J Fail. Anal. and Preven.

(2008) 8:3740
DOI 10.1007/s11668-007-9097-2

YP

R-R

Failure Analysis of Friction Weld (FRW) in Truck Axle

Application
Y. Huang Y. Zhu

Submitted: 17 September 2007 / in revised form: 16 November 2007 / Published online: 20 December 2007
ASM International 2007

Abstract A section of a fractured spindle from an axle

housing was metallurgically examined. The axle fractured
at the friction weld (FRW) interface and was from a truck
with only about 8,000 miles of service. SEM and metallographic examinations of the weld fracture revealed that a
ferritic band existed at the weld interface. A significant
amount of oxides and some micro-porosity was observed in
the ferrite. The existence of this type of microstructure
compromised the strength of the weld and resulted in the
overload fracture of the axle. FRW is supposed to be a
solid-state process; however, melting-solidification was
evident in this case, as suggested by the band of oxides
caused by liquid phase oxidation of the molten layer at the
interface and micro-shrinkage porosity formed during
solidification of the liquid film. Process conditions that
resulted in high heat input and liquidizing of the weld
interface before the upsetting stage of the friction welding
process should be mitigated.
Keywords Failure analysis
Friction weld

Axle housing
SEM

This article was published in Materials Science & Technology 2007

Conference and Exhibition: Exploring Structure, Processing, and
Applications Across Multiple Materials Systems (Detroit, MI),
September 1720, 2007, What We Can Learn from Failure Analysis
Symposium, ASM International, Materials Park, OH, 2007.
Y. Huang (&)
Key Safety Systems, Sterling Heights, MI, USA
e-mail: HuangY@KeySafetyinc.com
Y. Zhu
Ryobi Die Casting (USA), Shelbyville, IN, USA

Introduction
Friction welding (FRW) is a solid-state welding process in
which the heat for welding is produced by the relative
motion of the two interfaces being joined. In a typical FRW
process, a non-rotating work-piece is held in contact with
the rotating work-piece under constant or gradually
increasing pressure until the interface reaches the welding
temperature. The rotational speed, the axial pressure, and
the welding time are the principal variables that are controlled in order to provide the necessary combination of
heat and pressure to form the weld. These parameters are
adjusted so that the interface is heated and welding can
take place. Under normal conditions, no melting occurs at
the interface, although significant plastic deformation may
occur.
Because of FRWs capability to join dissimilar steel
grades [1], it has been widely used in axle applications,
where the spindles (medium carbon steel) are welded to the
axle housings (mild steel). This combination of material
selections would be somewhat difficult under traditional
fusion welding processes due to limited weldability of
medium carbon steel. After FRW, the axles are straightened and 100% eddy current inspected for possible
discontinuities.
A section of an axle housing was returned for failure
analysis because it fractured at the friction weld interface
after about 8,000 miles of service. As shown in Fig. 1, a
portion of slightly corroded area is evident on the bottom
side of the fracture. In a preliminary examination by a third
party, it was suggested that the presence of oxidation might
be an indication of a pre-existing crack that could have
occurred during the straightening process. However, this
possibility was discounted because it was unlikely that the
eddy current device would miss a discontinuity at this

123

38

J Fail. Anal. and Preven. (2008) 8:3740

Fig. 1 A fractured spindle

scale. Scanning electron microscope (SEM) examination of

the area with red rust revealed the same fracture characteristics of the rest of the fracture. Additionally, the lack of
paint on the fracture surface suggested that it was not a precrack, but rather had cracked during service.
Similar examination by another party concluded that no
material or welding defects were present to account for the
fracture. But the fracture was perplexing as FRW usually
imparts higher strength at the weld than the base materials.
Additionally, the weld was not the most highly stressed
area by design. Moreover, one of the important design
criteria was that no fracture should occur through the FRW
interface.

Fig. 2 An SEM photo, taken from the spindle side of fracture,

exhibiting fracture modes of different areas as indicated (*2009)

Observations and Discussions

The spindle fractured in an overload mode through the
welding interface and no evidence of fatigue was observed.
Fracture of friction welds through the weld interface is
rather unusual and is mostly like due to anomalies at the
weld that could have caused poor fusion. However, visual
examination of the fracture surface indicated the weld
exhibited reasonably good fusion, judging from bonding
area and weld curl size. Depending on the specific location
examined, the fracture manifested in mixed modes with
approximately half in ductile dimples and the other half in
cleavage as shown in Fig. 2.
Metallographic examination of the weld showed that the
fracture occurred at the interface where there was a distinctive ferrite band with oxides (gray spots) scattered
within it as shown in Fig. 3. SEM examination of the
corresponding area exhibited ductile dimple fracture with
numerous oxide particles situated at the centers of the
dimples (Fig. 4).
The ferrite layer was largely continuous and extended to
the weld curl as shown in Fig. 5. The microstructure of the
FRW interface from a reference sample is shown in Fig. 6.

