2. 93Journal of Failure Analysis and Prevention Volume 6(1) February 2006
debris surface, were torn off during re-entry. Exam-
ination along the inside of the panel, or inner mold
line (IML), revealed a plethora of spraylike debris.
The IML surface, in service,is coated with a green
Koropon (PRC-DeSoto International, Inc.) primer;
however, the IML displayed a gray matte finish
indicating heavy oxidation, primer burn-off, or
heavy amounts of debris sprayed over the surface.
Figure 2 displays the IML surface with areas of
interest labeled. The door frame edge (bottom) of
the IML exhibited relatively little damage to the
aluminum base metal plate, although on the surface,
a series of Inconel “baggie”clips are observed to have
melted, as displayed in Fig. 3(a). In Area 26 a
titanium fastener is observed to have melted at the
tip approximately 6.35 mm (0.25 in.) from the base
plate metal (Fig. 3b). Additionally, directional
slumping is observed to occur away from the edge
of the ET door frame. This indicates both a large
temperature gradient from fastener to base metal
and the direction of gas flow. The melted fasteners
are accompanied by the presence of multidirectional
spatter-type debris (Area 41, Fig. 3c).
Area 29, near the top of the fracture surface, shown
in Fig. 3(d), displayed a large amount of flame-spray
type debris; however, chipping of the
debris reveals a relatively undamaged
base metal surface. On the other hand,
Area 36 exhibited several patchy areas
of debris (Fig. 3e) that were observed
to be tightly adhered to the base metal,
suggesting bonding between debris and
base metal. In addition to the debris
and thermal damage, the forward edge
of item 33767 appeared to experience
a fast overload fracture exhibiting little
plastic deformation. This is in contrast
to the top of the plate, which appears
to have been subject to severe thermal
trauma.
Visual inspection of the OML also
revealed several areas of interest, as
illustrated by Fig. 4. Area 56, near the
top portion of item 33767, contained
one particular zone of interest exhibit-
ing deformation, displayed in Fig. 5(a).
Dark gray surface deposits suggest
heavy surface oxidation. Cracks through
the surface oxides as well as the woody
upset fracture edge indicate a possible
hot impact mechanism. Area 60,
depicted in Fig. 5(b), was observed to
experience localized surface melting and
re-solidification. This is indicative of a
local temperature along the edge of at
least 638 °C (1180 °F). Along the aft
edge of item 33767, Area 61 (Fig. 5c)
displays impact deformation occurring
on a “mushy” semi-melted base metal,
indicating local temperatures of approx-
imately 482 °C (900 °F). Figure 5(d)Fig. 2 Inner mold line (IML) surface
Fig. 1 Item 33767 overall position and orientation. (a) Outer side of panel and two
of the black tiles that are part of the thermal protection system. (b) Inside of the
panel, which exhibits a variety of spraylike debris
(b)
(a)
3. Columbia Accident Investigation Board (CAIB) Pathfinder Analysis of Item 33767 (continued)
94 Journal of Failure Analysis and PreventionVolume 6(1) February 2006
displays Area 47, along the ET door
edge. Bubbled Koropon primer is
observed along the surface, as well as
charred room-temperature vulcanized
(RTV) rubber. The charred RTV
indicates that there is a local temp-
erature of approximately 427 to 538
°C (800-1000 °F). Lastly, Area 49, an
area protected by thermal tile as shown
in Fig. 5(e), displays a darkening of the
Koropon primer; however, the RTV in
the region is reasonably undamaged.
The color of the Koropon primer implies that the
local temperature at the center of the plate was
approximately 204°C (400 °F).
Hardness and
Conductivity Testing
Hardness and conductivity tests were administered
to item 33767 in order to characterize the material
and compare it to known standards set for the alloy.
Figure 6 displays measured values of both the
hardness and conductivity including their relative
position on item 33767. The minimum hardness
requirement set for this alloy is 74 Rockwell “B”
hardness (HRB). Along the ET door frame the hard-
ness was at a maximum and was only measured to
be approximately 70 HRB. Near the top fracture
edge the hardness was observed to decrease to
approximately 40 HRB.
