Nó, S. Mayer, H. Clemens, J. San Juan, Study by mechanical spectroscopy of the relaxation phenomena due to mobility of defects in a Ti-Al-Nb-Mo-C-Si (TNM+) alloy (2017) to be published. [38] L. McCusker, R.B. von Dreele, D.E. Cox, D. Louёr, P. Scardi, Rietveld refinement guidelines, J. Appl. Crystallogr. (32) (1999) 36–50. [39] T. Klein, H. Clemens, S. Mayer, Advancement of Compositional and Microstructural Design of Intermetallic γ-TiAl Based Alloys Determined by Atom Probe Tomography, Materials 9 (9) (2016) 755. - 24 -

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

[40] J. Guyon, A. Hazotte, F. Wagner, E. Bouzy, Recrystallization of coherent nanolamellar structures in Ti48Al2Cr2Nb intermetallic alloy, Acta Mater. 103 (2016) 672–680. [41] L. Cha, C. Scheu, H. Clemens, H.F. Chladil, G. Dehm, R. Gerling, A. Bartels, Nanometerscaled lamellar microstructures in Ti–45Al–7.5Nb–(0; 0.5)C alloys and their influence on hardness, Intermetallics 16 (7) (2008) 868–875. [42] A.G. Evans, R.D. Rawlings, The Thermally Activated Deformation of Crystalline Materials, Phys. Stat. Sol. 34 (9) (1969) 9–31. [43] F. Appel, H. Clemens, F.D. Fischer, Modeling concepts for intermetallic titanium aluminides, Progr. Mater. Sci. 81 (2016) 55–124. [44] J. San Juan, M. Castillo-Rodríguez, L. Usategui, P. Simas, O.A. Ruano, J.A. Jiménez, H. Clemens, S. Mayer, M.L. Nó, A new methodology to study creep behaviour of materials by mechanical spectroscopy, Int. Conf. on CREEP, Toulouse, France, May 31 - June 4 (2015). [45] M. Weller, G. Haneczok, H. Kestler, H. Clemens, Internal friction of γ-TiAl-based alloys with different microstructures, Mater. Sci. Eng. A 370 (1-2) (2004) 234–239. [46] J. San Juan, P. Simas, T. Schmoelzer, H. Clemens, S. Mayer, M.L. Nó, Atomic relaxation processes in an intermetallic Ti–43Al–4Nb–1Mo–0.1B alloy studied by mechanical spectroscopy, Acta Mater. 65 (2014) 338–350. [47] S. Karthikeyan, M.J. Mills, The role of microstructural stability on compression creep of fully lamellar γ-TiAl alloys, Intermetallics 13 (9) (2005) 985–992. [48] X.-W. Du, J. Zhu, B. Wang, Y.-W. Kim, Creep‐induced phase transformations in a Ti-Al alloy, Philos. Mag. A 82 (1) (2002) 39–50. [49] E. Schwaighofer, P. Staron, B. Rashkova, A. Stark, N. Schell, H. Clemens, S. Mayer, In situ small-angle X-ray scattering study of the perovskite-type carbide precipitation behavior in a carbon-containing intermetallic TiAl alloy using synchrotron radiation, Acta Mater. 77 (2014) 360–369. [50] T. Klein, M. Schachermayer, F. Mendez-Martin, T. Schöberl, B. Rashkova, H. Clemens, S. Mayer, Carbon distribution in multi-phase γ-TiAl based alloys and its influence on mechanical properties and phase formation, Acta Mater. 94 (2015) 205–213. [51] R. Kainuma, Y. Fujita, H. Mitsui, I. Ohnuma, K. Ishida, Phase equilibria among α (hcp), β (bcc) and γ (L10) phases in Ti–Al base ternary alloys, Intermetallics 8 (8) (2000) 855–867. [52] M.E. Kassner, Fundamentals of Creep in Metals and Alloys, 3rd ed., Elsevier Science, Amsterdam, 2015. [53] R.E. Schafrik, Dynamic elastic moduli of the titanium aluminides, Metall. Trans. A 8 (6) (1977) 1003–1006. [54] Y.-W. Kim, Ordered intermetallic alloys, part III: Gamma titanium aluminides, JOM 46 (7) (1994) 30–39. [55] Y. Mishin, C. Herzig, Diffusion in the Ti–Al system, Acta Mater. 48 (3) (2000) 589–623. [56] J.D. Eshelby, The interaction of kinks and elastic waves, Proc. R. Soc. Lond. A 266 (1962) 222–246. [57] J.P. Hirth, J. Lothe, Theory of dislocations, John Wiley & Sons, New York, 1982. [58] D. Caillard, J.L. Martin, Thermally activated mechanisms in crystal plasticity, Elsevier Science Ltd, Amsterdam, 2003. [59] D. Holec, R.K. Reddy, T. Klein, H. Clemens, Preferential site occupancy of alloying elements in TiAl-based phases, J. Appl. Phys. 119 (20) (2016) 205104. [60] F. Appel, U. Christoph, Coherency stresses and interface-related deformation phenomena in two-phase titanium aluminides, Intermetallics 7 (1999) 1173–1182. [61] C.E. Wen, K. Yasue, J.G. Lin, Y.G. Zhang, C.Q. Chen, The effect of lamellar spacing on the creep behavior of a fully lamellar TiAl alloy, Intermetallics 8 (5-6) (2000) 525–529. [62] J. Wang, T. Nieh, The role of ledges in creep of TiAl alloys with fine lamellar structures, Acta Mater. 46 (6) (1998) 1887–1901. - 25 -