123

Fig. 3 The assembly fractured along the weld interface (a ferrite

band with oxides), etched, 2% natal

Fig. 4 The weld interface exhibited ductile dimple fracture, SEM

image (1,0009)

J Fail. Anal. and Preven. (2008) 8:3740

Fig. 5 A ferritic layer near the weld curl at the inner diameter,
etched, 2% nital

Fig. 6 A typical microstructure at weld interface, etched, 2% nital

In comparison, there was no ferrite band at the weld

interface of the reference sample, which also exhibited a
finer microstructure, indicating it was subjected to less heat
input and/or faster cooling rate during the welding process.
The cooling condition in the fractured sample also promoted Widmanstatten ferrite formation as shown in Fig. 5.
The formation of a ferrite layer along the weld interface
would certainly reduce the weld strength, as would the
oxide particles. It would be of great interest to understand
the mechanism of formation of the oxide band and ferrite
layer so that the root cause of the fracture can be identified
and mitigated.
To better illustrate the morphology of the oxide particles, the oxide band under unetched condition is shown in
Fig. 7. Some had suggested that the existence of this oxide
band at the interface indicated excessive oxidation on the
housing and/or the spindle prior to welding. Under this
assumption, both housings and spindles with excessive

39

Fig. 7 The oxide band between two dashed red lines at the weld
interface under unetched condition

oxidation were joined with the same FRW parameters used

in the axle manufacturing. However, the subsequent
mechanical and metallographic testing found that the welds
were sound and no oxide band was observed.
It should be noted that the work pieces are sufficiently
cleaned and substantially free of rust in normal manufacturing process. As described earlier, the weld curl of the
fractured sample exhibited sufficient size, indicating adequate material push-out during the FRW process. Even
if there were some rust on the work pieces prior to the
FRW, it is unlikely the rust will be retained at the center of
work piece; rather it should be pushed out into the curl
instead.
Another argument against this assumption of excessive
work piece oxidation prior to the FRW is the morphology
of the oxide particles in the weld interface. They are in the
form of fine-dispersed particles in the ferrite layer. Any
prior-FRW oxidation would exist in filmy or globular form
in a larger scale at the interface if it were not pushed out
into the curl.
As shown in Figs. 5 and 6, the two work pieces exhibited mixture of pearlite and ferrite microstructure. The
spindle had a pearlite-bainite-dominated microstructure
while the housing a ferrite-dominated one. Under a normal
FRW process, the work pieces go through a solid-state recrystallization process where limited carbon diffusion
occurs at the weld interface. As a result, the interface
exhibited transitional microstructures that are somewhere
in between both base materials as shown in Fig. 6.
Unlike the weld interface in Fig. 6, the interface of the
fractured sample in Fig. 5 exhibits a continuous layer of
ferrite. This indicates a complete depletion of carbon in this
band and it is against the principles of solid-state carbon
diffusion at the weld interface. Additionally, the morphology of the oxides within the ferrite layer suggests it did

123

40

Fig. 8 Crack-like shrinkage (porosity) observed at the fracture

surface (759)

J Fail. Anal. and Preven. (2008) 8:3740

show any signs of void and crack coalescence and lack

metallurgical bonding.
The porosity was ruled out to be from the base materials
as no porosity was observed and the inclusion levels were
fairly low in both base materials. The above evidences
suggested that the weld interface underwent a re-solidification process and a continuous liquid film existed at the
weld interface before and during the upsetting stage of the
welding process. When the weld interface was heated to
the liquid state during the heating phase of the FRW process, the liquid film was oxidized rapidly. This led to the
formation of the ferrite layer with oxide particles dispersed
within the ferrite grains. During the solidification after the
forging phase, the liquid film contracted in a closed space
(between the housing and spindle) and resulted in the
formation of micro-shrinkage porosity within the weld
interface.
The melting appeared to have occurred on the housing
because the ferrite layer was only evident on the friction
weld curl on the housing but not on the curl on the spindle
side. This soft ferritic layer with the presence of the microshrinkage and inclusions/oxides apparently had detrimental
effects on the strength of the friction weld as evidenced by
the axle fracture. Further, the formation of Widmanstatten
ferrite decreased the impact strength of the weld.

Conclusions

Fig. 9 Close look of the feature within the red circle further
suggested that the feature was shrinkage (porosity) formed when
the liquid film at the weld interface solidified (1,0009)

not form in the solid-state either as oxidation would most

likely start from the grain boundaries in solids, instead of
inside grains. The above offered hints that the weld interface might have briefly heated to a liquid stage and resulted
in carbon depletion and liquid oxidation.
Porosity is observed in the ferrite band in Fig. 5. In
Fig. 8, crack-like shrinkage (porosity) is observed on parts
of the fracture surface. It is clear that the crack-like
shrinkage was not the result of fracture as shown in Fig. 9.
This is because the areas surrounding these features do not

123

SEM and metallographic examinations of the fracture

revealed a band of oxides and microporosity along the weld
interface. The existence of microporosity and formation of
ferrite layer reduced the strength, and the formation of
Widmanstatten ferrite decreased the impact strength of the
weld and caused the overload fracture. The band of oxides
was caused by air exposure of a molten layer at interface,
and the subsequent solidification of the liquid film contributed to the micro-shrinkage. Conditions that resulted in
high heat input and liquidizing of the weld interface (e.g.
prolonged heating) before the upsetting stage of the FRW
process should be mitigated.

Reference
1. Metals Handbook, 9th edn., vol. 6. American Society for Metals
(1983)