Conductivity testing was conducted on item
33767 and compared to average conductivity values
of 2124-T851 aluminum which occur in the range
of 36 to 41% International Annealed Copper
Standard (IACS). Near the ET door frame values
of approximately 40% IACS were observed, whereas
along the top fracture edge the values diminish to
approximately 27 to 35% IACS. Both the hardness
and conductivity data suggest that item 33767 was
exposed to severe thermal trauma during the failure.
Fig. 3 (a) Inconel “baggie” clip. (b) Photo 26: Directional slumping of titanium fastener. (c) Photo 41: Multidirectional metal spatter on the
inner mold line surface. (d) Photo 29: Chipped spraylike debris. (e) Photo 36B: Tightly adhering spray debris
(a) (b) (c)
(d) (e)
Fig. 4 Outer mold line (OML)
4. 95Journal of Failure Analysis and Prevention Volume 6(1) February 2006
Metallographic Examination
Several metallographic sections were
taken from item 33767 in order to
characterize the microstructure of the
material. Figure 7 illustrates the areas
chosen for metallographic exam-
ination.
The first section examined, Sample
3, was taken from the forward edge of
item 33767.The fracture surface in this
region appeared to experience slight
plastic deformation and displayed a
course, granular type fracture. The
microstructure at the fracture surface
was observed to be typical of that of an
overaged 2124-T851 aluminum alloy
as shown in Fig. 8. The microstructure
present throughout the section indicates
a maximum temperature exposure of
approximately 232 °C (450 °F). The
lack of plastic deformation on the frac-
ture surface combined with the micro-
structure suggests that the failure of the
forward edge of item 33767 was most
likely an overload fracture and occurred
late in the fracture sequence. Debris was
Fig. 5 (a) Photo 56: Heavy surface oxides. Deformation and edge upset suggest hot impact. (b) Photo 60: Localized melting and
resolidification along the panel edge. (c) Photo 61: Impact on “mushy” base material. (d) Photo 47: Charred room-temperature
vulcanized (RTV) rubber and Koropon primer. (e) Photo 49: Koropon primer and RTV after tile removal
(a) (b) (c)
(d) (e)
Fig. 6 Conductivity and hardness decrease near the fracture edge.
Fig. 7 Sample locations: Sample 3-forward edge; Sample 17-center fracture; Sample
9-aft quarter fracture edge; Sample 11-aft edge fracture; Sample 12-external
tank door frame
5. Columbia Accident Investigation Board (CAIB) Pathfinder Analysis of Item 33767 (continued)
96 Journal of Failure Analysis and PreventionVolume 6(1) February 2006
also observed to have been deposited along the
fracture surface on Sample 3.
The next sample analyzed, Sample 17, was taken
from the midsection of the top fracture surface of
item 33767. Sample 17 was prepared and etched
using Keller’s reagent. The metallographic sample
displayed a rapid change in microstructure along
the transverse direction (Fig. 9). Beginning at the
fracture surface edge there is apparent necking of
the base material along with localized melting.
Examination of the microstructure from the top
fracture edge toward the ET door frame revealed
that the changes appeared consistent with that of a
high-intensity, short-duration heating followed by
rapid cooling, similar to the heat-affected zone
(HAZ) microstructure of a weld. Toward the bulk
material the microstructure was consistent with that
of an overaged material.The grain structure, accom-
panied by the apparent plastic deformation along
the fracture surface, indicates a hot tensile or hot
bending mechanism. Local melting near the fracture
surface also implies a surface temperature in excess
of 593 °C (1100 °F).
Sample 9 was acquired along the aft quarter of
the top edge of item 33767. Similar to Sample 17,
examination of Sample 9 revealed a dramatic change
in microstructure throughout the cross section from
fracture edge to bulk material, as shown in Fig. 10.
Likewise, the microstructure and large amount of
plastic deformation near the fracture surface imply
hot tensile failure mechanisms. Again the micro-
structure indicated local melting near the fracture
surface with a distinct transition approximately 12.7
mm (0.5 in.) from the fracture surface.
