IMDoin 090 ^ 00 Toxicity of the Pyrolysis and Combustion Products of Poly ( Vinyl Chlorides ) : A Literature Assessment

1

Wfa^XO] 14Sft_8ppiCfi*t^iso*Te^t v* ' "t 3«44TAfl3it1j#s(2laS'^H3 PROPERTIES OF PVC 5 3. THERMAL DECOMPOSITION OF PVC 6 4.              INTRODUCTION Poly(vinyl chlorides) (PVC) constitute a major class of synthetic plastics and are used in a wide variety of industrial and household products.

TOXICITY OF PVC PYROLYSIS AND COMBUSTION PRODUCTS
Approximately seven billion pounds of PVC resins were produced in the United States in 1984 [1]^. More than half of this amount, which corresponds to an average of about thirty pounds per person, is used in building products, including pipe and fittings, flooring, siding, and a wide variety of other construction products [2]. Much of this goes into residential construction.
Other major uses include packaging, electrical insulation for wire and cable, interior furnishings, and a host of miscellaneous household products. Because of its ubiquitous presence in the built environment, PVC is a possible source of fuel in unwanted fires. In a number of fires, injuries to firemen and building occupants have been attributed to its thermal decomposition or combustion products [3,4,5]. This, together with PVC's wide distribution, has led to numerous studies of the nature of the products, their occurrence in fires, and their physiological effects on living organisms.
A number of surveys of this voluminous literature have been performed.
One of the early surveys was performed at the Underwriters' Laboratories in 1963 [6]. An extensive survey was made by Tewarson at Factory Mutual in Numbers in brackets refer to the literature references listed at the end of this report.
In a supplement prepared in 1979, he cites an additional 56 references [10]. The literature has increased in volume at an accelerated pace since that time. Barrow, in his Ph.D thesis, provides an excellent review of the toxicity of the thermal decomposition products of PVC, with emphasis on experimental methods and physiological effects [11]. Boettner and Ball reviewed work on the composition of the volatile combustion products of PVC in 1981 [12]. Kent and van der Voort described the behavior of PVC in fires in 1981 [13]. Recently, Hinderer has reviewed the combustion product toxicity of PVC [14], Despite this volume of publications, it is difficult to obtain a clear picture of the role of PVC in fires. Reported data are difficult to compare and sometimes contradictory. A number of reasons may be cited for this difficulty: 1.
Pure PVC homopolymer (PVC resin) is almost never encountered outside the research laboratory. Most useful products are made from compositions compounded with a variety of additives designed to improve their properties. These additives include polymerization catalysts, plasticizers, lubricants, heat and light stabilizers, fire and smoke retardants, fillers and reinforcing agents [15]. The amount of plasticizer may vary from 10 parts per hundred of resin (phr) in semi-rigid PVC*s to 100 phr in very flexible PVC's. Any of these additives can affect the fire properties of the material and the composition of its combustion products. It is rare that two investigators use what can be identified as the same formulation. Often the samples are not adequately identified and, in some cases, the formulation may not be known to the investigators.
Methods of thermal decomposition or combustion of the samples vary widely from one study to another, particularly with respect to the thermal and atmospheric environment. These differences will affect the rate of product formation as well as the composition and concentrations of the combustion products. 3. In those cases where animals have been used to assess the physiological effects of the thermal degradation products, a variety of species, which may differ in their response to this exposure, have been employed. Even when the same species and strain is used in different studies, variations in the conditions of exposure can lead to differences in response.

4.
A plethora of methods for measuring responses to exposure have been employed. Different end points have been used, different parameters to characterize the experimental system are reported, and the results may be given in different and frequently non-comparable units.
Despite these difficulties, a fairly coherent picture of the fire behavior of PVC emerges from this mass of material. Minor inconsistencies may be attributed to one or more of the factors listed above. The present review concentrates on the more recent work (since 1969) and endeavors to be interpretive rather than comprehensive.

