Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 SMOKE TOXICITY TESTING INTRODUCTION This report is primarily concerned with toxic hazards associated with fires. However, the overall fire hazard (i.e., fatalities resulting from the inability of potential victims to escape from a fire) should actually be considered in terms of three major factors: (1) obscuration of vision, (2) heat, and (3) toxicity. The limits of human tenability for each of these factors have been reasonably well defined; whichever of the three factors first precludes escape sets the critical tenability limit for that fire scenario. Experience has shown that individual tenability limits are often reached in the order stated above (i.e., obscuration of vision and incapacitation due to heat generally occur before the tenability limit for inhalation of toxic gases is reached). This is especially the case with rapidly developing, flaming fires. In terms of lethality, however, about 70 percent of fire fatalities result from smoke inhalation (Hardwood and Hall, 1989~. Smoke most often Is defined as the airborne solid particulates and liquid aerosols and fire gases evolved when a material undergoes pyrolysis or combustion (ASTM, 1982; see also Appendix B of this report). The fire gases have received the most attention, while knowledge of the effects of inhaling particulates and aerosols from smoke is still quite limited. Although a wide variety of combustion products may be generated (Tables 4-1 and- 4-2), the toxicants are usually separated into three classes based on type of effect: asphyxiants; irritants, which may be sensory or pulmonary; and to~cicants exhibiting other or unusual toxic effects. Asphyxiants can cause central nervous system depression, with loss of consciousness followed by death. The effects of asphyxiant-producing toxicants depend on the 25
26 TABI E 4-1. Toxicological effects of combustion products from polymers. TOXICANT | SOURCES | TOXICOLOGICAL EFFECTS _ ~ Hydrogen cyanide (HCN) | From combustion of wool | A rapidly fatal asphyxiant | silk, polyacrylonitrile9 nylon' poison. polyurethane and paper. Nitrogen dioxide (NO,) adduced in small quantities | Strong pulmonary irritant other oxides of nitrogen from fabrics and in larger capable of causing immediate quantities from cellulose death as well as delayed nitrate and celluloid. injury. Ammonia (OHS) | Produced in combustion of | Pungent unbearable odor; | wood, silk, nylon and irritant to eyes and nose. melamine; concentrations generally low in ordinary building fires. Hydrogen chloride (MCI) ~ From combustion of ~ Respiratory irritant; potential polyvinyl chloride) (PVC) toxicity of HC1 coated on and some fire-retardant particulate may be greater treated materials. than that for an equivalent amount of gaseous HCI. , Carbon monoxide (CO) ~ From combustion of carbon- ~ Absorbed via the lungs into |i containing polymers. the bloodstream and CO2/CO ratio dependent on combining with hemoglobin oxygen supply. to the exclusion of oxygen resulting in arterial hypoxia. Other halogen acid gases l From combustion of | aspiratory irritants. (HF and HBr) fluorinated resins or films and some fire-retardant l materials containing bromine. Sulfur dioxide (SO2) From materials containing A strong irritant intolerable sulfur. well below lethal l concentrations. . Isocyanates From urethane polymers; Potent respiratory irritants; pyrolysis products such as believed the major irritants toluene-2 4-diisocyanate in smoke of isocyanate-based (TDI) have been reported in urethanes. small-scale laboratory studies; their significance in actual fires is undefined. Acrolein From pyrolysis of polyotefins Potent respiratory irritant and cellulosics at lower | temperatures (-400°C). l l Source: Adapted from Terrill et al. 1978.
27 TABLE 4-2. Volatile products from the pyrolysis of poly(m-phenyleneisophthalamide) fabric.. , . ~ COMPOUND QUANTITYb- 1 ' - 11 Carbon monoxide 32. Methane 2.9 Carbon dioxide 56. Nitrous oxide Ethylene Acetylene Ethane Cyanogen Propane Hydrogen cyanide Water Acetonitrile Acetone Propenenitrile Acetic Acid 3-Butenenitrile Benzene Butenenitrile Dio~cane Toluene Chlorobenzene Xylene Phenol Benzonitrile Toluonitrile Dicyanobenzene TOTAL 0.01 0.027 0.003 0.014 0.001 0.014 0.44 0.27 0.020 0.029 0.43 0.17 7.1 0.020 0.037 1.9 0.033 0.13 0.10 13.6 1.2 1.1 170.55 Pyrolyzed in nitrogen at 550°C. b Milligrams of compound produced per gram of polymer as measured by analytical techniques. Source: Adapted from Einhorn et al., 1974.
28 accumulated doses (i.e. both the concentration and the duration of exposure). The severity of the effects increases with increasing doses. Although many asphyxiants may be produced by combustion of materials, only carbon monoxide (CO) and hydrogen cyanide (HCN) have been measured in sufficient concentrations in smoke to cause significant acute toxic effects. An atmosphere deficient in oxygen, caused by its consumption in a fire, is also considered an asphyxiating condition. Irritant effects, produced in essentially all fire atmospheres, are normally considered by combustion toxicologists as being of two types: (~) sensory irritation,, or the illicitation of pain in the eyes and upper respiratory tract, and (2) pulmonary irritation, which may not cause a sensation of pain but does cause tissue damage in the lungs, leading to edema and a decrease in functional ventilation. Most irritants produce signs and symptoms characteristic of both sensory and pulmonary irritation. Eye irritation, an immediate effect that depends only on the concentration of an irritant, may be underestimated in its ability to impair a person's escape from a fire. Nerve endings in the cornea are stimulated, which causes pain, defied blinking, and tearing. Severe irritation may lead to subsequent eye damage. Victims may shut their eyes, which can partially alleviate these effects but may also impair their escape. Airborne irritants enter the upper respiratory tract, where nerve receptors are stimulated, with burning sensations in the nose, mouth, and throat, along with the secretion of mucus. Sensory effects are related primarily to the concentration of the irritant and do not normally increase in severity as exposure time increases. Following signs of initial sensory irritance, significant amounts of inhaled irritants may be quickly taken into the lungs, with symptoms of pulmonary or lung irritation being exhibited. Lung irritation often is characterized by coughing, bronchoconstriction, and increased pulmonary flow resistance. Tissue inflammation and damage, pulmonary edema, and subsequent death often follow exposure to high concentrations, usually after 6 to 48 h. Exposure to pulmonary irritants also appears to increase susceptibility to bacterial infection. Unlike sensory irritation, the effects of pulmonary irritation are related both to the concentration of the irritant and to the duration of the exposure. None of the smoke toxic potency tests to be described explicitly measures either sensory or pulmonary irritation. A standard test does exist for sensory irritation (ASTM, 1984) based on respiratory rate depression in mice; however, its relevance to incapacitation of humans expose to a fire atmosphere has been questioned (Potts and Lederer, 19781. In the absence of a test of demonstrated validity, mass loss tenability limits have been proposed for materials known to produce irritants in a fire (Purser, 1988~. This approach is described under Human Incapacitation Movie! in Chapter 5. The third general class of fire toxicants, those exhibiting either unusual toxic effects or unusual toxic potency, has few documented examples. One involved the formation of a neurotoxin from the thermal decomposition of a non-commercial rigid polyurethane foam (Voorhees et al., 1975), while another was the unusual toxic potency exhibited by polytetrafluoroethylene in certain laboratory tests. The latter case would now appear to be largely an artifact of the test method (Baker and Kaiser, 1991~.
29 BASIC PRINCIPLES OF SMOKE TOXICITY TESTING Under the assumption that the toxicity of smoke produced from the burning of materials could well be as complex as its chemical composition literally hundreds of constituents may be present early efforts in the 1970s were directed toward the development of laboratory tests for the smoke toxicity of materials. It was believed that only ~ bioassay (i.e.' an animal exposure procedure), would be a reliable wav of evaluating the combined effects of gases found in smoke. ,, ~ The various smoke toxicity tests have been extensively reviewed elsewhere (Kaplan et al., 1983), and, except for those having significant utilization today, a comprehensive review will not be repeated here. It is, however, of relevance to acknowledge that two key issues are common to smoke toxicity tests. 1. A relatively small laboratory combustion device (generally termed the "fire models) may or may not produce a fire effluent/combustion atmosphere that has the same composition as an uncontrolled fire. 2. Toxicity data obtained from the exposure of laboratory animals (usually rodents) may or may not be extrapolated both qualitatively and quantitatively to humans. The first issue addresses the relevance of the laboratory combustion device; the second, that of the biological model. The Laboratory Combustion Device In an effort to clarify the issues and bring a more systematic and scientific approach to the problem, the International Organization for Standardization (ISO)/Technical Committee 92/Subcommittee 3 (Toxic Hazards in Fire) established working groups to address separate aspects of the toxicity testing of fire effluents. (Technical reports are still in various stages of development, with only one having been issued by ISO as TR 9122-1, Toxicity Testing of Fire Effluents Part 1: General.. ~ The first major contribution of an ISO working group was to set forth a systematic classification of the types and stages of fires to be considered, along with the relevant characteristics of each. This Is shown in Table 4-3. The ISO group identified certain criteria for evaluating laboratory combustion devices used in the testing of materiab for smoke toxicity. Particularly significant was the criterion for relevance to uncontrolled! fires; the most impor~t considerations were identified as ventilation (oxygen availability), CO2 (carbon dioxide)/CO ratios, temperature "~/or heat nux, and residence times of fire effluents in the high-temperature zone. Realizing that no one laboratory combustion device (fire moclel) can replicate or simulate all the features of all fires, the ISO committee recommended that the selection of an appropriate fire mode} must be made for the particular fire conditions of interest. Although not specifically recommending any one fire model, the ISO committee directed that the choice of a laboratory combustion device be consistent with a good understanding of the characteristics of the real fire to be simulated. \
30 TABLE 4-3. Classification of fires. 1 . Fi re Oxygen Ratiob Temperatureb IrradianceC (%) (CO3/CO) ~ °C) (l`W/m2) _ Decompositon ~ ~ ~ ~ I. Smoldering (self-sustained) 21 N/A <100 N/A Nonflaming (ox~dative) 5 to 2 ~ N/A <500 <25 Nonflaming (pyrolytic) I <5 | N/A I ~ COO | N/A |- Developing fire (flaming) ~ 10 to 15 ~ 100 to 200 ~ 400 to 600 ~ 20 to 40 ~ _ _ Fully developed (flaming) Relatively low ventilation I to 5 <10 600 to 900 40 to 70 Relatively high ventilation ~ 5 to 10 ~ <100 ~ 6001' 1200 ~ 50 to 150 | a Mean value in fire plume near to fire. b General environmental condition (average) within compartment. c. Incident irradiance onto sample (average). Source: ISO, 1989. The Biological Mode! The issue of relevance of an animal in modeling human exposure was addressed by considering the effects of asphyxiants and irritants separately. Comparisons of the amounts of the major toxic gases producing a given effect (e.g., lethality or incapacitation) with rodents to those estimated to be required for humans revealed that laboratory animals are a reasonable model in the case of the asphyxiant fire gases, CO and HCN (Hartzell, 1989~. Similar analysis showed that rodents are also a reasonable movie! for lethality due to exposure to pulmonary irritants. For sensory irritants, the rodent models were judged to be inadequate (Purser, 1988~. Smoke toxicity should be evaluated on the basis of measuring the response of test animals exposed for a specified period of time (Klimisch et al., 19871. Rodents, usually rats or mice, are normally to be used. Lethality is to be the most commonly measured response, although some test methods also obtain a measure of the animal's incapacitation. For the biological response, the relationship Is determined between the response of the test animals and exposure to different concentrations of smoke. This Is accomplished by conducting a series of experiment in which the quantity of material combusted or the flow rate of diluting air Is varied to produce different concentrations of smoke. The number of animals that show a response, such as lethality or incapacitation, increases as the concentration is increased. In combustion toxicology, smoke concentration is traditionally expressed either as the quantity of test material used per chamber unit volume (the material charge concentration) or the material mass loss per chamber unit volume (the smoke concentration). Test methods that employ dynamic or flow-through generation of smoke usually express concentration simply as the mass of material charged. (The mass per unit volume can be calculated from the air flow rate.) If the percentage of animals responding within a specified time is plotted as a function of the logarithm of the concentration, a straight line is approximated. Such a plot represents the concentration-response relationship
31 of smoke produced by a material under the experimental conditions of the particular test method. The concentration that would produce a response or effect in 50 percent of the animals upon exposure for the specified time is obtained from the data by use of a statistical calculation. This concentration, commonly termed the EC50, is a measure of the potency of the smoke. The EC50 is a general term and may be used in reference to any measured response (i.e., effect) of the animal. When lethality is the measured response, the LC50 (lethal toxic potency measured in gums for a specified period of time) is used as a more specific term to denote the concentration of material or smoke that produced death in 50 percent of the animals. Although some test methods do report other response measurements, the LC~0 is the most commonly reported measurement of toxic potency in smoke toxicity testing. Being a statistical result, the LCS0 value is accompanied by confidence limits, usually at the 95 percent level. This gives a range of values for the LCSo, the significance of which is that an investigator testing a particular material would have 95 percent confidence that the LC50 would fall within the given range. Physiological responses are usually dose related (i.e., the magnitude of the effect increases with increasing dose or accumulated body burden of a physiologically active agent). Since the real dose of toxicants from smoke inhalation cannot be directly measured, it is assumed that the dose is a function of smoke (or toxicant) concentration (C) and exposure time (t) (MacFarland, 1976~. This Closes is really an expression of the insult to which a subject is exposed. Thus, Exposure doses (Ct) has become