The complete dissolution of the copper constit-
uents combined with grain-boundary melting near
the fracture surface in both Samples 17 and 9
indicated that the top edge of item 33767 was ex-
posed to a temperature of approximately 521 to 638
°C (970-1180 °F). Similarly, both of the samples
exhibited a fracture mechanism consistent with hot
tensile fracture. Additionally, both microstructures
exhibited void coalescence, necking, and localized
melting. Moving away from the fracture surface,
some incipient melting was observed to occur and
appeared to lessen toward the bulk material. The
incipient melting transformed to a microstructure
resembling a heavily overaged alloy. Both of these
sequences imply a high-temperature, short-duration,
rapid cooling scenario with a steep temperature
gradient, which was more evident in Sample 9.
Along the aft edge, Sample 11 was also removed
for metallographic examination. Figure 11 displays
a micrograph of the aft fracture surface on Sample
11. The fracture surface appeared to have been the
result of a hot bending type mechanism with the
mating piece apparently moving toward the top
edge. The fracture appeared to be woody in nature
and evidence of delamination was also detected. It
could be inferred that eutectic melting was taking
place along the fracture surface, again indicating that
this portion of item 33767 was subject to temp-
eratures in excess of 482 °C (900 °F).
Lastly, a metallographic sample was removed from
the ET door frame near the aft edge (Sample 12).
A portion of the sample included one of the several
Ti6Al4V fasteners.Initial observations indicated the
protruding end of the fastener had melted. Further
metallographic examination of the fastener revealed
a dramatic difference between grain size traveling
along the fastener from tip to base metal as shown
in Fig. 12. Large grains can be observed near the
melted end of the fastener; closer to the base metal,
a rapid decrease in grain size is detected. The large
grains near the tip combined with the rapid change
in grain size are indicative of a large thermal gradient.
Titanium will melt at approximately 1593 °C (2900
°F), while the base metal aluminum alloy will melt
at a temperature of approximately 638 °C (1180
°F); this would imply that a thermal gradient of
approximately 927 °C (1700 °F) was applied to this
portion of item 33767 over a distance of 6.35 mm
(0.25 in.).
Scanning Electron Microscopy
Scanning electron microscopy (SEM) and energy-
dispersive microanalysis were conducted on the
samples used for metallographic examination in
order to further characterize the material and fracture
surfaces. Several areas experiencing debris deposition
along the surface of item 33767 were analyzed in
order to determine the composition of the debris.
Along the fracture face of Sample 3 a large accumu-
lation of debris, approximately 152 µm (6 mils)
thick, reminiscent of a flame-spray application, was
analyzed (Fig. 13). The debris matrix consisted
mainly of aluminum and oxygen. Copper, iron,
nickel, titanium, chromium, calcium, silicon, and
6. 97Journal of Failure Analysis and Prevention Volume 6(1) February 2006
Fig. 8 Sample 3: Minimal deformation at the fracture face.
Maximum base material temperature approximately 232
°C (450 °F)
manganese were also present; however, examination
of constituent ratios did not yield conclusive results
as to compounds prevalent within the bulk debris
matrix. It should also be noted that concentrations
of debris containing large amounts of gold (common
braze material on the shuttle) and 300-series
corrosion resistant steel (CRES) were also found
within the deposition matrix. The deposit, which
resembled the 300-series CRES, would suggest a
deposition temperature of approximately 1371 to
1427 °C (2500-2600 °F).
The aft fracture surface present along Sample 11
was further analyzed in order to classify the failure
mechanism. Elemental mapping was employed to
segregate base metal from debris. Figure 14 shows a
distinct layer of debris trapped between layers of
base metal, combined with apparent delamination;
this additionally supports the previous notion of hot
bending failure. Surface damage was also prevalent
in the form of microcratering along Sample 11.The
observed microcraters appeared to be heavily oxi-
dated; furthermore, titanium debris was also detected
along the crater surfaces (Fig. 15). This indicates
that the surface damage occurred before the melting
of the titanium fasteners.
Along the ET door frame, near the aft portion of
item 33767, Sample 12 was also used to further
identify thermal conditions imposed on the subject
part. Spraylike debris was again detected along the
surface of the sample. The debris was comprised
primarily of titanium, aluminum, and vanadium,
indicating that the debris came from the Ti6Al4V
fasteners. On further examination, cracking was
observed within the Ti-Al-V debris, as shown in
Fig. 16, verifying that a rapid cooling effect was
present along the aft portion of the ET door frame.
Conclusions
At the outset of this investigation several questions
were proposed regarding the heating of the debris.