PROPERTIES OF PVC
Poly( vinyl chlorides) are high molecular weight poljnners formed by the free radical polymerization of vinyl chloride monomers.
The homopolymer contains 56.7% chlorine by weight. The structure is primarily the linear head-to-tail configuration but various irregularities can occur in the chain structure. The ends of the polymer chain may be occupied by a number of different catalyst fragments and chain transfer agents. Typical number average molecular weights lie in the range of 50,000 to 120,000.
The PVC homopolymer (PVC resin) is a hard, tough material that has little practical application. It has poor stability when exposed to heat and light and must be compounded with stabilizers to make it useful. These stabilizers are often compounds of heavy metals, e.g., barium, tin, cadmium, and lead.
Rigid PVC, with a minimum of additives, may contain on the order of 90% PVC homopolymer. It is used for pipes, ducts, building panels, siding, floor coverings, and other building applications.
The PVC resin can be compounded with a variety of plasticizers, usually high molecular weight organic esters, to give products the improved flexibility needed for use in films, tubing, wiring insulation, and the like.
Di-2-ethylhexyl phthalate (di-octyl phthalate, DOP) is perhaps the most widely used plasticizer, but a great variety of other compounds are also used to impart special properties. Phosphate esters, such as tri-cresyl phosphate, are sometimes used as plasticizers to avoid the increase in flammability caused by phthalate and other purely organic plasticizers [16].
Rigid PVC has an oxygen index of greater than 40 [16]. The addition of large amounts of plasticizer may lower this value to about 20 or 25. Fire retardants containing bromine, antimony, or phosphorus compounds may be added to counteract this effect. PVC compounds in general are among the more fire resistant common organic pol 3 miers, natural or synthetic.
In addition to PVC composites, a number of other chlorine containing polymers can be expected to show some similarity to PVC in their thermal decomposition properties and the toxicological properties of their combustion products. These materials will include vinyl chloride copolymers, poly-(vlnylidene chloride), polychloroprene (neoprene), chlorinated polyethylene and chlorinated PVC. 3.

THERMAL DECOMPOSITION OF PVC
It is to be expected that the thermal decomposition or combustion properties and the composition of the products from commercial items identified as "PVC" will vary widely depending on the nature and amount of additives contained in the formulation. Nevertheless, a certain consistency of behavior can be discerned if one focuses on the properties of the PVC homopolymer and considers the further effects of additives on this basic behavior.
PVC horaopolymer has relatively poor thermal stability, beginning to lose hydrogen chloride (HCi) slowly in the neighborhood of 100°C. The reaction is believed to start at imperfections in the polymer chain. Tertiary chlorine atoms, allylic chlorides and various oxygenated structures have been suggested [17].
The reaction appears to be weakly autocatalytic. Thermal stabilizers react preferentially with the weak points in the polymer chain, preventing development of the autocatalytic process. Through the use of stabilizers, useful service temperatures can be achieved.
The pyrolysis and oxidation of PVC have been studied by a number of investigators. Woolley [18] and O'Mara [19] have given the most detailed accounts and have summarized earlier work. The reported results depend on the particular formulation used and close agreement is not to be expected between investigators. However, a fairly consistent pattern of behavior emerges.
Stabilized PVC resins begin to lose HCJl slowly when heated to about 200°C.
The rate of HCi evolution increases with temperature, becoming fairly rapid at 230°C.
By 300°C, the dehydrochlorination is very rapid and results in an almost quantitative evolution of HCA. Woolley reports activation energies for the dehydrochlorination reaction of 174 kJ/mol and 151 kJ/mol in nitrogen and air, respectively, indicating the rapid increase of reaction rate with temperature The evolution of HC^. results in unsaturation in the polymer chain. The dehydrochlorination appears to progress in stepwise fashion along the polymer chain to produce conjugated unsaturation. This leads to color formation in the residue and cyclization reactions producing benzene and related compounds.
Both Michal [20] and Woolley [21] found about 70 compounds among the thermal decomposition products of PVC resin. Table 1 lists those products identified by Michal from PVC and some vinyl copolymers. Among those identified were aromatic hydrocarbons, unsaturated hydrocarbons, and chlorine-containing compounds. Benzene was the most prevalent organic product, amounting to several percent of the original sample weight [19,21]. Others were present in much smaller amounts.
Lat timer and Kroenke [22] have shown by isotopic labeling experiments that the benzene and other conjugated species result from intramolecular cyclization, while the mixed aromatic-aliphatic compounds, such as toluene involve intermolecular reactions. Only small amounts of the vinyl chloride monomer and other chlorine-containing organic compounds are found.
In the presence of oxygen, the decomposition is somewhat accelerated but primary products are little affected. Carbon dioxide (CO 2 ), carbon monoxide (CO) and water are formed from the dehydrochlorinated residue, particularly at high temperatures, thus decreasing the yield of organic products. Oxygenated organics are not found in significant amounts [21].
The presence of plasticizers and other additives will give additional products characteristic of the additives. There appears to be little interaction between the PVC and the plasticizer. O'Mara suggests that gas chromatography/mass spectrometry (GC/MS) "fingerprints" of plastisols can be used to identify the plasticizers [19]. Inorganic fillers capable of reacting with HCJl, such as CaC02 , can reduce the amount of HCJ, evolved [23]. Other inorganic additives, such as molybdenum compounds, have been promoted as smoke reducing compounds. Presumably, they catalyze the oxidation of particulates.
The presence of phosgene among the combustion products of PVC has been investigated because of its well known toxic properties. Woolley [21] was unable to confirm the presence of the compound among the thermal decomposition products of PVC in air. Brown and Birky [24] made a careful search for phosgene using a variety of decomposition conditions. Under most conditions, only trace amounts (0.01 to 0.16 mg/g) of phosgene were detected, however, greater quantities (up to 1 .6 mg/g) were formed when decomposition occurred as a result of electrical arcing.
Burning PVC can produce large amounts of particulate matter. The actual amount will depend on the conditions of combustion, particularly the oxygen supply, and the presence of other fuels. HCZ may be adsorbed on the particulates and be transported in that manner [25]. HCil may also be condensed by moist air to form a corrosive hydrochloric acid mist. This mist and smoke can obscure vision, thus interfering with escape and rescue. The pulmonary damage from PVC smoke inhalation has been shown to be more severe than that from HC£ alone probably because the particulates may be drawn into and deposited in the lungs where the adsorbed or dissolved HCZ enhances the toxic effect [54] .
On the other hand, the potential exists that a water spray applied to the fire gases may remove some of the HCJl [26] .

