Devising a Screening Test ]'or Toxic Fire Gases JAMES B. T E R R I L L , PH.D.; R U T H R. MONTGOMERY, M.S.; and CHARLES F. R E I N H A R D T , M.D. Haskell Laboratory for Toxicology and Industrial Medicine E. I. du Pont de Nernours and Company The basic fire triangle of oxygen, fuel, and heat becomes much more complicated when it is used as an experimental model. A discussion of major test variables introduces a comparison of the tube furnace, torch, heated cup (crucible), and radiant heater as fire models for laboratory tests. INTRODUCTION A
F I F T Y percent reduction of life and property losses due to fire within the next generation is the goal of the National Commission on Fire Prevention and Control. ~ Major efforts are underway in both government and industry to define specific fire hazards, educate the public on firesafety, develop improved products, specify proper use of products, improve and expand the use of early warning and fire suppression systems, and develop meaningful methods for measuring the fire performance of materials. Developing a more sophisticated, yet simple, laboratory screening test to characterize building cons~uction or furnishing materials in terms of the inhalation toxicity of their combustion/pyrolysis products is an important objective. Such a test would be a useful addition to other types of fire performance tests, such as ignition, rate and amount of heat release, and rate and amount of smoke production. We are taking this opportunity to review the basic parameters of a fire, to discuss pertinent toxicology test design, and to present a simple approach to a test model. To gain insight into the fire phenomenon as a whole, we chose the concept of the fire triangle, shown in Figure 1, as a discussion model. Fire occurs with varying amounts of oxygen, fuel, and heat, but will not occur when any one of these three ingredients is deficient. Varying amounts of NOTI$:Based on a paper presented by Dr. Terrill as an invited speaker in Symposium on the Flammability and Combustion of Non-Metallic Materials, 172nd National Meeting of the American Chemical Society, September 1, 1976, San Francisco.
95
96
Fire Technology
FUEL A
Figure 1. Traditional fire triangle,
B
C
OXYGEN
HEAT
these three ingredients give different fires and different combustion products. Figure 2 shows that our basic fire triangle becomes much more complicated when it is used as an experimental model. Various types, forms, and orientations of fuel become significant. Likewise, the transfer mechanisms of heat from the source and of oxygen from the atmosphere to the reacting fuel are primary factors influencing the chemistry of the resultant combustion/pyrolysis products. Furthermore, the presence or absence of a flame (combustion or smoldering) has a tremendous influence on the chemistry of the resultant products. Due to this myriad of factors, the problem of characterizing a "standard fire" is very complex. Any screening test can, at best, hope to encompass only a fixed set of conditions. Therefore, whenever possible, our test(s) should be modeled according to more common types of fires involving the product. If test conditions can be type,form ,orientation in test
FUEL: A
~
nl QXYGEN:
level,static vs forced ventitotion
t
Testsrun of different
points
\r, (contact,radiant, T" ( co vective,orcombinotions HEA • t of two or eli three types
GOAL OF SCREENING TEST: Survey various points within Z~ Figure 2. Fire triangle as a test model.
Toxic Gases
97
easily modified for operation at a limited number of fuel, oxygen, and heat ratios, the screening test will be more reproducible without significant increase in cost or complexity. It is now recognized, although unfortunately not yet universally accepted, that small-scale laboratory tests using externally applied heat will not define the life hazard in most real fire situations. However, such tests provide a means of comparing or ranking products and, therefore, provide a logical first step in any screening of comparable materials. A particular advantage is that materials yielding unique, highly toxic products can be detected or identified. We have (facetiously) hypothesized the ideal small-scale test as follows: • Simulates significant real conditions; • Provides fast answers; • Provides accurate answers; • Uses inexpensive, readily available equipment; • Requires nonskilled operators; • Is inexpensive to run; • Gives animal results but does not require animals, i.e., can be run anywhere; • Uses material in any form, i.e., production material results from a single batch; • Requires less than 10 grams of material; and • Is reproducible (inter- and intralaboratory). We are frankly pessimistic about developing a practical model t h a t incorporates all ten attributes. Thus, reduced to more conventional reasoning, we will discuss some of the major variables inherent in a small-scale, combustion product toxicity test. (1) A n i m a l Selection. Species, sex, number, weight, and strain are basic variables in any toxicological experiment. (2) A n i m a l Exposure Regimen. • Duration of exposure. This is one of the most critical test variables, since heat stands out as the unavoidable limiting factor in real fires, and the interval until such heat is generated becomes an absolute maximum for exposure. In small-scale tests, an arbitrary, generalized, exposure time is selected. • Chamber temperature. Animal environment must be below that temperature which w~ll produce lethal or appreciable contributory stress from heat. • Exercise. Fires may prompt exposed individuals to sudden vigorous physical activity that can add significant stress. (Persons with coronary disease are particularly vulnerable.) Increase in respiration rate from exercise can increase the rate at which gaseous toxicants enter the blood. • Oxygen concentration. If the chamber atmosphere is overwhelmed by
98
Fire Technology
fire "effluents," oxygen concentration will dip suddenly and critically in experimental tests. (3) Animal Observations. Clinical data, neurological examination to assess incapacitation, blood tests to monitor absorption of carbon monoxide and cyanide (CN), and pathological examination may all be desirable, depending on the scope of the test. (4) Selection and Sensitivity of Analytical Methods. These are governed by the scope of the tests as well as the methods available. Analytical data should be correlated with biological data and not used independently in order to "determine" composite toxicity of a polymer from thermal degradation. (5) Test Material Configuration. Shape, form, and density can effect heat and oxygen supply. (6) Mode of Sample Decomposition. Combustion/pyrolysis conditions are perhaps the most complicated test variables, since they specify the particular mechanics selected to generate a "fire toxicity" atmosphere and will directly influence temperature and fire gas atmosphere. (7) System Configurations. The arrangement and use of the test apparatus other than that used for the actual combustion/pyrolysis m a y comprise: (a) length of connecting tubes; (b) lining of chamber and connecting tubes; and (c) venting, recycling, heating, dilution and/or flow rate of off gases from the sample. (8) Likely End Use Conditions. Whether use is minimal or extensive, industrial or consumer, in furnishings or concealed in many building structures, can, in many cases, govern the type of testing warranted. As a particular example, compare the use of textile fibers in draperies to their use in carpets. Recent reviewers ~-4 provide a comprehensive introduction to experimentation involving animal exposures and/or analytical studies of smoke. In some experimental studies, the terms pyrolysis and combustion have often been interchanged. Basically, pyrolysis specifies thermal degradation, and combustion specifies thermal degradation in at least enough oxygen to support a flame. 4 Pyrolysis testing may either be distinguished from, or include, combustion testing. In our own laboratory, we have been primarily concerned with manufacturing processes that involve polymers heated for prolonged periods (one to four hours) at given temperatures. In such cases, a tube htrnace is a convenient and representative pyrolysis model. However, in other areas, such as accidental fire situations, a torch (which is really an elaborate Bunsen burner), a heated cup (or crucible), or a radiant heater similar to that developed by the National Bureau of Standards (NBS) may be a more appropriate model than the tube furnace. The tube furnace is very limited as a fire model for many reasons. It is impossible to operate under "flaming" conditions. Secondly, very rapid pyrolysis of the sample will exceed the capacity of the system to deliver adequate oxygen, and abnormally high levels of carbon monoxide may be produced. Thirdly, some samples may react explosively inside a red hot
Toxic Gases
99
furnace. Fourthly, the tube furnace does not appear to simulate a simple aerodynamic modeP for combustion processes. Prior to an exothermic reaction, the maximum temperature in the tube furnace is at the tube wall; whereas in a fire, maximum temperature is in a reaction zone near the fuel surface. The "torch test," which is modeled on the "classic picture" of fire, was described in an earlier report. 4 In this work, the clinical observations on exposed rats were correlated with measured atmospheric levels of common toxicants (carbon monoxide, hydrogen cyanide, halogen acid, and sulfur dioxide). The chief disadvantages of this somewhat qualitative test were the sizable (~-~2 ×) scatter of data in replicate runs and the use of a fan to circulate the combustion gases between two 2-ft ~ (2 × 0.06 m 3) rectangular chambers, shown i n Figure 3. Impaction of oily and tacky smoke particulate by the fan will increase particle size, which, in turn, decreases both the atmospheric concentration of particulate and the respirable nature of the particulate.
Circulating Figure 3. Sketch of a two-chamber exposure apparatus. Volume is 4 #3. Heater on right may be replaced by a torch.
~ 1 . x 1 . x 2 , ~
RatExposure
Pyrolysis Chamber
Chamber
RECENT TEST
DEVELOPMENTS METHODOLOGY
Fan
IN
A procedure of improved precision for sample pyrolysis is the electrically heated cup (see Figure 4). Tests were run using the same 4-ft 3twin chamber as previously used in the torch test. Samples (3 to 10 g) can be either cut into pieces or wound into tight rolls and pyrolyzed at a constant cup temperature. Incorporated into this more recent experimental protocol is a measurement of carboxyhemoglobin (COHb) in all dead rats immediately after exposure. If no rats die, two are sacrificed following exposure for this determination. The COHb values show the body burden due to varying concentrations of carbon monoxide. If COHb is >_ 70 percent in the dead rats (without the aggravation of heat stress), the contribution of other asphyxiants to death is minor. Secondly, we select experimental protocols (varying sample size, pyrolysis temperature, and exposure time) so that only part of the group survives the exposure. These survivors are held for a
100
Fire Technology Test Sample
Thermocouple To Powerstat And Heater Controller Heater Base, Connect To Variac
Figure 4. Cutaway drawing of a pyrolysis cup as used at Haskell Laboratory. (Designed by C. Rapisarda, Carothers Laboratory, Du Pont.)
Container is - Stainless Breaker ~4x6"
two-week observation period. Deaths during the pos~exposure recovery period can be produced by various factors, such as pulmonary edema. The results of four films pyrolyzed at 550 ° and 650 ° C in the heated cup show that the atmospheric levels of CO during the exposure are slightly different for the various films (Figure 5). Since lethal levels of COHb (greater than 75 percent) were found in exposed rats, and no rats died after surviving the initial exposure, significant amounts of toxicants other than F,I~ o
P 6000
,~/F~,mc
II II
5ooo ~
/ / / 5000
/f
L _
4000
/
5000 -
.,~.....~ F,tmA
/
j~/~
~"
/ " . y - " ~, ~,,~o
/
/
B~
I;
.o e,lme
II
4000
• /-"
I/~ ~;
:'
g i E --:
3000
=C~{.' ' ' ' 0
--
Film B
8
..o." /
2000
d
3000
2000-
I
."
•/
i
x :/
~ ':
..o /
.,'
,ooo ,~ y / / .~
~
1
10
20
~
I
I
l
30
40
50
60
Pyrolysis, min,
10
o;
.oc
20
30
40
50
60
Pyrolysis, rnin,
Figure 5. Atmospheric levels of carbon monoxide in pyrolysis of four 5-g film samples at 550°C and 650°C using the apparatus in Figure 3. Film A is the base polymer and Films B, C, and D are the base polymer plus different "flame retardants" at comparable performance loading levels.
101
Toxic Gases
carbon monoxide were probably not present. In a second study of polymers containing carbon, oxygen, hydrogen, nitrogen and a "flame retardant," decomposition produced levels of carbon monoxide and hydrogen cyanide that differed for the two polymers (Figure 6). In these cases, the COHb level found in animals dying during exposure (69-70 percent) appeared lowered, suggesting a contribution by hydrogen cyanide. The postexposure pulmonary distress suggested the presence of pulmonary irritants (perhaps tacky particulate). In general, precision of results for runs with the heated cup ( ~ __ 25 percent) is greater than those for the "torch test" ( ~ _+ 50 percent). However, since the cup can be viewed as an open-ended tube furnace, some of the same drawbacle~q apply - - limited aerodynnmic model for combustion, potential flashing of large samples in red hot cups, and only contact heating (no radiation). Furthermore, a flaming test is difficult to operate, since most materials burn poorly in horizontal configurations and the cup hinders uniform transport of oxygen to the flame. Our most recent research, still developmental, uses a single 10-ft 3 (0.28-m 3) stainless steel chamber 5 ft by 1 ft by 2 ft (1.5 m by 0.3 m by 0.6 m) with a radiant heater of the type used in the NBS smoke chamber, s A 3-in. by 3-in. (7.6-cm by 7.6-cm) sample up to 1 ~ in. (3.2 cm) thick m a y be pyrolyzed at fluxes up to 7w cm -2. The smoke rises between the sample and the heater and flows along the top of the chamber to the opposite end, which contains exposed rats. The rats are restrained from wandering by an expanded metal screen. View ports constructed of Lucite ® or glass permit continuous observation. Mixing of gaseous products appears rapid;
200 180 160
4000
X
140 3000
1K ~ [ f ~ ' ' ' ~ t t ' ' ' ' ' l K ~ l K
a..v.z
~
120
,,o,,~, 2,
~
1oo
gE 0--~
R~ e
2000
l I muzy~, ~1
g ~--
80
,o
1000
2O t 10
I 20
1 30
I 40
Exposure,rain.
I 50
t 60
0
10
I 29
I 30
I 40
t 50
I 60
Exposure, rain,
Figure 6. Atmospheric levels of carbon monoxide and hydrogen cyanide for polymers based on C, H, O, and N . Polymer 1 has no "flame retardant." Polymer 2 is "flame retarded." R u n s were in the 4-# ~ chamber with the heated cup using 20 g at 650°C cup temperature.
102
Fire T e c h n o l o g y
80 percent equilibrium for test gases is reached in less than 2 rain. Ten cubic feet appears to be an optimum size, because in smaller sized exposure chambers the heat buildup from the radiant heater is too intense. In larger chambers, toxic gas levels are too low to produce adverse effects. Performance of this system has been demonstrated by testing a wide variety of materials - - low and high density plastics and natural products. In tests with white pine, the rate of carbon monoxide production appears proportional to flux (Figure 7). In addition, post-exposure pulmonary distress is often more pronounced in rats exposed in runs at lower fluxes. In these studies, carbon monoxide data for runs at 2.5 or 5.0 w cm -~ indicate the precision of replicates to be about +_ 20 percent. A major disadvantage of the vertical sample holder is that thermoplastic samples can flow easily out the open face. Tilting of the entire assembly 45 ° backwards helps reduce this problem. Also, Fiberglas ® mesh may be added to impede the dripping. Tests with red oak (2.5 or 5.0 w cm -~, no Fiberglas ®) in both the conventional vertical position and the 45 ° recline give similar carbon monoxide levels and rat mortality data. However, arranging the sample in a horizontal position with the heater parallel and directly above has the disadvantage that combustion products must rise towards the hot heater. Contact of these gases with the heater elements would likely change their chemistry. A flaming test condition can be incorporated into this system b y use of an air/propane torch, match, or electrical resistance wire igniter. Tests at the same radiant flux with flaming and nonflaming conditions using red oak indicate carbon monoxide production is markedly diminished when a
12000
u
*,~,
9coo
Figure 7. Atmospheric levels of carbon monoxide from white pine pyrolyzed at different fluxes. At. 2.5 W cm -2, sample was purposely ignited and allowed to flame for 15 rain while receiving 2.5 W cm -~ radiation. After the flame self-extinguished, radiation continued at 2.5 W cm -2. Runs were in 10-# 8 chamber with radiant heat as developed by the National Bureau of Standards.
80oo
~ ~o I
"
'~"
3OO0
5
I0
15
20
25
~
35
~0
4~
Toxic Gases
103
flame is present. Also, oxygen depletion is more rapid with a flame present. If the torch's air/fuel mixture is not properly adjusted, the torch could produce carbon monoxide directly. CONCLUSION This developmental work continues because in testing new products or product applications, the question, "Have we picked the best test model?" must constantly be raised and satisfied. A recent study by Cornish, Barth, and Hahn 7 of two pyrolysis toxicity test designs showed that the resulting rank orders of ten common materials varied significantly and depended strongly on test procedure. A yet unaccepted test end point is "incapacitation" 8.9 as opposed to the much simpler test end point of death. Also, the relationship between various existing small-scale fire studies needs to be established. Overall, we conclude that two additional criteria - - (a) design and use of materials, (b) test procedure - - combine with the basic fire triangle of heat, oxygen and fuel to form the fire pyramid, shown in Figure 8. Evaluation of a material's performance under a series of appropriate test conditions and procedures should be more beneficial than using a single test. We believe this presentation may further the resliT.ation that pyrolysis toxicity test results depend strongly on test procedure. Kind and Amount of Fuel A
Figure 8. Fire toxicity pyramid.
•
~ /// B
Pyrolysis
. . . . .
/
Ventilation (Oxygen)
(Exclusive of Heat)
E Design and Use of Material
REFERENCES ~National Commission on Fire Prevention and Control, America Burning; The Report of the National Commission on Fire Prevention and Control (U.S. Government Printing Office, Washington, DC, 1973), p. 8. Birky, M. M., "Review of Smoke and Toxic Gas Hazards in Fire Environment," International Symposium, Fire Safety of Combustible Materials, University of Edinburgh (October 1975). Birky, Merritt M., Philosophy of Testing for Assessment of Toxicological Effects of Fire Exposure," Journal of Fire and Flammability/Combustion Toxicology, Vol. 3, No. 1 (February 1976), p. 5.
104
Fire Technology
Montgomery, Ruth R., Reinhardt, Charles F., and Terrill, James B., "Comments on Fire Toxicity," Journal of Fire and Flammability~Combustion Toxicology, Vol. 2, No. 3 (August 1975), p. 179. Emmons, Howard W., "Fire and Fire Protection," Scientific American, Vol. 231, No. 1 (July 1974), p. 21. 6 Chien, W. P., and Seader, J., "Smoke Development of Different Energy Flux Levels in an NBS Smoke Density Chamber," Fire Technology, Vol. 10, No. 3 (August 1974), p. 187. , . 7 Cornish, Herbert H., Barth, Mary L., and Hahn, Kolman L., 'Comparative Toxicology of Plastics During Thermodecomposition," International Symposium on Toxicity and Physiology of Combustion Products, University of Utah, Salt Lake City (March 1976). s Spurgeon, Joe C., "A Preliminary Comparison of Laboratory Methods for Assigning a Relative Toxicity Ranking to Aircraft Interior Materials," U.S. Department of Transportation, FAA-RD-75-37 (October 1975). Alarie, Yves C., Wilson, E., Civic, T., Magill, Joseph H., Funt, John M., Barrow, C., and Frohlinger, J., "Sensory Irritation Evolved by Polyurethane Decomposition Products," Journal of Fire and Flammability~Combustion Toxicology, Vol. 2, No. 2 (May 1975), p. 139.