International Journal of Refractory Metals & Hard Materials 73 (2018) 79–84

Contents lists available at ScienceDirect

International Journal of Refractory Metals

& Hard Materials
journal homepage: www.elsevier.com/locate/IJRMHM

Rotary friction welding of molybdenum components T

a,⁎ a b b c b a
M. Stütz , F. Pixner , J. Wagner , N. Reheis , E. Raiser , H. Kestler , N. Enzinger
a
IMAT, Graz University of Technology, Austria
b
Plansee SE, 6600 Reutte, Austria
c
Klaus Raiser GmbH & Co. KG, Germany

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

Keywords: Joining of TZM components by inertia rotary friction (IFW) welding is an established industrial process for
Refractory metals welding cross-sections up to 1500 mm2. Up-scaling to medium-size components up to 5000 mm2 in a direct drive
Molybdenum friction welding process requires a better understanding of the influencing factors of the weld procedure, e.g.
TZM machine parameters, weld preforms, and the clamping system. Based on the existing IFW process for TZM tubes,
Welding
welding parameters were transferred to tubular components of pure molybdenum (OD: 150 mm, ID: 130 mm,
Rotary friction welding
4400 mm2). Successful welds were produced showing a fine-grained, defect-free microstructure. However,
molybdenum proved to be more challenging than TZM. Particularly high upset rates and motor overload oc-
curred during the friction phase. Therefore, a study was carried out to reveal the underlying mechanisms with
small-size samples under laboratory conditions. It was shown that extensive plastic deformation of the entire
weld zone occurred due to higher thermal diffusivity and lower strength of molybdenum compared to TZM. This
high upset rate reduces the process window for a reproducible welding procedure significantly. In conclusion, a
concentrated energy input during the transition from friction to forge phase is required to countervail the high
thermal diffusivity of molybdenum. Based on these observations, the feasibility of friction welding of medium-
size molybdenum tubes will be discussed.

1. Introduction [4]. Therefore, minimizing impurities is an aim in the production of Mo

Rotary friction welding (RFW) is a solid state joining process, where
the required heat is generated by friction between the joining partners.
When the shear strength is reached, plastic deformation occurs and the
characteristic weld flash starts to form. In general, two concepts exist:
inertia friction welding (IFW) and direct drive friction welding (DDFW).
In IFW the rotary part and an attached flywheel are accelerated, then
the drive is decoupled and the welding partners are brought in contact.
The speed of the parts decelerates during the whole welding process
continuously. Towards the rotary halt, the axial force is increased. In
direct drive friction welding (DDFW), a motor drives the rotary welding
part during the whole process and at the braking of the rotation, the
axial pressure is increased whereby the flash is fully formed and the
process is completed shortly after the spindle halt. Based on the his-
torical development of the friction welding process, IFW is mainly used
in North America, while in Europe DDFW is predominant [1–3].
Molybdenum is a refractory metal and shows unique properties,
such as high melting point, high thermal conductivity and low thermal
expansion coefficient. However, Mo is brittle at room temperature and
highly susceptible to brittle failure due to impurities, most of all oxygen
and Mo-based alloys [4,5]. The dispersion strengthened alloy TZM is
the most used Mo-alloy. Small additions of titanium, zirconium, and
carbon lead to the formation of carbides and act as getter elements for
oxygen. As a further consequence, recrystallization temperature and
high temperature strength are increased compared to technical pure Mo
[4]. In the temperature range between 900 °C and 1500 °C the ultimate
tensile strength (UTS) of TZM is up to five times higher compared to
Mo, and the service temperature for TZM is specified up to 1400 °C [6].
Mo and Mo-base alloys generally exhibit limited weldability. Fusion
welding of molybdenum and TZM causes an embrittlement of the heat
affected zone due to grain coarsening and precipitated impurities at the
grain boundaries [7,8]. In contrary, the technique of friction welding
avoids the typical defects of fusion welding and the resulting embrit-
tlement. Since the grain size is reduced by high deformation and de-
formation rates during the process, the strength and toughness are
improved compared to fusion welding processes [9,10]. An example for
the industrial application of friction welding is the manufacturing of
TZM target-stem assemblies for X-ray tubes. IFW is used to join these
TZM-components, where the weld energy is provided by the rotating
flywheel. Typical cross-sections are in the range up to 1500 mm2 TZM

⁎
Corresponding author.
E-mail address: markus.stuetz@tugraz.at (M. Stütz).

https://doi.org/10.1016/j.ijrmhm.2018.02.004
Received 29 November 2017; Received in revised form 22 January 2018; Accepted 4 February 2018
Available online 06 February 2018
0263-4368/ © 2018 Published by Elsevier Ltd.
M. Stütz et al. International Journal of Refractory Metals & Hard Materials 73 (2018) 79–84

[10,11]. applied upon deceleration start. This is common practice in friction

The aim of this work is to weld pure molybdenum tubes with cross- welding and referred to as “forging into the turning spindle” [3].
sections up to 5000 mm2 on a DDFW-machine. Process parameters were 2. Forge force gradient: the rate of force increase from friction force to
investigated based on the experience with IFW of TZM. However, pre- forge force.
vious investigations have shown that DDFW of Mo is challenging. 3. Spindle deceleration: the deceleration rate from welding speed to
Preliminary spindle halt, insufficient drive power and clamping pro- spindle stop.
blems occurred. A detailed parameter study was not possible [12].
Hence, experiments with small samples of Mo and TZM were in- In order to derive information about the upscaling to larger cross-
vestigated under laboratory conditions in order to: sections, the specific energy was calculated from the motor load and
weld speed. The in-situ pre-heating phase was not included in this
• Identify reasons for the occurring motor overload and preliminary calculation, since the heat dissipation reached an equilibrium with the
spindle stall. inserted power by friction and caused no significant sample deforma-
• Examine the influence of weld parameters on the joint quality tion. The calculated values therefore represent the energy input during
• Examine the influence of the material by comparing Mo and TZM. the friction and forging phase.
The welded samples were inspected visually to evaluate the flash
The up-scaling of the process requires information about the power shape, the geometric symmetry of the weld and the visual appearance.
and energy demand during the RFW process. In IFW, the energy E that Metallographic investigations were carried out on welded samples
is converted into heat in the friction surface originates from the inertia using light optical microscopy to investigate the quality and the mi-
moment J of the flywheel, and can be calculated according to: crostructure of the welds.
1
E (t ) = ·J ·ω (t )2 3. Results & discussion
2 (1)
Scaling of IFW processes is done using the specific weld energy and 3.1. Molybdenum medium-size format
keep it constant [13]:
E Welding experiments with different parameters (weld speed, fric-
= const.
A (2) tion pressure) were conducted; however, regardless of the welding
parameters, the experiments showed a preliminary spindle stop due to
In DDFW, the weld energy, can be calculated from the product of
insufficient motor power. The results have been published previously
motor load, weld speed, and time [1]. The specific weld energy and
[12] and are summarized below showing the welding parameters of one
weld parameters of the laboratory scale experiments were used to
representative weld in Fig. 2. The spindle speed decreases immediately
propose a DDFW process for Mo-tubes of 5000 mm2 cross-section.
when the friction phase starts. The exponential increase of the upset
during the friction phase with an upset rate up to 7 mm/s should be
2. Experiments
mentioned. Fig. 3 shows an image of the friction welded Mo-tube. Al-
though the weld flash indicates a good weld, defects were found in
The investigated materials were commercially pure Mo (> 99.97 wt
metallographic cross-sections.
%) and TZM (0.5% Ti, 0.08% Zr, 0.01–0.04% C, all wt%). The pro-
duction route of the material was sintering, hot forging, stress-relieve
3.2. Molybdenum and TZM small-size formats
annealing, and machining.
Welding experiments with medium-size Mo tubes of 4400 mm2
The laboratory scale experiments enabled the variation and mea-
cross section with an outer diameter (OD) of 150 mm, and an inner
surement of the welding parameters: friction pressure, welding speed,
diameter (ID) of 130 mm were performed. The welding machine was a
and forge condition. Fig. 4 shows the influence of the specific power on
heavy-duty DDFW machine with 150 kW drive power and 1250 kN
the upset behavior of Ø12 mm Mo rods for three different weld speeds.
maximum forge force. Axial force, speed, upset, and motor load para-
The specific weld power, and consequently the specific weld energy,
meters were recorded. Investigated parameters were friction pressure,
decreases with decreasing weld speed. As a result, the upset decreases
welding speed, and friction time. The Mo tubes were pre-heated in a
from 5.2 mm at 100% weld speed to 3.9 mm and 2.7 mm for 95% and
furnace to 400 °C.
91% weld speed, respectively.
In order to investigate the material behavior during the process a
The specific weld power curves of Mo in Fig. 4a show two separate
friction stir welding machine (MTS I-STIR BR4) was adapted for rotary
peaks that are the result of the friction torque during the RFW process.
friction welding (RFW). The machine capabilities were a maximum
The first peak can be attributed to the friction phase and results from
axial force of 35.6 kN, spindle speed 3200 min−1, 180 Nm torque, and
the interlocking and breaking of asperities and subsequent softening of
axial travel of 15 mm. Small-size friction welds with Ø12 mm rods of
the material by frictional heating. The second peak is the result of the
Mo and Ø9 mm rods of TZM material were conducted. To exceed the
forge phase where the torque rises with increasing forge force, and
ductile-brittle transition temperature, an in-situ pre-heating phase was
sharply falls off to zero with decreasing weld speed. No steady-state
employed [14]. Friction at low pressures (< 10 MPa) resulted in a
could be observed between these two peaks that would result from the
constant temperature of 350 °C in 8 mm distance from the weld inter-
balance between material softening and strain hardening [2].
face without significant plastic deformation. The welding process
Fig. 5 shows the upset as a function of the specific weld energy of all
started by increasing the pressure from pre-heating to friction pressure.
performed Mo welds. Insufficient bonding can be observed if the spe-
Fig. 1 shows the input parameters pressure and speed for Mo and TZM.
cific energy is low and the resulting upset is < 4 mm. It is assumed
It should be noted that TZM requires a higher friction and forge pres-
that < 4 mm upset the impurities are not completely expelled into the
sure than Mo to produce successful welds due to the higher strength.
flash and, therefore, the joint quality is poor. If the welding energy is
Varied welding parameters were friction pressure, welding speed, and
increased above 120%, excessive plastic deformation leads to an in-
forge condition to determine optimal welding conditions. The forge
stable process with high upset rates and upsets > 12 mm. Between
condition is determined by three parameters:
4 mm and 12 mm upset, sound welds were produced and the image
insert in Fig. 5 shows an example. The solid line represents process
1. Forge point: the initiation of the forge phase in relation to the cur-
conditions with maximum spindle deceleration and low forge force
rent spindle speed. In the small-size experiments the forge force was
gradient. If the forge force gradient is increased, the upset increases at

