---

The launch of the Columbia mission STS-107.

Author’s note: This document was prepared under the sponsorship of the National Aeronautics and Space Administration. Neither the United States government nor any person acting on behalf of the United States government assumes any liability resulting from the use of the information contained in this document, or warrants that such use will be free from privately owned rights.

Materials analyses of key forensic evidence helped unlock the mystery of the loss of space shuttle Columbia that disintegrated February 1, 2003 while returning from a 16-day research mission. Following an intensive four-month recovery effort by federal, state, and local emergency management and law officials, Columbia debris was collected, catalogued, and reassembled at the Kennedy Space Center. Engineers and scientists from the Materials and Processes (M&P) team formed by NASA supported Columbia reconstruction efforts, provided factual data through analysis, and conducted experiments to validate the root cause of the accident. Fracture surfaces and thermal effects of selected airframe debris were assessed, and process flows for both nondestructive and destructive sampling and evaluation of debris were developed. The team also assessed left hand (LH) airframe components that were believed to be associated with a structural breach of Columbia. Analytical data collected by the M&P team showed that a significant thermal event occurred at the left wing leading edge in the proximity of LH reinforced carbon carbon (RCC) panels 8 and 9. The analysis also showed exposure to temperatures in excess of 1,649°C, which would severely degrade the support structure, tiles, and RCC panel materials. The integrated failure analysis of wing leading edge debris and deposits strongly supported the hypothesis that a breach occurred at LH RCC panel 8.

INTRODUCTION

The Space Shuttle Columbia was returning from a 16-day research mission (STS-107) with nominal system performance prior to the beginning of entry interface and during coastal crossing. Approximately one minute and 24 seconds into the peak heating region of entry interface (GMT 2003:32:13:52:17), an off-nominal temperature rise was observed in the left main landing gear (LMG) brake line temperature D sensor. Twenty-four seconds later, LMG brake line temperature A and C sensors exhibited offnominal temperature increases as well. Subsequently, numerous off-nominal bit flips and off-scale readings were recorded in multiple orbiter left wing components until the vehicle loss of signal was recorded at GMT 2003:32:13:59:32.136 (see the Loss of Columbia sidebar for details of the shuttle’s final minutes). Debris was observed periodically exiting Columbia’s flight path throughout the reentry profile over California, Nevada, New Mexico, and Texas.

Following an intensive four-month recovery effort by federal, state, and local emergency management and law officials, Columbia debris was collected, catalogued, and reassembled at the Shuttle Landing Facility at Kennedy Space Center, Florida. Approximately 83,900 items were collected representing an estimated 38 percent of the orbiter’s dry weight. A Materials and Processes (M&P) Failure Analysis team was concurrently formed to assess recovered debris and analyze selected component materials. The M&P team supported all materials analyses directed by the Columbia Accident Investigation Board.

The M&P team initially assisted Columbia subsystem experts in cleaning and assessing the fracture surfaces and thermal effects of selected airframe debris. Due to the forensic nature and sensitivity of handling single-source debris samples, “Pathfinder” samples that exhibited similar characteristics to the damaged components of interest were selected for evaluation. Additionally, these parts helped validate proposed sampling and materials analysis techniques. Effects of the breakup and reentry were observed upon examination of these aluminum, titanium, Inconel, and stainless steel components.

Following the extraordinary recovery and extraction of wing sensor data from the modular auxiliary data system/orbital experiments (MADS/OEX) recorder, emphasis was placed on identification and evaluation of left wing leading edge components. Damage patterns observed on select wing leading edge component debris included erosion of reinforced carbon carbon (RCC) wing panel pieces, slumping of thermal protection system (TPS) tiles, and metallic deposits. The extent of isolated damage to select leading edge components suggested that major thermal events occurred in the left wing leading edge near RCC panels 8 and 9. Samples of deposits from these areas and similar areas on the right wing were carefully chosen from extensive examinations and interpretation of radiographic images to minimize the quantity of sampling. Detailed procedures and sampling techniques were developed to preserve hardware and critical evidence, and a myriad of analyses were undertaken to determine the most timely and cogent methods for material determination of as-received deposits and prepared metallographic specimens. After considerable testing, three techniques were down-selected for all subsequent analysis. Powder x-ray diffraction (XRD) was chosen to provide crystalline compound information, scanning electron microscopy (SEM) x-ray mapping and back-scattered electron images were selected to show the distribution of similar elements, and microprobe analysis of cross sections through deposits was found to provide accurate compositional information which, when combined with other data, would yield content and layering information.