the preferred term in combustion toxicology to quantify an exposure either to smoke or to pure fire gases. Concentrations of common fire gas toxicants, such as CO and HCN, are usually expressed as parts per million (ppm) by volume. Therefore, the exposure dose can be expressed as the product of the concentration and time (i.e., ppm · min). In the case of a changing concentration of a gaseous toxicant, the exposure dose is the integrated area under a concentration vs. time curve. Often, the concentrations of fire gas toxicants may not be known. In that event one can still deal with the concept of exposure dose as it applies to smoke. Since smoke concentration cannot be quantified, an approximation is made that it is proportional to the mass loss during a fire. The integrated area under a mass loss per unit volume versus time curve thus becomes a measure of smoke exposure dose (e.g., g~m~~.min). Of the numerous smoke toxicity test methods that have been used, only a few are plausibly relevant to uncontrolled fires, are adequately documented in the literature, and/or are in sufficiently common use to warrant description here. TEST METHODS National Bureau of Standards Cup Furnace Test The National Bureau of Standards (NBS) test (Levin et al., 1982) employs a cup or crucible furnace, often referred to as the Potts furnace,. named for the investigator who first reported its use in combustion toxicology (Figure 4-~. Heating is considered to be . . . .. ., . ., , .. . . ~ . ~ , ~ ~ ~ largely conductive, with the bottom and lower portions ot the quartz cup constltutmg tne hot zone. Test materials of up to ~ g are introduced into the cup, which has a volume of about ~ I. Procedures for testing materials involve combustion at just below (nonflaming)
32 and just above (flaming) an autoignition temperature. Concentration- response (lethality) relationships for 30-min exposures (14-day postexposure observation) are determined by using rats held in tubular restrainers for head-only exposure. Six rats are used per test. Concentration is controlled by varying the sample weight charged to the cup. There has been · . ~ PORTS FOR ANIMALS W0 W~ a T r ~ ~ it, EXPOSURE CHAMBER FURNACE ENCLOSURE particular concern regarding air flow into the cup, although it does communicate with a volume of 200 1 of air contained in FIGURE d-1. National Bureau of Standards (NBS) apparatus. Source: the exposure chamber. With Lenin et al., 1982. the variable sample sizes used and an ill-defined fuel/air ratio, the system has been criticized for failing to carry out combustion in a well-characterized manner. A further criticism has been that various materials are not tested under the same conditions but often at considerably different temperatures. The method provides a good mode! for nonflaming oxidative decomposition. It is also a good model for simulating the decomposition conditions during a well-ventilated, early- stage fire. It cannot produce the high-temperature, oxygen-vitiated conditions of a fully developed, postflashover fire. Once in fairly common use in a number of laboratories in the United States, the NBS cup furnace test has been well documented, with considerable data available (Levin et al., 1983~. Its use has declined, however, in recent years. University of Pittsburgh Test In the University of Pittsburgh (UPTrr) test method (Alarie and Anderson, 1979; New York State, 1986), the combustion device is a muffle or box furnace, which is often used in an inverted position to provide for a pedestal connected to a mass sensor (Figure 4-2~. With this arrangement, continuous monitoring of sample weight is conducted. Combustion is accomplished by using a linearly rising temperature of 20°C/minup to as high as 1100°C, . . < ~ ~ ~ ~ . ~ . . c. ~ . ~ ~ . ~ ~ ,~c ~ ~ ~ ~ ~ . while 1 1 1/mln ot~ air is pulled through the furnace. ( line smoke atmosphere produced IS diluted with an additional 91/min of cool air prior to exposure of test animals.) The smoke concentration is varied by changing the weight of material charged to the furnace. The fuel/air ratio can therefore vary widely depending on the weight of the sample and its rate of decomposition. Exposures of mice (four in each test) are for 30 min. plus 10 min postexposure, starting when ~ percent of the sample weight has been lost. Bioassays include concentration
33 to response and time to death for lethality. An early version of the test utilized respiratory rate depression as a measure of sensory irritation (Barrow et al., 19761. In the UPI11 test, the concentration term in the LCso refers to the weight of test specimen charged to the furnace. Thus, LC50 values for this test are not directly comparable with those from the NBS method. The test begins in a nonflaming oxidative mode, and at some stage transition to flaming usually occurs. At this time the CO2/CO ratios tend to be low (under 20, usually less than 10), while the temperature is still low (less than 600 °C). This combination of conditions does not, therefore, fit well into the scheme of fire classes shown in Table 4-3. It can represent the rather special situation of a small fire load under restricted ventilation. Although the oxygen concentration is rather low, there are some similarities to the chemical decomposition environment in the early stages of ~ developing fire. The method does not simulate the conditions of a large, fully developed, postflashover fire; however, it could be used to measure the toxic potency of products resulting from the decomposition conditions of developing fires. A problem of small explosions has been encountered with some materials that thermally decompose or burn very rapidly once ignition occurs. This phenomenon, possibly the rapid emission of gases whose volume exceeds that being pulled from the furnace, has not been reported to be sufficiently severe to rupture the apparatus. Another criticism of this method is the 20°Cimin temperature increase, which is quite stow and is not representative of uncontrolled fires. It results in a gradual fractionation of the pyrolysis products, with a disproportionately high percentage of low-temperature decomposition products in the fire effluent that is analyzed and presented to the test animals early in an exposure. Despite these problems, the UPt t-T test Is required for certain construction products in the state of New York (New York State, 1986~. As a result, the test is offered by eight ~ . ~ ~ I FLOW METER PUMP _ F I LTER EXPOSUR E r CHAMB: W ~ ~ 7\ LellGTH: 10cm l.D. 19mm AN I MA L PLETH YSMOGR APH SAMPLING PORT LENGTH: 25 cm DIAM.: 10.5 cm FIGURE d-2. University of Pittsburgh apparatus. DILUTING AIR 1~1 ICE BATH | FURtMACE 1 25x23x34cm r rS WE IG HT SENSOR ~ 'C RECORDER
34 laboratories certified by New York state. The requirement is such that only test data be submitted. The data are not used for classifying products. DIN 53436 Test The DIN 53436 test is characterized by the use of a moving annular tube furnace operating at a constant temperature from 200 ° to 1000°C (German Standards Institution, 1981, Prager, 1988) (Figure 4-3~. The furnace is programmed to travel the length of a quartz tube containing the sample. Decomposition, taking place in an air stream countercurent to the direction of furnace travel, is intended to result in the continuous flow of fire effluents of constant composition. Radiant heat is the major source of energy transfer. This test device, used in six European countries, offers a rather wide range of well-controlled combustion conditions. Ratios of CO2/CO can be widely varied through the choice of the air flow rate; thus, both freely ventilated and ventilation limited fires may be simulated. Specimens may undergo either flaming or nonflaming combustion, depending on the imposed heat flux level and/or the presence of an ignition device. Some difficulty has been experienced in controlling, flaming conditions; however, work using sample segmentation has shown promise with this mode of combustion (EinbroUt et al., 1984~. Such procedures can alter the type of atmosphere produced. An air stream countercurrent to the direction of flame propagation, in contrast generally to the uncontrolled fire situation, is used to prevent uncontrolled preheating DISTRIBUTION CHAMBER DILUTION AIR PRECISION FLOW-METER \ C777777 1 COMBUSTION _ AIR _ \ MOBILE FURNACE QUARtZ CRUCIBLE AS SAMPLE HOLDER FIGURE dog. DIN SS4" apparatus. Souce: Klimisch et al., 1980. PRECISION-FLOW METER ~II i' /1 ~ \ ID G LU Cal In ANIMAL TUBES
35 effects by the combustion products. However, cocurrent conditions have been used (Boudene et al., 1976~. The lack of continuous monitoring of sample weight is compensated by the continuous decomposition process, which enables one to relate the exposed mass, volume, or surface area to the bioassay and/or analytical test data. The DIN 53436 test is clearly useful for toxic potency testing. It is capable of producing the chemical decomposition environment of any of the fire types shown in Table 4-3, and it is the only method that closely simulates the high-temperature, oxygen~vitiated, conditions in a postflashover fire. Care must be taken with simulating the CO2/CO ratios of growing, well-ventilated fires, since such ratios tend to be rather small under standard operating conditions. Self-sustained smoldering is also difficult to simulate in small samples, as is the case with most small-scale methods. The DIN 53436 test has been well documented and widely used. Considerable work demonstrating its validity has been reported (Prayer et al., 1987; Prager, 19881. It is quite flexible in being able to accommodate a range of controlled ventilation conditions, with CO2/CO ratios from below 5 to more than 200 being obtained. The major disadvantages of the system are the difficulties in quantifying the exposure dose of smoke (although it can be estimated) and in controlling both smoldering and flaming combustion. U.S. Radiant Furnace (Modified) Testt The combustion device used In the U.S. radiant furnace (modified) methodology (Grand, 1990) consists of a horizontally mounted, cylindrical quartz combustion cell, 130 mm inside diameter and approximately 320 mm in length (Figure 4-4~. It is connected to an animal exposure chamber through a rectangular stainless steel chimney, which is approximately 30 tic 300 x 300 mm in height. The chimney is divided into three channels, creating a heat I, pump action by inducing COMBUSTION smoke to flow up the center CELt channel, while air from the FIGURE dot. U.~. pant flee (modified). exposure chamber circulates down the outer channels. External to the combustion cell are four tungsten-quartz radiant heat lamps focused onto the plane of the specimen. A platform, accommodating test specimens of 76 x 127 mm and up to 51 mm in thickness, is connected to a load cell located ~ SMOKE SHU l l en ANIMAL EXPOSURE TORTS ~ 1- ~ -- A--- i--- A--- ~ -- A---- ~ ~ i] ~ ~ SPECIMEN PLATFORM / \~ LOAD CE' L This test is also referred to as the NIBS (National Institute of Building Sciences) test.
36 underneath the combustion chamber to monitor continuously the specimen weight. A high- energy spark plug is used as an ignition source. Heat fluxes ranging from 0 to 75 kW/m2 have been used. Ratios of CO2/CO have ranged widely, usually being well in excess of 20. This combustion device, therefore, appears to produce good simulation of an early developing, flaming fire. However, insufficient data are presently available to establish the exact characteristics. Developmental work still in progress, should permit better characterization of the properties of this combustion device. Unique to this method is a proposal that the test method be used to develop an linden of potential toxic hazard. for screening of test materials by procedurally combining time to ignition, rate of decomposition, and toxic potency. Alternatively, it has been proposed that these separate parameters be extracted from the test for use in toxic hazard engineering calculations. Such a proposed test is currently undergoing the ASTM consensus process. In this method, a test specimen is subjected to ignition while exposed to radiant heat, with the smoke produced being collected for 30 min within a 200-! chamber communicating with the combustion assembly. Concentrations of the major gaseous toxicants are monitored over the 30-min period, with Ct products for each being determined from integration of the areas under the respective concentration-time plots. The Ct product data, along with the mass loss of the test specimen during the test, are then used in calculations to estimate the 30-min LC50 of the test specimen. The estimated LC50 is then validated by use of exposure of rats. Validation assures that the monitored toxicants account for the observed toxic effects and that there is no unusual toxicity. Although there Is still a lack of documentation and availability of data from the test method, several laboratories have the apparatus. Considerable work using an earlier version of the radiant furnace test is described by Alexeeff and Packham' 1984. ROI`E OF ANALYTICAL CHEMISTRY Babrauskas et al., 1987, stated that early attempts to explain the toxicity of smoke by analysis of the combustion products were fraught with problems: (~) the number of identified species was overwhelming (over 400 compounds have been reported from the decomposition of wood), (2) the toxic potency of every one of these compounds is not known, and (3) the approach becomes difficult in particular in view of the multiplicity of combinations of these gases. On the other hand it was once perceived that exposure of animal models was the only reliable methodology. As an example of support for this contention' relatively sophisticated instrumental methods failed to detect a highly toxic bicyclophosphate material produced from the nonflaming degradation of a fire retarded polyurethane foam (Voorhees et al., 1975~. The bicyclophosphate compound was identified only after grand mat seizures were observed in rats exposed to the combustion atmosphere (Petajan et al., 1975~. This example, though rare, serves to illustrate that care must be taken in relying solely on analytical results. Experience has since suggested that the toxicity of most fire effluent atmospheres can be explained largely on the basis of only a few major components. Thus, along with the advent of engineering calculations for hazard and risk assessment, analytical methods in combustion toxicology are extremely important in minimizing animal experimentation. In fact, chemical analysis of fire effluent atmospheres is deemed sufficiently important that it
37 is the subject of an ASTM Standard Guide (ASTM, 1988) and a forthcoming ISO Technical Report (ISO, 1989~. Since such guidance is available, this report will not go into detail on the methods. The most common fire effluent constituents, CO and CO2, are routinely measured on a continuous basis by nondispersive infrared spectroscopy. Oxygen is continuously measured by use of a paramagnetic detector. HCN and hydrogen chloride (MCI) are normally sampled with syringes, together with some type of noncontinuous analysis. Commonly used are gas chromatography, ion-specific electrode titrimetry, and calorimetric procedures for HCN. Titration and ion chromatography are employed for HCI. However, even HCN and HC! have been successfully analyzed continuously in some systems by use of continuous automatic titration methocIs (Grand, 1988~. Fourier transform infrared methods show considerable promise for continuously analyzing several fire gases simultaneously (Kallonen, 1990~. Gas chromatography/mass spectrometry may be used for analyses of fire effluent components that do not lend themselves to direct instrumental methods. CONCLUSIONS Toxicity data for materials and/or products are obtained from laboratory tests that, upon utilizing a combination of analytical data along with the exposure of rodents, yield two useful types of information: l) the principal typets) of intoxication, i.e., asphyxiation, sensory or pulmonary irritation, etc. and 2) the LC50 of the smoke in units of gums over a given exposure period. Toxic potency data obtained from laboratory tests are, however, subject to the following limitation and/or considerations. 1. No single laboratory combustion device Is appropriate for all materials and products under the condition of all fire types and stages. Therefore, there can be no universal Smoke toxicity test.. The laboratory combustion device used in a test should be chosen and operated to approximate as closely as possible the conditions of the type of fire being examined. (For example, laboratory scale combustion furnaces may, under certain conditions, produce less CO per unit mass of sample burned than would occur in a real fire. Thus, laboratory LC~0 values may need to be Adjusted for use in hazard calculations.) 2. All LC ;O values have an associated level of statistical confidence. Furthermore, interiaboratory comparisons suggest LC~0 determinations can vary by a factor of about 2.5. 3. Although calculated from data based on exposure of rodents, LCS0 values can be extrapolated to human exposure with reasonable confidence for asphyxiants and for pulmonary irritation. Sensory irritation Is not addressed with current laboratory smoke toxicity tests. Also, its relevance to incapacitation of humans has not been demonstrated. (Hazard assessments currently tend to set threshold tenability levels for acid gases and other combustion products known to have irritant properties.) 4. Currently employed laboratory smoke toxicity tests do not directly measure incapacitating effects of smoke inhalation. Incapacitation must be inferred from LC~0 values.
38 REFERENCES Alarie, Y. C. and R. C. Anderson. 1979. "Toxicological and Acute Lethal Hazard Evaluation of Thermal Decomposition Products of Synthetic and Natural Polymers," Toxicology and Aoolied Pharmacology, Vol. 51, pp. 341-362. Alexeeff, G. V. and S. C. Packham. 1984. "Evaluation of Smoke Toxicity Using Concentration-Time Products," Journal of Fire~ciences, Vol. 2,pp.362-379. AST M. 1982. Standard Terminology Relating to Fire Standards," ASTM E 176-82, American Society for Testing and Materials, Philadelphia, PA. ASTM. 1984. "Standard Test Method for Estimating Sensory Irritancy of Airborne Chemicals," ASTM E 981, American Society for Testing and Materials' Philadelphia, PA. ASTM. 1988. "Standard Guide for Measurement of Gases Present or Generated During Fires," ASTM E 800, American Society for Testing and Materials, Philadelphia, PA. Baker, B. B., Ir. and M. A. Kaiser. 1991. "Understanding What Happens in a Fire," Analytical Chemistry, Vol. 63, pp. 79-83. Barrow, C. S., Y. C. Alarie, and M. F. Stock. 1976. Sensory Irritation Evoked by the Thermal Decomposition Products of Plasticized Poly-(viny} chloride)," Fire and Materials, Vol. 1, pp. 147-153. Babrauskas, V., B. C. Levin, and R. G. Gann. 1987. HA New Approach to Fire Toxicity Data for Hazard Evaluation, Fire Journal, March/April, p. 22. Boudene, C., I. M. Jouany, and R. Truhout. 1976. Protective Effect of Water Against Toxicity and Pyrolysis and Combustion Products of Wood and Polyvinyl, 2nd Symposium on PVC, Lyon, France, July; I. Makromol. Sci. Chem., A 11/~. EinbroUt, H. I., I. Hupfeld, F. H. Prager, and H. Sand. 1984. "The Suitability of the DIN 53436 Test Apparatus for the Simulation of a Fire Risk Situation with Flaming Combustion," Journal of Fire Sciences, Vol. 2, pp. 427-438. Einhorn, I. N., D. A. Chatfield, R. W. Michelson, and ]. H. Futrell. 1974. "Pyrolysis and Combustion Products of Nomex, Durette, and Tediar Polymers, Final Report NASA Contract No. NAS2-8244, University of Utah, Salt Lake City. German Standards Institution. 1981. ~Erzeugung Thermischer Zersetzungsprodukte von Werkstoffen unter Luftzufuhr und ihre toxikologische Prufung. Part 1, Zersetzungsgerat und Bestimmung der Versuchstemperatur. Part 2, Verfahren zur thermischen Zersetzung, Part 3, Entwurf-Verfahren zur inhalation-toxikologischen Untersuchung thermischer Zersetzungsprodukte,~ DIN 53436. Grand, A. F. 1988. "Continuous Monitoring of Hydrogen Chloride in Combustion Atmospheres and in Air," Journal of Fire Sciences, Vol. 6, pp. 61-79. Grand, A. F. 1990. "Evaluation of the Corrosivity of Smoke from Fire Retarded Products," Journal of Fire Sciences, in press.
39 Hartzell, G. E. 1989. Prediction of the Toxic Effects of Fire Effluents, Journal of Fire Sciences, Vol. 7, pp. 179-193. Harwood, B. and ]. R. Hall, Jr. 1989. "What Kills in Fires: Smoke Inhalation or Burns?, Fire Journal, May/]une, pp. 29-34. ISO. 1989. "Toxicity Testing of Fire Effluents Part 1: General, ISO/TR 9122~1, International Organization for Standardization, Geneva, Switzerland. Kallonen, R. 1990. "Smoke Gas Analysis by FIlR Method. Preliminary Investigation, Journal Refire Sciences, Vol. 8, pp. 343-360. Kaplan, H. L., A. F. Grand, and G. E. Hartzell. 1983. Combustion Toxicology: PrinciDles and Test Methods. Technomic Publishing Company, Inc., Lancaster, PA. Klimisch, H. J., I. E. Doe, G. E. Hartzell, S. C. Packham, I. Pauluhn, and D. A. Purser. 1987. Bioassay Procedures for Fire Effluents: Basic Principles, Criteria and Methodology," Journal of Fire Sciences, Vol. 5, pp. 71-145. Klimisch, H. I., H. W. M. HolIander, and I. Thyssen. 1980. "Generation of Constant Concentrations of Thermal Decomposition Products in Inhalation Chambers. A Comparative Study with a Method According to DIN 53436. I. Measurement of Carbon Monoxide and Carbon Dioxide in Inhalation Chambers. Journal of Combustion Toxicology, Vol. 7, p. 243. Levin, B. C., A. I. Powell, M. M. Birky, M. Paabo, A. Stolte, and D. Malek. 1982. Further Development of a Test Method for the Assessment of the Acute Inhalation Toxicity of Combustion Products,. NBSIR 82-2532, U.S. National Bureau of Standards, Washington, DC. Levin, B. C., M. Paabo, and M. M. Birky. 1983. "An InterIaboratory Evaluation of the National Bureau of Standards Test Method for Assessing the Acute Inhalation Toxicity of Combustion Products,. NBSIR 83-267S, National Bureau of Standards, Gaithersburg' MD. MacFarIand, H. N. 1976. Respiratory Toxicology, EssaYs in Toxicology, Vol. 7, W. _, , , Hayes, Jr., Ed., Academic Press, New York, Chapter 5, pp. 121-154. New York State Uniform Fire Prevention and Building Code. 1986. "Combustion Toxicity Testing, Office of Fire Prevention and Control, Department of State, New York State, Albany, NY, Article 15, Part 1 120. Petajan, ]. H., K. ]. Voorhees, S. C. Packham, R. C. Baldwin, I. N. Einhorn, M. L. Grunnet, B. G. Dinger, and M. M. Birky. 1975. Extreme Toxicity for Combustion Products of a Fire-Retarded Polyurethane Foam," Science, Vol. 187, pp. 742-744. Potts, W. I. and T. S. Lederer. 1978. Smoke Limitation in the Use of the Sensory Irritation Method as an End-Point in Measurement of Smoke Toxicity, Journal of Combustion Toxico1Qgv, Vol. 5, pp. lS2-195. Prager, F. H. 1988. Assessment of Fire Mode! DIN 53436,~ Journal of Fire Science,, Vol. 6, pp. 3-24.
40 Prager, F. H., H. ]. Einbrobt, I. Hupfeld, B. Muller, and H. SancI. 1987. Risk Oriented Evaluation of Fire Gas Toxicity Based on Laboratory Scale Experiments~The DIN 53436 Methodic Journal of Fire Sciences, Vol. 5, pp. 308-325. Purser, D. A. 1988. Toxicity Assessment of Combustion Products and Modeling of Toxic and Thermal Hazards in Fired SFPE Handbook of Fire Protection Engineering, National Fire Protection Association, Quincy, MA, Section 1, pp. 200-245. Terrill, I. B., R. R. Montgomery, and C. F. Reinhardt. 1978. Toxic Gases from Fires," Science, Vol.200,pp.1343-1347. Voorhees, K. I., I. N. Einhorn, F. D. H'1eman, and L. H. Wojcik. 1975. The Identification of a Highly Toxic Bicyclophosphate in the Combustion Products of a Fire-Retarded Urethane Foam. Journal of Polvmer Science, PolYmer Letters Edition' Vol. 13, p. 293. -