80
M. Stütz et al. International Journal of Refractory Metals & Hard Materials 73 (2018) 79–84

Fig. 1. (a) Input pressure course for Mo and TZM samples with indicated phases: Pre-heating, friction, and forging. (b) Friction speed was varied between 91% and 100% speed to adjust
weld energies.

almost constant specific weld energy. The more intense forge force
gradient resulted in sound joints (dashed line), but also in instable forge
conditions with an asymmetric weld flash and an angular misalignment
of the welding interface (triangle marks). The results show that rotary
friction welding of Mo is sensitive to small changes of the process
parameters. The parameter window for sound welds is small and poor
bonding or instable plastic deformation or instable forging may occur
easily. The specific weld energy has a significant influence on the upset
and weld quality, but the forge conditions (spindle deceleration and
forge force gradient) as well.
The TZM experiments were performed analogously to the Mo ex-
periments, but almost a twice as high level of friction- and forge pres-
sure (see Fig. 1) was required for successful welds due to the higher
strength of TZM in comparison to Mo. TZM shows a similar course of
the specific weld power as Mo with two distinct peaks (Fig. 6a). Fur-
thermore, lower weld speed results in lower weld energy and lower
upsets (Fig. 6b). Fig. 3. Medium-size friction welded Mo-tube (OD: 150 mm, ID: 130 mm).
TZM allows a larger variation of welding parameters compared to
Mo to produce good welds. It was possible to vary the friction pressure,
TZM welds. A sound weld zone and flash formation is observed in the
force gradient, and spindle deceleration. The correlation between spe-
polished sections. The etched cross-sections show the typical micro-
cific weld energy and upset for TZM is shown in Fig. 7. Analogous to
structure of friction welded materials with the transition from severe
Mo, low specific energy resulted in upsets < 4 mm and insufficient
plastic deformation in the contact zone to the unchanged base material.
bonding of the welds. Very high specific weld energies caused up-
The elongation of the grains exhibits the direction of plastic flow [2].
sets > 12 mm and the process became instable due to excessive plastic
Fig. 8 also shows a hardness profile across the weld for (a) Mo and (b)
deformation. The experiments with TZM show that higher friction
TZM. The hardness profile shows no significant differences between the
pressures are required compared to Mo. Similar to Mo, the specific weld
base material and the weld zone.
energy may be suitable to compare welds with identical forge condi-
Fig. 9a and b shows light optical micrographs of the welded samples
tions. The alteration of force gradient and deceleration influence the
near the weld interface of Mo and TZM, respectively. The contact zones
resulting upset independently from specific weld energy.
of Mo and TZM (Fig. 9c and d) exhibit a fine-grained microstructure
Fig. 8 shows light optical micrographs of representative Mo and
caused by severe plastic deformation and recrystallization. Fig. 9e and f

Fig. 2. Welding parameters of Mo welding trial, pre-heated to 400 °C/900 min−1: (a) axial pressure, (b) weld speed, and (c) upset.

81
M. Stütz et al. International Journal of Refractory Metals & Hard Materials 73 (2018) 79–84

Fig. 4. Influence of the weld speed (91%, 95%, and 100%) of Ø12 mm Mo samples on the (a) specific weld power and (b) upset (2.7 mm, 3.9 mm, and 5.2 mm).

Fig. 5. Influence of the specific weld energy on the upset of Ø12 mm Mo rods. Sound Fig. 7. Influence of the specific weld energy on the upset of Ø9 mm TZM rods for different
joints were achieved with high spindle deceleration and low forge force gradient (solid weld parameters: low and high friction pressure, maximum forge force gradient, and
line) and partially with higher forge force gradients (dashed line). reduced braking gradient.

Fig. 6. Influence of the welding speed (91%, 95%, and 100%) of Ø9 mm TZM samples on the (a) specific weld power and (b) upset (3.9 mm, 5.4 mm, and 7.0 mm).

82
M. Stütz et al. International Journal of Refractory Metals & Hard Materials 73 (2018) 79–84

(a) (b)
Fig. 8. Micrographs of friction welded (a) Mo (100% speed) and (b) TZM (100% speed). The left and right pictures show a polished and an etched cross-section, respectively. The position
of the hardness profiles is indicated by the white lines. The white boxes show the location of the details in Fig. 7 (see below).

exhibit grains oriented towards the direction of plastic flow. The mi-
crostructure becomes coarser because strain and strain rate are lower
compared to the adjacent contact zone [2].
3.3. Upset behavior of Mo and TZM
The laboratory experiments have shown that the process window
for sound welds is small for Mo and TZM. Particularly, Mo is sensitive to
high specific energies and high forge force gradients that cause high

(e)

(f)
(c)

(c)
(d

(a) (b)

(c) (d)

(e) (f)

Fig. 9. Detail of the microstructure friction welded samples: Overview of (a) Mo and (b) TZM with details (c, d) of the contact zone with severe plastic deformation and (e, f) partly
deformed zone.

83
M. Stütz et al.
Fig. 10. Original process for medium-size Mo-tubes (solid line) and the proposed process
adaptions (dashed line): (I) Pre-heating phase at low friction pressure, (II) raised friction
pressure, (III) softer forge force gradient, (IV) forge point “forging into the turning
spindle,” (V) shorter friction time.
upsets and instable plastic deformation. It is assumed that during the
friction phase, the temperature rises very fast over a large volume due
to the high thermal diffusivity of Mo (a = 53 mm2/s). The yield
strength decreases within this volume and, if the friction pressure ex-
ceeds the yield strength, the material starts to flow and causes the high
upsets and upset rates. Therefore, the deformed volume should be kept
to a minimum to ensure a stable flash formation [15]. In comparison,
low-carbon steel with a four times lower thermal diffusivity
(a = 14 mm2/s) exhibits lower upset rates. In a previous work, upset
rates up to 7 mm/s were observed for Mo in comparison to 0.1 mm/s for
steel [12]. The dispersion strengthened alloy TZM shows a comparable
thermal diffusivity, but a higher strength than Mo. Therefore, RFW of
TZM enables a wider variation of welding parameters, such as higher
forge force gradients and friction pressures. It can be concluded that
friction welding of Mo and TZM requires short process times with high
energy input to concentrate the energy in the weld zone.
3.4. Upscaling from small size to medium size
Fig. 10 shows the input parameters of the conducted medium-size
experiments. From the small-size experiments, a series of process im-
provements are suggested and indicated by dashed lines in the figure:
I. A pre-heating phase should be inserted to safely and reproducibly
exceed the DBTT prior to welding. This can be realized by friction at
low pressures that cause pre-heating but no plastic deformation.
II. The friction pressure should be raised, analog to values used in the
small-size experiments.
III. The transition from friction force to forge force must follow a soft
gradient.
IV. The forge phase start must coincide with the spindle braking start to
create a “forging into the turning spindle” condition. The decel-
eration remains unchanged at maximum achievable deceleration.
V. The friction time should be reduced because of the increased fric-
tion pressure.
With these improvements, DDFW of medium-size Mo-tubes should
be possible and reproducible. However, up-scaling from 113 mm2 to
4400 mm2 increases the required weld energy accordingly that largely
exceeds the used 150 kW RFW machine in the medium-size experi-
ments.
4. Conclusions
Friction welding is established for many different materials like
steels, or titanium alloys, among others. Compared to these materials,
84
International Journal of Refractory Metals & Hard Materials 73 (2018) 79–84
Mo and TZM proved to be more challenging to establish a friction
welding process due to their different thermal and mechanical prop-
erties, most of all their high thermal diffusivity. This work presents
experiments using medium-size tubes on an industrial DDFW machine
and small-size samples on an adapted laboratory machine. The fol-
lowing was revealed:
• TZM requires higher axial pressures than Mo to achieve comparable
plastic deformation due to the higher strength of TZM compared to
Mo.
• It is assumed that the temperature increases very fast in a large
volume due to the high thermal diffusivity of Mo and TZM. The yield
strength decreases within this volume and the material starts to flow
and causes the high upsets and upset rates. This narrows the oper-
ating window for a successful process.
• Upset thresholds were determined between 4 mm and 12 mm for
small-size specimens of Mo and TZM. Below 4 mm insufficient
bonding was observed, above 12 mm the weld process became un-
stable.
• The specific weld energy was an insufficient parameter to com-
pletely describe the DDFW process. If the forge condition (forge
force gradient and spindle deceleration) was changed, the resulting
upset changed despite similar specific weld energies. These devia-
tions demonstrate that the manner of energy input—thus the forge
condition—is important for predict the welding process.
•
A decrease of the spindle deceleration results in an increased process
time and upset. Therefore, a fast spindle deceleration is preferred.
•
The maximum force gradient was applicable to TZM but not to Mo,
because the plastic deformation became unstable. With a low force
gradient successful welds were realized.
Acknowledgment
The K-Project Network of Excellence for Metal JOINing is fostered in
the frame of COMET – Competence Centers for Excellent Technologies
by BMWFW, MBVIT, FFG, Land Oberösterreich, Land Steiermark, Land
Tirol, and SFG. The program COMET is handled by FFG.
References
[1] M.M. Schwartz, Friction welding, Metals Joining Manual, McGraw-Hill Book
Company, 1979, pp. 6.1–6.51.
[2] M. Maalekian, Friction welding – critical assessment of literature, Sci. Technol.
Weld. Join. 12 (8) (Nov. 2007) 738–759.
[3] Manufacturing Technology Inc., Friction Welding, (1999).
[4] E.M. Savitskii, G.S. Burkhanov, Physical Metallurgy of Refractory Metals and
Alloys, Springer US, Boston, MA, 1970.
[5] M. Nagae, T. Yoshio, J. Takada, Y. Hiraoka, Improvement in recrystallization
temperature and mechanical properties of a commercial TZM alloy through mi-
crostructure control by multi-step internal nitriding, Mater. Trans. 46 (10) (2005)
2129–2134.
[6] Molybdän Werkstoffeigenschaften und Anwendungen, (2013).
[7] A.J. Bryhan, Joining of Mo Base Metal and Factors Which Influence Ductility, Exxon
Production Research Co., Houston, TX, 1986.
[8] S. Primig, H. Leitner, H. Clemens, A. Lorich, W. Knabl, R. Stickler, On the re-
crystallization behavior of technically pure molybdenum, Int. J. Refract. Met. Hard
Mater. 1 (2009) 1–9.
[9] A. Ambroziak, Friction welding of molybdenum to molybdenum and to other me-
tals, Int. J. Refract. Met. Hard Mater. 29 (4) (Jul. 2011) 462–469.
[10] B. Tabernig, N. Reheis, Joining of molybdenum and its application, Int. J. Refract.
Met. Hard Mater. 28 (6) (Nov. 2010) 728–733.
[11] G. Reznikow, Method for making a rotating anode with an integral shaft, Patent
US5,592,525; 1997.
[12] M. Stuetz, N. Enzinger, J. Wagner, N. Reheis, H. Kestler, E. Raiser, Rotary friction
welding of large molybdenum tubes, Proc. 10th Int. Conf. Trends Weld. Res. 2016,
pp. 153–156.
[13] V.R. Dave, M.J. Cola, G.N. Hussen, Heat generation in the inertia welding of dis-
similar tubes, Weld. J. 80 (10) (2001) 246s–252s.
[14] F. Pixner, M. Stütz, J. Wagner, N. Reheis, N. Enzinger, Comparing molybdenum and
steel in rotary friction welding using an FSW machine, 22. Erfahrungsaustausch
Reibschweissen, 2017.
[15] K. Mucic, F. Fuchs, N. Enzinger, Process optimization for linear friction welding of
high strength chain, EUROJOIN Conference, 2012, pp. 157–166.