FIGURES
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

SEARCHING FOR CLUES—EARLY ANALYSIS OF LEFT WING COMPONENTS

Debris received at the Shuttle Landing Facility was identified, photographed, and placed on a reconstruction grid (Figure 1). Due to indications of erroneous shuttle telemetry and imagery of foreign object debris impacts along the left side of the orbiter, the M&P team performed early assessments of airframe components believed to be associated with a possible breach of Columbia. The components exhibited varying degrees of thermal effects, and the M&P team was tasked to evaluate the significance of the damage and its possible relation to the breakup. These components included the midbody panel, uplock roller, main landing gear (MLG) strut, left wing carrier panel fasteners, and left wing tiles. A cross section of the orbiter’s left wing is shown in Figure 2. Debris assessments recorded by the M&P team later appeared to correlate well with the sensor data obtained from the MADS/OEX recorder.

Midbody Panel

The midbody panel was identified to be the lower aluminum wing skin and tiles from the inboard forward corner of the left hand (LH) MLG door (Figure 3a). Unique flow patterns were observed on portions of the tiles, and there was evidence of localized heat erosion at the outer mold line (OML) along the panel’s edge. The surface of the tiles eroded by the flow was glazed and hardened, and some metallic deposits were observed on the tile surface. The patterns observed in the tile were approximately 90 degrees from the nominal reentry flow pattern. The corners of the tiles near the inboard corner of the gear door were cratered and eroded; however, there were no visible deposits on the tiles.

The edge of the panel at the inboard corner was also cratered, and a small hemispherical erosion pattern was observed at the panel’s edge (Figure 3b). The flow patterns observed in the tile near the forward inboard corner of the panel were approximately 90 degrees from the nominal reentry flow pattern. Additionally, the OML of the panel opposite the midbody panel (forward outboard corner) and the OML of the aft inboard corner showed highly localized heating and erosion at the corners.

Main Landing Gear Door— Uplock Roller

The M&P team evaluated additional landing gear door and wheel well hardware believed to be relevant to the investigation. One of four left landing gear uplock rollers (Figure 4) was recovered; several metallic deposits were observed on the frame and roller portions. A thin, uniform, metallic coating was observed on all surfaces of the inner and outer titanium flanges and approximately the lower third of the cylindrical shaft. Additionally, some discoloration/heat tinting was observed on the cylindrical shaft adjacent to the metallic deposits. Energy-dispersive x-ray spectrometric (EDS) analysis of the coating indicated major amounts of metallic aluminum with minor amounts of copper, titanium, manganese, and iron. No surface features or markings could be identified that would aid in identifying the location of the roller within the wheel well.

Landing Gear

A portion of a landing gear strut (Figure 5) was recovered during search operations and identified as a left MLG component. The backside and bottom of the cylindrical strut had very localized regions of erosion and burning and were heavily coated with resolidified molten metal deposits, termed slag. The front side (faces forward when deployed) showed no signs of burning or erosion, with portions of the chrome plating still intact. The outboard axle showed thin, uniform slag deposits, while the inboard axle was heavily eroded approximately 9 cm to 10 cm along its axis.

Carrier Panel Attach Fasteners

During the debris assessment, it was discovered that several steel fasteners that attach the upper and lower aluminum access panels to the wing spar appeared to have brittle fracture characteristics (Figure 6). The aluminum panels were protected with tile and secured to the RCC spar attach fittings with two stainless-steel fasteners. The lower panels had an aluminum box beam as a spacer between the access panel and the spar fitting.