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

[63] L.M. Hsiung, T.G. Nieh, Creep deformation of fully lamellar TiAl controlled by the viscous glide of interfacial dislocations, Intermetallics 7 (7) (1999) 821–827. [64] P.M. Hazzledine, Coherency and loss of coherency in lamellar Ti-Al, Intermetallics 6 (7-8) (1998) 673–677. [65] L. Zhao, K. Tangri, TEM investigation on the interfacial boundaries in as-cast Ti3Al+TiAl alloy, Acta Metall. Mater. 39 (10) (1991) 2209–2224. [66] D.G. Pettifor, R. Podloucky, The structures of binary compounds: II. Theory of the pd-bonded AB compounds, J. Phys. C: Solid State Phys. 19 (1986) 315–330. [67] A. Dlouhý, K. Kuchařová, Creep and microstructure of near-gamma TiAl alloys, Intermetallics 12 (7-9) (2004) 705–711. [68] J.N. Wang, T.G. Nieh, Effect of minor Si addition on the creep of a fully lamellar TiAl alloy, Scripta Mater. 37 (10) (1997) 1545–1550. [69] F.-S. Sun, F. Froes, Precipitation of Ti5Si3 phase in TiAl Alloys, Mater. Sci. Eng. A 328 (1-2) (2002) 113–121. [70] D. Seidman, On the Genesis of Nuclei and Phase Separation on an Atomic Scale, MRS Bulletin 34 (2009) 537–542.

- 26 -

ACCEPTED MANUSCRIPT Figures captions

RI PT

Fig. 1. SEM images (BSE mode) and TEM BF micrographs of the fully lamellar microstructure. In (a) the microstructure after the solution heat-treatment and the annealing procedure is discernible. The microstructure consists of colonies comprising of fine lamellar structures with only little βo phase in between. Further, regions of discontinuous precipitation (DP) are evident situated along colony boundaries. ζ-silicides present in the microstructure can be discriminated by their bright appearance. In (b) and (c) regions of DP in higher magnification are evident. Clearly, a globularization of the microstructural constituents can be observed. The colonies’ interior (d) consists of a nano-scaled lamellar structure with even interfaces (see inset).

M AN U

SC

Fig. 2. Tensile creep curves determined for the fully lamellar TNM+ alloy at (a) a constant temperature of 800 °C and (b) a constant stress of 200 MPa. Minimum creep rates evaluated are given next to the respective curves. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 3. Internal friction (IF) spectra of the nano-lamellar TNM+ alloy at 0.03 Hz (magenta) and 0.1 Hz (blue). The good agreement of experimental curves and fitted HTB curves (cyan and violet) is visible (see text). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