This analysis has shown that the answer is one that
varies by location, as depicted in Fig. 17. Hardness
and conductivity testing indicated that the top of
item 33767 was affected most severely, while along
the ET door frame the material was only marginally
affected. In addition to the hardness and conduc-
tivity testing, the Koropon color at the center of the
plate and near the forward end of the piece indicated
that the base metal material reached a temperature
of approximately 204 to 260 °C (400-500 °F).
Charring of the RTV primer occurred in several areas
along the top edge of the debris and near the aft
edge. The charring suggests exposure to a
temperature of approximately 427 to 538 °C (800-
1000 °F). Melting of the base metal detected along
the upper aft edge of item 33767 would occur due
to temperatures in excess of 593 °C (1100 °F).
Lastly, the observed melting of theTi6Al4V implies
a temperature of approximately 1649 °C (3000 °F)
was reached on some surfaces.
7. Columbia Accident Investigation Board (CAIB) Pathfinder Analysis of Item 33767 (continued)
98 Journal of Failure Analysis and PreventionVolume 6(1) February 2006
Fig. 9 Sample 17: Series of micrographs showing the progressive microstructure changes
A D G
B E
C F
Fig. 10 Sample 9: Series of micrographs showing rapid microstructure changes
8. 99Journal of Failure Analysis and Prevention Volume 6(1) February 2006
The duration of the heating appears to have been for
a relatively short time. Assuming the shuttle remained
largely intact until the time of the Loss of Signal event,
itisreasonabletoconcludethatitem33767wasexposed
to temperature for approximately five minutes.
However, with the presence of severe temperature
gradients and the microstructure along the top edge
indicating rapid heating and cooling, it could also
be feasible for the duration of heating to be much less
than the suggested five minutes. The direction of the
heating was not readily apparent outside of the tenet
that heat would travel from hotter areas to cooler areas
or in relation to the part, from the top and aft edges
toward the ET door frame.
Fig. 11 Sample 11: Cross section of the aft fracture surface
Fig. 14 Sample 11: Element mapping reveals a layer of trapped
heavy metal debris.
Fig. 12 Sample 12: Microstructure of 6-4-Ti fastener indicates a
steep thermal gradient between the fastener tip and the
plate.
Fig. 13 Sample 3: Globular deposit. Large bright deposit is
corrosion resistant steel.
100µµµµµm
1mm Mix
Fracture analysis implies that item 33767 failed
along the top edge first.The material is thinner along
the top edge, suggesting that it would tend to heat
more rapidly than other sections of the piece.
Localized heating resulted in a loss of mechanical
strength for the top edge and melting at and near
the fracture surface. Microstructures of the areas
along the top edge of the debris confirm that rapid
heating and cooling occurred. The top edge and
apparent first failure appears to have been the result
9. Columbia Accident Investigation Board (CAIB) Pathfinder Analysis of Item 33767 (continued)
100 Journal of Failure Analysis and PreventionVolume 6(1) February 2006
Fig. 15 Sample 11: (a) Microcraters occurred first, followed by (b)
titanium alloy spatter debris.
(b) 300µµµµµm
(a) 300µµµµµm
Fig. 16 Cracking in theTi-Al-V debris
Fig. 17 Thermal summary
Fig. 18 Failure sequence
of hot tensile failure. The aft failure was observed
to exhibit signs of a bending mechanism through
delamination and debris trapped between layers of
base material. The aft failure most likely occurred
second in the failure sequence. Lastly, the forward
failure displayed limited ductility and was most
likely the site of final, fast, overload fracture. The
failure sequence is illustrated in Fig. 18, and fracture
mechanisms are labeled near the respective areas.
Conventional failure analysis using traditional in-
vestigation techniques was able to provide useful
information regarding the heating duration, temp-
erature, and probable failure sequence. Conductivity
and hardness testing were successful in detecting
locations at which intense thermal trauma occurred.
These approximations were later verified by metal-
lography, SEM imaging, and EDS. However, it
should be noted that SEM and EDS were of limited
value until the sectioned samples were analyzed.
Acknowledgment
Eric Weishaupt of Metals & Materials Engineers,
LLC is gratefully acknowledged for his contribution
in preparing this manuscript from the transcribed
presentation.