Nine failed and four unfailed fasteners were selected for failure analysis. Seven of the nine failed fasteners were determined to be high-temperature failures, while the two remaining fasteners appeared to have failed at lower temperatures. Of the seven high-temperature failures, four were melted at the head end, indicating localized temperatures in excess of 1,315°C (2,400°F). The remaining three failures exhibited intergranular fractures on a large-grained structure, indicating temperatures between 1,038°C (1,900°F) and 1,204°C prior to fracture.

The two lower-temperature failures appeared to have been due to ductile bending. The grain sizes of these two fasteners indicated moderate temperature exposure between 704°C and 927°C. Because these were not intergranular fractures, a time of failure could not be correlated to the period of exposure.

Main Landing Gear Door— Corner Tile

A corner tile on the left main landing gear door tile demonstrated a similar flow pattern as the left midbody panel described previously. Visual evaluation of the OML of the tile revealed apparent thermal flow erosion characteristics, such as melting, flowing, and lifting of the reaction cured glass coating of the outboard edge, directly adjacent to the outboard thermal barrier. The flow direction appeared to be inboard and slightly forward (Figure 7). This flow pattern was oriented approximately 90 degrees from the nominal flow direction expected in this area. In addition, the inner mold line (IML) showed similar evidence of thermal flow erosion, but indicated the flow direction to be from inside the forward outboard corner of the main landing gear cavity, outward, and forward. X-ray radiography did not detect any notable features aside from the surface features noted; therefore, no sampling or chemical analysis was performed.

CLUES EMPHASIZE SHUTTLE’S WING LEADING EDGE

Evidence of extreme overheating and heavy deposits observed on specific pieces of wing leading edge hardware appeared to correlate with the OI and MADS/OEX sensor data anomalies, and the team began to support detailed analyses of hardware in this area. In order to validate the proposed breakup scenarios under consideration, the investigation concentrated on three areas of interest associated with the wing leading edge (Figure 8): shuttle wing and leading edge subsystem (LESS) carrier panel tiles, RCC panels, and wing substructure attach hardware. Damage patterns observed in these areas suggested that major thermal events occurred in the left wing leading edge.

Several left wing leading edge components exhibited unique indications of heat damage relative to other wing leading edge parts. These components included excessive overheating and slumping of LESS carrier panel tiles, eroded and knife-edged RCC rib sections, and heavy deposits on select pieces of RCC panels.

Radiography of both tile and RCC panel pieces showed that x-rays were an excellent method of characterizing the location and shape of metallic deposits, melt-flow patterns on tile, and imbedded debris not visible on the surface. Following radiography, samples of high-density surface deposits were removed and materials analyses were employed to identify their composition. Data obtained from the analyses enabled estimates to be made of the environment and temperatures that were necessary to create the surface deposits.

Thermal Effects of LESS Carrier Panel Tiles

Visual Observations

Evidence of overheating and slumping was observed on three lower left carrier panel 9 tiles adjacent to LH RCC panel 9. Figure 9 shows the simulated configuration of the carrier panel tiles. Depressed/slumped and eroded regions were observed in two of the three tiles. The forward-facing sidewalls of samples 16692 and 22571, which nominally seal against the lower RCC panel 9 heel, were severely slumped and eroded.

Dark-colored deposits were observed on all three outer mold line (OML) tiles (samples 16692, 22571, and 57754). The thickness of the deposits varied across the tile surfaces. In the case of 22571 and 57754, the deposits produced visually apparent flow-like patterns oriented in the aft/outboard direction (Figures 10 and 11). Visual evaluation showed evidence that in some locations on the tile sidewalls, the deposits had built up over adjacent soft goods, such as insulation and batting material. This was supported by the presence of entrapped ceramic fibers in the deposits.

An internal tile was recovered from upper left LESS carrier panel 8. This tile exhibited a greenish coloration and heavy slumping (Figure 12). The surface deposits on internal tiles 50336 and 50338 were not as thick as those observed on the lower left LESS carrier panel 9 OML tiles.

X-ray radiography of the carrier panel tiles did not detect any notable features aside from the surface deposits noted. A typical example is shown in Figure 13. Sampling and chemical analysis were therefore initiated for surface deposits only.

Chemical Analysis

Samples of the surface deposits were removed and chemical analysis was performed using scanning electron microscopy/energy dispersive spectroscopy (SEM/EDS) and electron spectroscopy for chemical analysis. The results indicated that the elemental components of the deposits were primarily aluminum, nickel, niobium, and carbon.^1–3 Although the precise composition of the source alloys/compounds could not be identified with certainty, the elements found were consistent with the compositions of 2000 series aluminum alloy, Inconel 601, Inconel 718, and the internal wing Cerachrome insulation. Electron spectroscopy for chemical analysis results indicated that the outermost layer was highly carbonaceous. This indicated that the carbonaceous outer layer was deposited after the metallic layer, which had in some cases fluxed into the cured glass tile coating.

Deposits were also found on the threaded internal surface of the ceramic insert in tile sample 16692. The fused silica plug and lock cord were observed to be intact at the OML end of the insert. This indicated that the deposits were introduced from the internal side of the tile. The elemental composition of the deposits was essentially the same as that of the deposits found on the OML of the tile. The deposits may have occurred after the tile’s protective strain isolation pad facing the IML had been partially eroded away or debonded.

Thermal Effects

Tile slumping and surface deposits on the LH lower LESS carrier panel tiles are consistent with flow occurring from inside the RCC cavity out through the upper and lower carrier panel locations in that vicinity. The surface deposits on lower LH carrier panel 9 tiles are consistent with a flow direction exiting from RCC panel 8. The thermal degradation of the internal tiles recovered from upper carrier panel 8 and lower carrier panel 9 suggests that the flow was in excess of 1,649°C. The composition of the tile surface deposits suggests that the flow contained molten/vaporized materials from the LESS internal insulators, attachments, carrier panels, and/or wing spar.

Thermal Effects— RCC Panel Deposits

Deposits similar to those observed on the LESS carrier panel tiles were also evident on the inner surfaces of several LH RCC panels (Figure 14). The deposits resembled solidified metallic slag and were strongly adhered to the internal surfaces of the panel segments. The quantity and thickness of the deposits also varied according to the RCC panel number.

The M&P team noted marked differences in the appearance and quantity of deposits between the LH and right hand (RH) RCC surfaces. Figure 15 presents a qualitative summary of slag observed on left wing RCC panel pieces, ribs, and T-seals.

Figures 14 and 15 illustrate that the relative severity of the left wing leading edge deposits approached a maximum at RCC panel 8 and decreased on either side. Heavy deposits were also observed on the inner surfaces of the outboard ribs of panels 4, 5, and 7; however, very few deposits were observed on the inboard ribs of these panels.

Very few deposits were observed on RCC panels past panel 12, and there was more evidence of mechanical damage than thermal effects on the remaining panels outboard of panel 12. Although the quantity of deposits was considerably greater on the LH leading edge panels than the RH panel sections, medium grade deposits were also observed on an upper panel portion and outboard rib section of RH RCC panel 8.

Metallurgical Analysis

The relative differences observed between the amount of slag deposits on the left and right RCC panels prompted a metallurgical analysis. The analysis included the following: a review of the chemistry of high-temperature reactions associated with the wing materials, non-destructive radiography of the RCC panel surfaces, and a metallurgical evaluation of samples removed from the RCC panels. Cross sections of deposits from left and right RCC panels were analyzed to identify and characterize their composition, composition gradients, and any layering effects on the inner surfaces.

The high-level objectives of the analysis were the following: