MFES Homepage Wade Bartlett's CV Vetronix CDR Info Back to the Technical Essays Page Publications Some Web-Links How to contact MFES
Mechanical Forensics Engineering Services, LLC, 88 Tebbetts Road, Rochester NH, 03867


Review of Lamp Examination for
ON or OFF in Traffic Accidents

TEXT TITLE: Topic 823: Lamp Examination for ON or OFF in Traffic Accidents
PUBLISHER: Northwestern University Traffic Institute

INTRODUCTION: A comprehensive review of this manual revealed that though it does contain much useful information, there are some statements which are incorrect and/or misleading and can lead an investigator to incorrect conclusions. This critique is designed to be reviewed parallel to the manual.

Pg 23-5 In Exhibit 1, new filaments are described as having "evenly spaced coils."
In small lamps, i.e. other than headlamps, it is common for the filament pigtails to be clamped with poor axial alignment. As manufactured, at the end with the poorest alignment the coils are not evenly spaced.
Pg 23-5 In Exhibit 1, new filaments are described as having "longitudinal draw lines."
The statement is correct, however analytically it is not as clear cut. Filament wire is deliberately drawn so as to produce highly elongated grains. In non-halogen lamps, after extended service, grain boundary grooving develops which gives an array of axial lines. The investigator can distinguish between draw lines and grain boundary grooving by using oblique illumination at high magnification and noting whether the lines are raised above the surface or depressed. Grain boundaries always become incised. Both raised and incised draw lines will be observed. Grain boundary grooving lines can seldom be traced for more than about four wire diameters whereas draw lines are continuous. When working in the scanning electron microscope (SEM) the investigator must use great care to unambiguously determine whether the electron beam is revealing a groove or a ridge. Because of the SEM's high magnification, the investigator usually must translate and reorient the sample to track the lines for several wire diameters.
Pg 23-5 Exhibit 1 suggests that only new lamps have undarkened glass.
Providing that they are operated at normal voltage, the glass in halogen lamps remains undarkened throughout their life. This becomes a useful tool in that if a halogen lamp has a darkened envelope then it was operating at too low a temperature for an extended period. Either the lamp's operating voltage was mismatched for the application or there was an electrical defect resulting in low voltage. If all of the halogen lamp envelopes in a vehicle were darkened, then there was probably a system defect.

Small vacuum lamps exhibit darkening which is uniform all around the inside of the envelope and stem. Lamps which are back filled with gas, as most are, are darkened directly above the filament. The location of the darkening can be useful in establishing lamp installed orientation.
Pg 23-5 Exhibit 1 suggests that only new lamps have a "bright base."
Moisture and dirt cause corrosion. In dry locations such as inside well-sealed lamp assemblies or inside a well-sealed trunk, the bases will remain bright.
Pg 23-5 Exhibit 1 suggests that only aged bulbs exhibit "possible downward sag."
Downward sag of a filament in a halogen lamp causes the convex side of the coil to run colder while the concave side runs hotter. Tungsten deposition takes place preferentially on the hot side. Under a high power optical microscope or in the SEM the investigator will observe a smooth buildup on the concave side and an eroded but bright surface on the convex side. The deflection shown in exhibit 3 is sufficient for a preferential buildup to develop over time in a halogen lamp.

In halogen lamps, the preferential buildup on the concave side is critical in determining whether a filament's curvature developed as a part of an impact event which was historical or one that was recent. A preferential buildup is sure evidence that a curvature is historical. The absence of buildup shows low operating voltage or little lamp use or that the impact was recent. A buildup which is angularly displaced from the concave side is probably significant, however it requires careful interpretation.

The filament supports are bright in a new lamp. During aging the hot filament supports and adjacent, seldom used filament's "getter" the contaminating oxygen and nitrogen which outgas from lamp components and slowly leak into the lamp. The supports and adjacent filaments acquire colorful tarnishes, starting with dark blue to gray closest to the filament and tending to pink and yellow in cooler locations. The overhanging ends of the filament show similar colors. The light-element electron dispersive spectroscopy (LE-EDS) attachment found on many scanning electron microscopes (SEMs) will show the presence of the bound oxygen and perhaps the nitrogen.

On the surfaces of filament supports in better quality lamps, there is often a rough surfaced, multifaceted layer of gettering material.
Pg 23-5,
Pg. 23-20
In exhibit 1 and on the later page, the issue of "rounded or ball ends" associated with burned out lamps is discussed.
The rounded or ball ends are due to arcing as the filament separated. The arc beads which form on initial separation are large and symmetrical. Because the ends of the filament remain electrically active, and aided by having been stretched during impact, as they "wave around" they frequently contact other surfaces and create additional arc beads. These beads are small and asymmetrical, with much less melting because of the low initial filament temperature. Magnifications of at least 40X are needed to see these small beads. As long as the circuit remains energized these additional arc beads can develop long after an impact event.

Normal lamp burnout occurs when metal has evaporated from the filament, which creates a local band of high resistance. The higher resistance causes a higher voltage drop in that region of the filament, creating a higher power dissipation and a local hot spot. The hot spot soon burns out.

After a filament breaks in mid-span the remaining filament ends are subject to cantilever loading at the posts. If the metal has become embrittled then the filament will often subsequently fracture adjacent to a post. Therefore, observation of a brittle fracture adjacent to a post does not show that the lamp was "off" at the time of impact unless the entire filament has been shown to be free from arc beads, fused glass particles, or other direct evidence that the filament was hot.
Pg 23-7 Aged lamp bulbs
The statement is partially correct. Both small and large, high time, non-halogen lamps including sealed beam headlights frequently show tungsten darkening. The EDS accessory of the SEM easily detects this deposited tungsten on fragments of the envelope.
Pg 23-7 Aged lamp filaments
Note elsewhere the special characteristics of aged filaments from halogen lamps.
Pg 23-7 Age sag in aged lamps
Analysis shows that the rate of "age sag" depends critically on the filament's operating temperature and metallurgy. Thus age sag is often non-existent on filaments taken from older cars having corroded wiring systems. On the other hand, a defective alternator voltage regulator or a high resistance or sulfated battery can overvoltage the system and cause every filament to quickly sag. Plain tungsten filaments sag most rapidly. The most sag resistance are thoriated tungsten filaments and the AKS doped tungsten, both of which have highly elongated and interlocked grain structures. (AKS represents the chemical symbols for the three doping elements, Aluminum, Potassium, and Silicon.)

The elongated, interlocked grain structure is often visible on the surface of aged, non-halogen filaments. The presence of thorium in the thoria particles (ThO2) is easily detected by EDS since the amounts are in the 1-2% range. The dopants in the AKS alloys are usually below the sensitivity limits of EDS analysis systems. The Metals Handbook, p327, v3 9th edition, lists the approximate dopant amounts as follows:
        Aluminum (A), 15 ppm
        Potassium (K), 91 ppm
        Silicon (Si), 47 ppm
Pg 23-8 Lamp failure...burnout
Line 1 is in error. The melting point is not reached unless and until filament separation has occurred and there is arcing. The failure takes place through evaporation of tungsten from the overheated band.
Pg 23-8,
Pg 23-25
Pg 23-28
Lamp failure, Brittle Fracture
Filament deformation, Broken Filament
The description is only partially accurate. The paragraph describes recrystallization. While recrystallization can and does occur, it is not as much a problem as in days gone by since manufacturers have found ways of suppressing it through adjustment of metal chemistry. At operating temperatures above and below the recrystallization temperature grain growth occurs. Grain growth leads to the grain boundary embrittlement which can cause both high and low temperature brittle fracture. Grain growth also reduces the metal's creep strength, thereby facilitating more rapid "age sag." In extreme cases single grains may extend across the filament's entire section. When this occurs, unfavorably oriented grains are subject to more rapid creep, allowing their cross sections to be preferentially reduced. Although creep generally lowers a lamp's wattage and average filament temperature, a grain experiencing preferential creep will run hotter.

While it is not difficult for a trained observer to differentiate between ductile and brittle fracture, the services of a metallurgist experienced in the interpretation of fracture morphology are required to ascertain the cause of a particular brittle fracture.

It is incorrect to assert that brittle fracture reliably indicates low temperature fracture. Brittle fracture in the absence of grain growth more probably than not indicates low temperature fracture, but it is not a sure thing. The ductility shown for example in Exhibit 25 is restricted to the pure tungsten filaments used in lamps which are smaller and operate with lower temperature filaments.

As pointed out earlier, it is important to check the small filament pieces for high temperature indicators. This requires magnifications of at least 40X in a stereo microscope. While the conventional, upright stereo microscope works well for intact filaments, when examining debris inside an envelope the pieces lie on the opposite side of the bulb. Looking through the glass ripples in the near side causes distortions which seriously obscure the fine details. It is therefore imperative that if a claim of low temperature fracture is to be made, the investigator meticulously examine the pieces by either opening the envelope or using an inverted stereo microscope. The inverted unit is preferred since it is non-destructive. Exam results based on upright microscope examination through an intact envelope should be discounted.
Pg 23-12 Lamp abnormalities resulting from accidents, last paragraph
It is inaccurate to state that the principles for halogen lamps are the same as for regular lamps. The redeposition phenomenon does not occur in filaments from regular lamps so that important aspects of the analysis of filament surfaces are markedly different.
Pg 23-12
Pg 23-34
Exhibit 22
Exhibit 84
These exhibits are seriously deficient in that they fail to preserve the orientation of the lamps. In interpreting sag it is important to know whether a particular lamp's base is forward, rearward, up, down, left, or right. In addition, the rotational orientation must be noted. This is most conveniently done with a small mark on the top of the envelope. Ball point metal markers and "whiteout" are both convenient.
Pg 23-13 Exhibits 23 & 53, Glass Broken, White Oxide
When a lamp envelope is broken while the filament is ON, or the filament is energized subsequent to the envelope having been broken, the compounds which form and their colors are:
        Tungsten Nitride, (WN2), Brown
        Tungsten Dioxide, (WO2), Brown
        Tungsten Pentoxide, (W2O5), Blue-violet
        Tungsten Trioxide, (WO3), Yellow

EDS and xray diffraction studies as well as the color of the deposits indicate that tungsten nitride and tungsten trioxide predominate in lamps which are broken while ON. Tungsten nitride decomposes in water so that it is vitally important that lamp fragments be protected from exposure to rain.

Much of the discussion of "oxides" extending into page 23-14 probably actually reflect nitrides. The test is easy. Just soak the fragments in water. What remain are oxides.
Pg 23-17 Filament Deformation
The statement is inaccurate. As discussed earlier, due to grain boundary embrittlement following grain growth, hot filaments may not be ductile. Barring low quality metal, the determination hinges on whether or not the filament is thoriated and/or AKS doped, and whether or not grain growth has occurred.

Not receiving the attention the topic deserves, is the issue of when in an accident electrical power is lost. There are only two sources of electrical power: the battery and the alternator. Early in a sequence of collision events rotation of the alternator may be arrested, the battery may be ruptured, or the battery cables may be severed, thereby instantaneously interrupting all electrical power. In these circumstances, the absence of high temperature failure in lamps remote from the collision zone does not indicate that they were OFF prior to the loss of electrical power.
Pg 23-19 Broken Filament, next to last paragraph
The statement reads "Old filaments which have become crystallized and..." Every metallurgist cringes at such a misleading statement. All ordinary metals are crystalline at all times except when molten or vaporized. Perhaps the authors were referring to intergranular cleavage fracture due to grain boundary embrittlement.
Pg 23-20
Pg 23-22
Force of Impact
Investigators should be cautioned that research has shown that under service conditions the degree of "hot shock" and acceleration do not correlate. The metallurgical explanation for the lack of correlation is widely varying aging rates due to different filament operating temperatures, different filament geometry,and differences between alloys.

Investigators must not extrapolate from aircraft lamps due to numerous differences. Aircraft lamps are shorter life due to higher operating temperatures. Except for older, lighter aircraft, they are 24 volts and higher. Except in the smallest taxi lights and the smallest, 12 volt personal aircraft they are much higher wattage.
Pg 23-22 Age Sag
Age sag is observed in high time, halogen headlights and all high time headlights exposed to rough service and higher operating temperatures.
Pg 23-26 Tampering
The swirling of the gasses inside the lamp is not due to vehicle motion but rather to the jet of oxygen and nitrogen entering through cracks in the envelope. The swirling is readily reproduced on the test bench without any motion of the envelope relative to the surrounding air.
Pg 23-31 Opening Motor-vehicle Lamps
Because of the risk of losing very small arc bead fragments, opening an envelope should not be permitted until examination with an inverted stereo microscope has been completed.

Frederick G. Hochgraf is professor emeritus of Materials Science at the University of New Hampshire, and is the senior scientist at New Hampshire Materials Laboratory, in Dover, New Hampshire. He can be reached at (800) 334-5432.

Mechanical Forensics Engineering Services, LLC.

This page last modified 08-NOV-2001