EP

TE D

Fig. 4. TEM BF micrographs and STEM HAADF image of the crept specimen condition. In (a) and (b) the lamellar structure is visualized, whereby degradation is clearly visible by an increasing interface roughness (macro-ledges (1), protrusions (2)) and the occurrence of terminating lamellae (3). Image (c) corresponds to a detail of the lamellar structure, where ledges along the interfacial boundary are discernible. One of these has been highlighted with a red circle and the interfacial structure is represented by the red line. Image (d) evidences the formation of precipitates in the lamellar structure, which are only visible when tilting the specimen. Using HAADF imaging, precipitates are found to form in the interior of the α2 phase as visible in (e). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

AC C

Fig. 5. APT reconstructions of the microstructure (a) prior and (b) after creep at 800 °C and 150 MPa for ≈ 375 h. The lamellar structure is clearly delineated by the differences in Al concentration. Prior to creep no precipitates are discernible in the lamellar structure in agreement to the TEM investigations. After creep ζL-silicides and βo,L precipitates that nucleated and grew at the expense of the α2 lamellae are visible. For the purpose of delineating predominating transformation products the alloying elements Mo and Si have been indicated. The ζL-silicide is represented by a cSi = 10 at.% isoconcentration surface and the βo,L particles is indicated by a cMo = 2 at.% isoconcentration surface. Quantitative chemical compositions of the lamellae prior and after creep as well as of the precipitates can be found in Table 2. As marked by an arrow a fine γ lamellae terminates at the ζL precipitate. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

- 27 -

ACCEPTED MANUSCRIPT Fig. 6. Proximity histogram for Mo and Si calculated from the APT data of the crept specimen condition. Both elements tend to accumulate in the α2 phase. Upon creep exposure pile-ups in the vicinity of the interface on the α2 side are visible, which correspond to the direction of interfacial displacement. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

SC

RI PT

Fig. 7. Scheme of the proceeding interfacial processes after Ref. [62] and precipitation phenomena. In (a) an interface between γ and α2 is displayed. Creep conditions result in positive or negative ledge migration, which occurs accompanied by dislocation migration by jog-pair formation and climb to the newly created terrace. Simultaneously, phase compositions are adjusted by elemental diffusion, particularly of strongly partitioning elements, such as Mo and Si, resulting in interfacial pile-ups (see Fig. 6); (b) eventually, these accumulations culminate in the precipitation of βo,L particles and ζL-silicides. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

AC C

EP

TE D

M AN U

Fig. 8. Concentration changes of alloying elements Mo and Si within the α2 and γ phases upon creep exposure. Mo and Si concentrations are reduced in the α2 phase, while they remain similar in the γ phase. The respective concentration reductions are due to the precipitation of βo,L particles and ζL-silicides within the α2 phase, which are rich in Mo and Si, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

- 28 -

ACCEPTED MANUSCRIPT Tables Table 1. Phase fractions as derived from XRD and quantitative SEM image analysis as well as Vickers hardness values prior and after creep.

HV10 ± σ

γ

α2

βo

Before creep

77.5

22

< 0.5

480 ± 3

After creep

78.5

20.5

<1

470 ± 3

RI PT

Phase fraction [vol.%]

Phase

Element concentration [at.%] Ti

Nb

Mo

C

Si

Other

3.92

0.82

0.18

0.21

0.27

3.70

1.26

0.74

0.65

1.05

γ

50.9 43.7

α2

61.8 30.8

γ

51.4 43.1

4.09

0.80

0.19

0.17

0.25

α2

63.2 29.8

3.95

1.09

0.53

0.56

0.87

β o ,L

55.4 32.0

5.41

6.44

0.08

0.31

0.36

ζL

51.3 9.82

8.82

1.27

2.55

25.4

0.84

AC C

EP

After creep

Al

TE D

Before creep

M AN U

SC

Table 2. Quantitative chemical analysis based on APT data of the constituent phases before and after creep. The chemical composition of the γ phase changes very little during exposure, while the composition of the α2 phase changes noticeably. In particular the respective concentrations of the elements that can be found in the precipitates, βo,L and ζL, Mo and Si are reduced in the α2 phase during exposure as these particles form from this phase. Moreover, consistently with literature [50] the βo phase shows a low solubility for interstitial element C, while a substantial amount of Si is dissolved. The ζL-silicide observed is enriched with Al and Nb and, moreover, with C in comparison to its stoichiometric composition of Ti5Si3.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT