Spontaneous Ignition, Part II: Investigation

by
John D. DeHaan

DeHaan, John D. Spontaneous Ignition, Part II: Investigation.
Fire and Arson Investigator.Vol. 46, No. 4 (June 1996). p 8-11.

[interFIRE VR note: Figures referred to in the text have not been reproduced in this reprint.]

INTRODUCTION:

Investigators must be aware that spontaneous ignition can take place given the right fuels and the right circumstances. Investigation of fires caused by self-heating, because of the balances of time and heat necessary, can be very difficult. The first step in any investigation is a careful determination of the actual area of origin. Before a great deal of time and effort is expended in considering spontaneous ignition processes, the investigator must be absolutely certain that the fire actually started in that suspected origin. Eyewitness reports, information from fire suppression people, and a careful interpretation of burn patterns, relative charring, calcination, beveling, and all the other indicators must be thoroughly carried out. Once the area of origin is determined the hunt usually focuses on the presence of significant residues.

PHYSICAL EVIDENCE

It should be remembered that there often are no significant post-fire residues left by a self-heating process, because there are no additional elements of fuel for ignition necessary other than the material itself. Because the entire mass of reacting material may be at or close to ignition when flaming combustion ensues, the chemical detection of residues of the preliminary reactants may be impossible. Laboratory tests conducted by the Center of Forensic Sciences after observed ignition of linseed oil on cotton cloth revealed that extraction of the unburned cloth yielded virtually no unreacted linseed oil which would have permitted the chemical identification of the mechanism responsible. ¹ Over time, the temperature throughout the reacting mass has had a chance to reach the point at which all of the reactant has begun to be involved, and a subsequent fire, even of short duration or subject to extinguishment attempts, may obliterate any significant chemical residues. Because there is a possibility of detecting unreacted materials, samples of the debris should always be collected and preserved in a sealed can or glass jar for later examination. One alternative is to determine if there are any similar containers or equipment in the vicinity or elsewhere in the area that can be sampled. One successful investigation involving furniture refinishing rags revealed several cans in the area of origin to be less damaged than the ones that actually ignited. Examination of those cans revealed all the preliminary stages from discoloration right through charring, resulting in identification of the mechanism involved. ²

Sometimes a residue is discovered, not of the original reacting material but of the final reactant products. This is sometimes found after fires in large accumulations of hay as a glassy, irregular, green or gray matrix that the investigator usually suspects is the residue of an incendiary device. Such a "hay clinker" is illustrated in Figure 1. This clinker was recovered from the center of a large haystack fire and subsequent laboratory analysis by this author revealed the presence of silicon oxide, calcium oxide, and sodium oxide. All organic material contains small quantities of silicon, calcium, sodium, magnesium, and other trace elements. Those materials oxidize as all of the carbonaceous material is burned away. These materials, when heated together at the high temperatures produced in the center of a reacting mass like a haystack, can fuse together to form a naturally occurring glass. Such materials do not include any incendiary residues.

DIAGNOSTIC SIGNS:

The most diagnostic sign of spontaneous ignition is the presence of more damage to the center of a mass of material than at the edges, as in Figures 2 and 3. In a "normal" fire, exposure from outside the fuel mass results in more damage to the outside layers of fuel than to the interior. This is the result of the low heat conductivity of the typical cellulosic fuel. In spontaneous ignition the most significant fire damage will most often be in the middle of the fuel mass, often isolated from the surrounding container or edges of the mass itself. In very large masses, there may be multiple isolated pockets of smoldering with occasional flares reach the exterior (Bowes, p.388).

This damage in haystacks, masses of cotton fabrics, and similar cellulosics is often accompanied by a brown discoloration and an acrid odor. It has been suggested that this brown discoloration is not simply charring but the product of reactions between amino acids and reducing sugars. Such reactions, called "Maillard" reactions, produce dark brown pigments (Bowes, p.401). Such discoloration, acrid odors and properties often characterized by investigators as "acidic" are associated most predominantly with drying oils on cotton cloth or in haystack or feedstock ignitions. They will not be expected in the ignition processes of pyrophoric carbon and similar solid materials.

The heat generated by materials as they self-heat towards spontaneous ignition evaporates moisture from the surrounding fuel, which may be seen as smoke. Such smoke is usually detectable for an hour or more prior to the actual ignition of the material. In later stages of the ignition process, this smoke will include aldehydes (pentanal, hexanal, etc.) and possibly acrolein and other chemicals which have distinctive acrid odors ³ Anyone in the vicinity with a normal sense of smell will be able to detect these odors for sometime prior to the actual flaming ignition. In tests conducted by the Center of Forensic Sciences and the Ontario Fire Marshal's Office, boiled linseed oil on cotton rags in large waste containers at an ambient temperature of 16-18°C produced smoke and odors within one hour of initiation of the test followed typically by flaming/ignition some 4 to 5 hours later ⁴ . In one instance where there was a low ambient temperature (12-14°C) the smoking was followed by ignition some 12 to 14 hours later. The interior temperature of the reacting mass was on the order of 100°C at the time this smoking was observed. In addition to the acrid odor, there are some lachrymatory components. Watering eyes of people discovering the fire or attempting to fight it may be an indicator of an ongoing chemical process. Such organic volatiles are not produced by chemical processes such as catalytic reactions and will not be detectable in such fires. Experts familiar with the chemistry of suspected chemical processes may be able to advise the investigator as to what volatiles may be produced prior to flaming ignition.

One problem which may give misleading clues is the possibility of a very hot ember, created externally, which drops into the middle of a loosely divided mass of material such as coal, feedstock or sawdust. Such hot embers could come from welding slag, industrial processes, overheated bearings (from ventilation equipment), or even a burning cigarette or cigar. Dropping into a loosely divided mass of material they can bury themselves deeply. The charring induced locally by a hot object may take hours or even days to percolate through to the surface before it is detected. The detection of residues of hot material may require careful sifting of the ashes of the reacting material (even down to very fine mesh sieves) before such particles are found. In such instances, other signs of spontaneous ignition such as the discoloration and the acidity mentioned before will be absent.

Finally, the fire's behavior when suppression is attempted is a helpful indicator. Because the reacting mass is deep within the mass of material, initial firefighting such as the application of water, carbon dioxide or solid chemical fire extinguisher may result in only momentary suppression of the flames followed by re-ignition. In larger masses of material such as haystack or feedstock fires, the exposure of the material by shovel, tractor, or even by straight-stream hose application may expose enough preheated fuel to open air to cause a flash of flame followed by subsequent difficulty in suppression. The reaction of the fire to firefighting efforts can only be documented by interviewing the first-in fire companies.

FOLLOW-UP:

If the scene investigation has confirmed the particular area of origin and the preceding inquiry has resulted in detection of a possible mechanism, there are a number of additional steps the investigator can take. First of all, the materials involved can be compared against the NFPA Fire Protection Handbook list cited previously (Table 5-11). ⁵ The physical form of the materials involved is critical. Bulk solids need to be porous and produce a rigid char. Bulk liquids will only be suspect if there is a potential for bulk process heating such as catalytic reactions. Other chemical reactions such as hypergolic mixtures or drying oils (in suitable form) may be identifiable from the surroundings or by the circumstances. The size and time considerations are interlinked with the type of material suspected of involvement. Generally the larger the mass, the longer the time required to initiate the problem. Situations involving hot materials (such as freshly made plywood or hot laundry) require a fairly large mass but may also result in flaming combustion in short spans of time because the "preheating" essentially "short-circuits" the early portion of the fire development. Materials that will react in small masses such as hypergolic mixtures or drying oils, of course, will react in time frames much shorter than the larger mass materials. Ignition in less than an hour is not unheard of for reactive materials like linseed oil on cotton rags, especially at high ambient temperatures.

A more difficult determination is the estimation of sources of extra heating which can increase the ambient temperature of the materials. This extra heating can be obvious things like a spell of sunny or very hot weather, a change in location of the material from a cool environment to a much warmer one, or equipment running which creates heat (either directly by contact or indirectly, through wires or cables through the storage area). Even light bulbs can generate localized heat when buried in finely divided materials or increase the ambient temperature conditions enough to trigger the onset of spontaneous ignition when located above the material in the same compartment.

LABORATORY TESTS

Chemical or thermal laboratory tests may help establish the susceptibility of some materials. The frequency of fires in cotton and linen which had been treated with various oils to make processing easier created a demand for a suitable test in the 19th century to establish which oils were safe to use for that application. The Mackey test was first documented in 1895. ⁶ It involved taking a sample of a fabric with the oil on it and exposing it to a steam bath at constant 100°C for two hours. The temperature of the test mass was measured directly by thermometer. If that test mass reached 200°C in the test time (typically two hours) the material was considered unsuitable for use. The original Mackey test was a good start, processes which take more than two hours under 100°C ambient conditions would generally escape detection under that scheme. Additives such as antioxidants would delay the oxidation of the oils for some period of time not necessarily duplicated in the test. If the actual processing of the fabric resulted in exposure of the treated fabric to high temperatures for longer periods of time, ignition could ensue, even when the material actually passed the original Mackey test. With those limitations, the Mackey test was modified for the circumstances in each potential fuel situation. It is now conducted at various temperatures with varying times to try to make it applicable to a wider range of problems.

A better understanding of the mechanisms involved brought improved tests since the 1930s. Because the likelihood of spontaneous ignition varies with the mass of material as well as its size and shape, a more generalized evaluation is necessary. The Frank-Kamenetski evaluation carries out tests similar to the Mackey test on fuel masses of various sizes, and then the reactions of the fuel at those various sizes is used to extrapolate to the likelihood of ignition conditions for larger masses. The tests involve multiple tests of, for instance, fuel elements three centimeters, six, twelve and thirty centimeters on a side. Then those test results are plotted and can be extrapolated to fuel masses as large as three meters on a side with some regularity and reliability. ⁷ This bracketing method takes into consideration the effect the bulk of the fuel has on L versus Q, that is the surface-versus-volume balance that was discussed in Part I of the paper.

Processes which had been on-going for some time (either repeatedly or continuously) with no previous problems involve a more delicate investigative approach. Here the investigator must determine if any changes occurred which could affect either Q (the heat generated) or L (the heat lost) as a result of those changes. Changes such as the ambient temperature of storage (of either raw materials or the finished product) would be an obvious factor, especially changes to a higher ambient temperature which would have the predictable effects on the L factor. A more subtle problem is one involving sudden downward changes in the ambient temperatures where, for instance, material is stored for long periods of time at high temperature and then moved to a much colder climate. This change could result in the condensation of moisture inside the storage compartment. Some reactions, particularly those involving coal, charcoal, and grain or feedstock products, are triggered by the sudden increase in moisture via condensation. Changes in processing conditions such as the temperature of either the raw material or the process, or any change in the amounts of materials (especially to larger batches then had been involved before) may trigger a self-heating problem. Changes in formulation may have obvious effect on the heat generated. Raw materials from new sources could include contaminants or trace materials that could trigger exothermic reactions that had not occurred previously. Changes in the start-up or shutdown procedure may affect the heat loss rate, especially for continuous operations where the finished product is normally drawn off at a fixed rate and the shutdown procedure results in the storage or accumulation of some of the finished product. If the material has been stored for a longer period than usual that may be enough time to reach critical conditions. If materials are moved out, shipped or divided up at a particular time, the process may have been operating in the critical range, with the problem not detectable until a strike or a interruption in rail service or distribution causes delays in shipping. If the stock has been moved and repacked (typically taking something that was spread out and moving it into a single storage-bin, thereby consolidating a number of small masses into a large mass), this will increase the volume to surface ratio. Modifications to equipment or changes in the ventilation conditions may affect the rate at which heat is dissipated.

POSSIBLE SCENARIOS

There are numerous situations that are subject to self-heating and of concern to the fire investigator. The first is the formation of a solid carbonaceous material sometimes described as "pyrophoric" carbon. It may seem odd to take wood with an ignition temperature of 200-300°C and have it ignite upon exposure to temperatures as low as 120°C. The charring, or pyrolysis of wood can result in the formation of a 'new' material whose Q curve is very different from that of freshly cut wood, as in Figure 4. Such a change could result in spontaneous ignition even if the corresponding L curve has not changed from "normal" conditions. The formation of pyrophoric carbon occurs upon exposure to low temperatures (120-150°C) for very long periods of time, typically weeks, months, or even years. Carbon or char produced under open flame conditions is produced under much higher radiant heat exposures and its Q curve will be correspondingly different from the most reactive or "pyrophoric" form.

Investigations involving the contributions of drying oils are more common for the ordinary fire investigator, but also much more difficult to identify. The drying oils involved are those non-saturated oils found in vegetable and animal products. The saturated hydrocarbon oils such as found in petroleum products are not subject to self-heating or spontaneous ignition at ordinary temperatures. The materials of concern here are vegetable oils such as linseed, rapeseed, cottonseed, peanut, and sometimes even corn oil and safflower oil (such as those found in cooking products). Fish oils and fish meal are highly subject to self-heating. In addition, lards and animal oils are also subject to self-heating but at a much lower rate. The key here is that these materials are not subject to significant self-heating when present as bulk liquids When spread across large surface areas such as on porous cloth, however, they are exposed to a great deal of oxidizing air. It is the combination of the oil with the oxygen in the air that allows the drying process to occur, and it is that process which generates the heat. Even very small quantities (grams or ounces) of such oils have been documented in laboratory tests to be able to lead to spontaneous ignition of their support materials when the large-surface-area support is reduced to a small volume by being wadded up in a can or pile rather than being spread out to dry.

Tests conducted by the Center of Forensic Sciences and Ontario Fire Marshal in September 1991 revealed the susceptibility of boiled linseed oil when placed in covered buckets and exposed to ambient temperatures as low as 15°C. ⁸ Temperatures were recorded via computer-monitored thermocouples. The warming process started immediately upon assembly of the tests. In some instances smoke and acrid smells were reported at the end of one hour with temperatures on the order of 100°C measured near the center of the mass. Ignition into open flame combustion occurred several times at the 4-6 hour mark at temperatures on the order of 400°C.

Howitt, Zhang and Sanders have examined the self-heating of linseed oil in some detail and report that while linseed oil contains the mixed glycerides of oleic, linoleic and linolenic acids, it is the oxidation of linoleic and particularly linolenic acids which occur most rapidly at room temperatures. ⁹ It is the linolenic acid which is principally responsible for oxidation and polymerization. Raw linseed oil will oxidize and polymerize but it does so at a very slow rate (such that self-heating is so low that ignition is rarely observed). Its drying properties can be accelerated by boiling and separating out the denser components. In modern coatings, the raw linseed oil is modified by the addition of chemical drying agents or catalysts. These are often cobalt or zirconium salts, but salts of other heavy metals such as lead can be used. These chemical additives can accelerate the oxidation process much more than boiling. As a result, modified linseed oils can polymerize and self-heat at much faster rates, with the resultant increased likelihood of ignition. Howitt, et al report that a single rag the size of a handkerchief dampened with one of these drying agent-modified linseed oils can burst into flames some 6-8 hours after exposure (at room temperatures) and can continue to burn for an hour or more after flaming ignition. The presence of volatile petroleum solvent in consumer products may delay auto-ignition by cooling the reacting mass by evaporation but may not prevent it. The specific product potentially involved should be identified by the scene investigator.

Another mechanism is most commonly associated with commercial laundry facilities, but once in a while is seen in laundries of hospitals, military, and even residences. In these instances, laundry removed from a dryer prior to the cool-down cycle is dumped or stacked hot in hampers. Cotton or linen, being cellulosic materials, will undergo oxidation even at normal temperatures but this is rapidly accelerated as the ambient starting temperature is greatly increased by the laundry process. If the materials are stacked or binned at high enough temperatures (>90°C), the heat accumulated in the center of the mass may be adequate to trigger spontaneous ignition of the cotton materials. There is at least one mention in the literature of synthetic materials igniting under similar circumstances; however, most synthetics do not form rigid char and tend to melt as they approach their ignition temperatures. Unless supported by a rigid-char material (such as being mixed with cotton fabrics) they are less likely to sustain self-heating beyond that melting temperature. Any fire in a laundry facility should bring both possibilities in mind.

Since clean cotton self-heats with some reluctance, it has been suggested that the residues of cooking oils in fabrics (especially in clean up towels from restaurants and similar food processing facilities) would make them more likely to ignite spontaneously when the contaminant is corn, peanut, cottonseed, or other edible oil. Fish oils and animal fats could also contribute to the onset of spontaneous ignition. It has been suggested that with today's laundries (even home laundries) operating at lower water temperatures with less aggressive detergents, there may be a higher concentration of these oils left in the fabrics after the laundry process. If that less-than-total cleaning is followed by drying at high temperatures such as those found in commercial laundries, a possibility of spontaneous ignition would be correspondingly higher. The CPSC has issued a warning regarding such laundered materials - CPSC Recall #92-40. ¹⁰ One case involving new (unused) cotton "table linens" has suggested another possible mechanism. These items were imported and it is entirely possible that the oils used in the production of fabrics in foreign countries need not necessarily pass any of the tests (including the Mackey test described earlier). After laundering a quantity of tablecloths and napkins were placed in large bags and were found to be charred a few hours later. ¹¹

Dixon recently reported that a food product using grain flour fried in canola (rapeseed) oil was found to be subject to self-heating when the initial high temperatures of cooking were not dissipated properly. ¹² He suggests that fires in kitchens be carefully evaluated for self-heating mechanisms when edible oils have been exposed to high temperatures and then presented on porous, combustible substrates such as food or towels.

Hay, feedstocks, and other agricultural waste such as lawn trimmings, silage, and bagasse have to be considered in their turn. These materials require larger quantities than is typically found in residential property, and correspondingly long storage conditions.

The microbiological mechanism (requiring presence of moisture) is a typical ignition scenario but according to Gray's work not completely essential. ¹³ Properly prepared silage is too wet and too well-compacted to allow aerobic processes (the desired decomposition being anaerobic). If the silage is too dry or loosely packed, oxidation and self-heating can set in. Once again, if the mass of material and the time necessary fits, then spontaneous combustion should be considered a possibility with these materials.

Coal dust and charcoal, especially in finely powdered forms, are some of the most notoriously self-heating materials in the industry. Finely divided coal has been known for more than a hundred years to ignite spontaneously. Such products are especially susceptible when wetted and then dried back out again, and the drying process creates the most significant temperature rises. As the material dries, it is less conductive to heat. This would result in lower loss rates which could be enough to trigger spontaneous ignition. The presence of water may also add to the risk of ignition due to the heat released when water is absorbed by charcoal.

Materials which are produced hot, such as plywood and fiber-reinforced plastic moldings, if stacked and stored in large quantities while hot, may be susceptible to spontaneous ignition. In this case no degradation is required; the oxidation of the material triggered by the high temperature, coupled with stacking or storage conditions which do not allow heat to dissipate, is all that is required. Heat generated by that oxidation is then trapped in the matrix of material and triggers more rapid oxidation. Given enough insulation (and limited convective loss) the process could result in ignition of the entire mass. The reader is referred to references by Bowes and Beever for comprehensive discussions of mechanisms and chemistry of other possible reactions.

Chemical reactions, because of the variety of exothermic reactions possible within solid materials, gases and bulk liquids, have to be considered a possibility depending on their chemical nature. Once again, the possibilities are beyond the scope of this preliminary treatment and experts in the processes involved should be contacted.

SUMMARY

Spontaneous ignition is a perfect expression of the delicate balance between factors involved in any fire: the fuel present, the heat that fuel generates upon oxidation, the ventilation conditions necessary for both introduction of oxygen as well as the removal of heat and products generated, and time. Because this delicate balance is sometimes difficult to identify by scene examination, it is easily overlooked by the fire investigator who is used to short time frames and intense fires. Nevertheless, it is a potential cause of fires that all investigators need to consider when conducting a thorough investigation.

References

1. Center of Forensic Sciences and Ontario Fire Marshal "Spontaneous Combustion" (Video). Ontario Fire Marshal, Toronto, Canada, 1992.

2. BATF - Milwaukee, WI. Private communication, 1991.

3. Whirlpool Corp., "Spontaneous Combustion of Vegetable Oils on Fabrics," Whirlpool Corp., 1991.

4. Center of Forensic Sciences "Report on Self-Heating Tests - Sept. 26, 199" (Unpublished data). Also, BM Dixon, Spontaneous Combustion: The Journal of the Canadian Association of Fire Investigators, March 1992.

5. NFPA, Fire Protection Handbook, 17th Edition, Quincy, MA, 1991, pp 5-130 - 5-133.

6. Mackey. WM, J. Soc. Chem. Ind., Vol. 14, 1895, p. 940.

7. Drysdale, D, Introduction to Fire Dynamics, John Wiley & Sons, 1985, p. 195.

8. Center of Forensic Sciences "Report on Self-Heating Tests - Sept. 26, 1991" (Unpublished data). Also, BM Dixon, "Spontaneous Combustion: The Journal of the Canadian Association of Fire Investigators, March 1992.

9. Howitt, DG, Zhang, E, and Sanders, BR, "The spontaneous combustion of linseed oil." Presented at the 20th International Conference on Fire Safety, Burlingame, CA, Jan. 1995.

10 Fire & Arson Investigator, Vol. 42, #4, June 1993 p. 30.

11. Brogan, R, New South Wales Fire Brigade, Private communication, 1992.

12. Dixon, BM, "The potential for self-heating of deep-fried food products" presented at the FBI Arson Symposium, Fairfax, VA, Aug. 1995.

13. Gray, BF, "Spontaneous Combustion and its Relevance to Arson Investigations." Paper presented at IAAI-NSW Annual Conference, Sydney, Australia, Sept. 1990.

Suggested Reading

Bowes, PC, (1984), Self-heating: Evaluating and controlling the hazards, Elsevier, Amsterdam, 1984.

Beever, PF, "Self-heating and spontaneous combustion," SFPE Handbook on Fire Protection Engineering, 1st Ed. SFPE pp 1-341 - 351.

Dixon, BM, "Spontaneous Ignition, Part II," The Journal of the Canadian Association of Fire Investigators, June 1992.

About the Author

A native of Chicago, Illinois, John DeHaan graduated from the University of Illinois-Chicago Circle in 1969 with a B.S. degree in Physics and a minor in Criminalistics. DeHaan worked for four and a half years as a criminalist with the Alameda County Sheriff's Department Crime Lab in Pleasanton, California and nine and a half years with the California Department of Justice - Sacramento Regional Lab. He then became a Criminalist/Physical Scientist with the Bureau of Alcohol, Tobacco and Firearms, U.S. Treasury Department at the San Francisco Laboratory Center. Over these years he developed considerable expertise in arson and explosive evidence, human hair, shoe prints, instrumental analysis and crime scene reconstruction. Additionally, John pursued extensive individual research projects in each of these areas and published over thirty research papers and technical reports in a variety of journals in the forensic, police and fire literature. In 1987, John DeHaan was selected as program manager at the California Criminalistics Institute, responsible for the Institute's training programs in low explosives, arson accelerant detection, hair, microscopy, and latent fingerprints. He was awarded his PH.D. in Pure and Applied Chemistry (Forensic Science) by Strathclyde University in Glasgow, Scotland in 1995.

His initial interest in fire-related evidence was kindled by analyses of fire debris conducted while at Alameda County but has since extended to include research into fire behavior and diagnostic signs. This research is based on first-hand fire experiments involving full scale structure fires under controlled conditions. He has been involved in hundreds of fire and explosion cases and has been an expert witness in civil and criminal trials across the U.S. He has helped produce several training videotapes through the California Department of Justice and the Arson-Bomb Investigation Unit of the California Fire Marshal. Lectures have been given by him before police, fire, and forensic audiences across the U.S., Australia, Canada and the United Kingdom. Among his publishing accomplishments are portions of the California District Attorneys Association Manuals Arson Investigation and Arson Prosecution, and co-authoring the chapter on "Gas Chromatography in Arson and Explosives Analysis" in Gas Chromatography in Forensic Science, (Ian Tebbett, Ed.). John authored the widely read textbook, Kirk's Fire Investigation, 2nd. Ed. (1983) and 3rd. Ed. (1991).

To maintain international participation in fire investigation, John has been a Fellow of the American Academy of Forensic Sciences since 1975, and serves on its editorial board; is a member of the International Association of Arson Investigators; is active on the IAAI Forensic Science Committee; and is a member of the National Fire Protection Association and serves on its Committee on Fire Investigation. He was recently elected Associate of the Institution of Fire Engineers, (United Kingdom). He has been a member of the California Association of Criminalists since 1971 and was its President 1983-1984. He holds a Diploma in Fire Investigation from the Forensic Science Society and is a Fellow of the American Board of Criminalistics (Fire Debris Speciality).

Reprinted with permission.