TOXICITY OF PVC PYROLYSIS AND COMBUSTION PRODUCTS
Many studies of the toxicity of PVC combustion products have been performed. A recent review by Hinderer [14] presents a summary of the findings. Again, the drawing of quantitative conclusions is made difficult by variations and uncertainties in the materials tested, differences in proce-dure, and dissimilar ways of presenting the results. Nevertheless, it is possible to form a useful qualitative picture of the toxicological behavior of PVC in fires.
According to the literature discussed in detail below, the acute inhalation toxicity during and following exposures to PVC thermal degradation products appears to be due almost entirely to the presence of CO and HC2.. The chronic effects of multiple exposures to PVC thermal decomposition products have only been examined in a few cases and are not well understood [27].
Additives contained in PVC compositions and other fuels that accompany PVC may produce other toxic products. For example, an experimental PVC composition containing zinc ferrocyanide was shown to produce a rapidly acting toxic combustion atmosphere [28] . This effect was probably due to hydrogen cyanide (HCN) which formed by a reaction of HC£. with the ferrocyanide (Table 9).
Since carbon monoxide and hydrogen chloride exert their physiological effects through different mechanisms, direct interaction between them is not expected and the assessment of their individual roles in casualty production is difficult. Carbon monoxide is an asphyxiant causing incapacitation and death primarily during exposure. HCJl, on the other hand, is both a sensory and pulmonary irritant. As a sensory irritant in the upper respiratory tract, HCil can depress the respiration rate, thus slowing the ingestion of CO.
The sensory effect is rapid and concentration-dependent [11,54]. HC2. can also act as a pulmonary irritant, causing severe damage to the lower respiratory tract and post-exposure death [54].
Hydrogen chloride is extremely soluble in aqueous body fluids.
Therefore, much HCi is removed from the inhaled gases in the nose and upper respiratory passages. This occurrence affords some protection against pulmonary damage. Rats and mice, used in most experiments, are obligatory nose breathers. Humans, on the other hand, when confronted with a sensory irritant, tend to bypass the nasal passages by breathing through the mouth.
Some investigators have thus predicted that pulmonary damage from HCJl would be much more severe in humans than in rodents [29].
Alarie and his coworkers examined the effects in mice exposed to HCJl or PVC decomposition products both by normal breathing and through a tracheal cannula that delivered the gases directly to the lungs. They found the PVC LL^q (30 min exposure plus 3 hour post-exposure period) to be about seven times greater for the nose breathing mice than for the cannulated mice [30].
The HC^. LC^q (10 min exposure plus 3 hour post-exposure period) values for mice with and without tracheal cannulas were 1100 and 10100 ppm, respectively. From these results, they postulate that the acute lethality of materials that produce water soluble acid gases will be underestimated by experiments with small rodents and that the actual toxicity to humans will be underestimated by this ratio. 2 LC^q is defined by Alarie as the mass of tested material which when thermally decomposed via the University of Pittsburgh test method causes 50% of the animals to die in the specified exposure time plus a short post-exposure observation period; however, in combustion studies, LC which stands for lethal concentration is usually defined as the mass of tested material divided by the available volume of the combustion products or the concentration of a specific gas.
Therefore, for the purposes of this report, the results of these experiments will be presented as LL^q indicating that a lethal loading of material is the figure of merit. Work by Kaplan et al. [31] using juvenile baboons, which like humans may breathe through their mouths, does not support this conjecture. They showed the lethal amount of HCil to be about the same for baboons (5 min at 16600 ppm) and nose-breathing rats (5 min at 15200 ppm). These deaths were postexposure. Moreover, exposure to high concentrations (baboon: 17300 ppm; rat: 77000 ppm) of this sensory irritant for 5 min did not incapacitate either animal, as measured by an escape task.
Barrow et al. [32] have made extensive studies of the irritant effects of inhaled gases including HCil and the total thermal decomposition products of PVC . Both caused a reduction in respiration rate. The maximum percent decrease in the respiration rate was found to be a linear function of the log of the concentration over a considerable range of concentrations of the irritant gas. They define the RD 5 Q as the concentration of gas necessary to cause a 50% decrease in the respiratory rate from that of the animal prior to exposure. Since pulmonary irritants, in contrast to sensory irritants, increase the respiratory rate while decreasing the tidal volume, it can be judged that PVC products act more like sensory than like pulmonary irritants.
The decomposition products of PVC were reported to be somewhat more potent than pure HCi. Unfortunately, the method of exposure in the two cases was different so no quantitative comparison is possible. The RD 5 Q depends on the maximum concentration of products in a flowing air stream rather than the total amount. The PVC samples were decomposed in a ramp-heated furnace where the concentration of decomposition products varied with time and no determination of the maximum concentration was made. Experimental details, including extensive observations of physiological and pathological effects, are given in a thesis by Barrow [11]. Mice and rabbits were used; some of the rabbits had tracheal cannulas to study pulmonary irritants. The extensive damage noted in the upper and lower respiratory tract 24 hours following exposures was similar for both HCil and the PVC decomposition products. At high concentrations, eye damage resulted.
Wong et al. [33] have examined the long term pulmonary performance of guinea pigs following 30 min acute inhalation exposures to different concentrations of PVC thermal decomposition products. They used a nonlntrusive and noninvasive method in which the animals are exposed to the toxic combustion products and then challenged with a mixture of 10% CO 2 , 20% ©2 and 70% N 2 .
In control animals, this mixture will increase both tidal volume and respiratory rate. In the exposed animals, these responses are depressedan indication of pulmonary damage. The gradual return to normal indicates the recovery of the animals. Wong et al. found that only the group exposed to the greatest amount of combustion products from PVC (10 grams or~16 mg/il) still showed lowered pulmonary performance 57 days following the exposure. Animals exposed to lower concentrations recovered prior to that time. 4.1 Laboratory Measurements of PVC Decomposition Product Toxicity Cornish and Abar [34] were among the first to investigate the toxicity of PVC decomposition products. They used a flow system with a temperature slowly rising (3°C/min) to a maximum of 550°C. The air flow was 2.8 Jl»min^and the exposure chamber (a dessicator) volume was 7 Jl so the products were quickly swept through and out of the chamber where 4 rats were exposed for approximately 140 minutes or until death. They were observed for 24 hours following the exposure. The results of these studies on seven different PVC samples are shown in Table 2. The samples identified as pure homopolymers of PVC were more toxic on a sample mass basis. Carbon monoxide appeared to be a factor in causing lethality, but the data were inconsistent.
In a later study, Cornish et al. [35] compared results obtained in their dynamic system (rising temperature of 3°C/min until a maximum of 800°C was reached; five animals were exposed for 140 min and observed for 7 days; airflow was 3 Jl/min) to those obtained in a 1500 I static system during a 4 hour exposure of 15 rats (temperature reached in 1-2 min; material exposed for 6-10 min). The observed LC^q values were 4 mg/ in the dynamic system and 36 mg/ in the static system.
At about the same time, Hofmann and Oettel [36] and later Hofmann and Sand [37] studied the toxicity of PVC decomposition products. They subjected rats to the products from a rigid PVC (5 g samples) decomposed in the DIN toxicity test furnace at a series of fixed temperatures. This apparatus supplies an approximately constant concentration of decomposition products to a flowing air stream (100 Jl/hr) during a 30 minute exposure period. The mass loading density was 50 mg/£., and in the later experiments 20 mg/i and 10 mg/l due to subsequent air dilutions. The expected variation of toxicity with temperature (Table 3) was observed and correlated with the decomposition temperature of the polymer. For a mass loading of 50 rag/i, no fatalities occurred from exposure to the products generated at 200°C, but 10 of 12 animals died when PVC was decomposed at 300°C. Deaths were reported during the 30 minute exposure, but no post-exposure observations were taken. Since PVC decomposition products have been found, in other methods, to produce many post-exposure deaths of animals that survive the exposure, we can assume that the specimens would have been found to be much more toxic if the animals that survived the exposure had been observed post-exposure. The carboxyhemoglobin content of the blood in the dead animals was below the lethal level expected for pure CO, an indication of the presence of toxicants in addition to CO.
Kishitani [38] studied the toxicity of smoke from 30 grams of a rigid PVC board to mice.
The material was decomposed at temperatures rising from ambient to 740°C in 15 min. Two liters per min of the smoke flowed through the 56 I exposure chamber in which one mouse was exposed for 15 min. Five tests were run on each material.
It is difficult to make a quantitative interpretation of his data. All the animals died during or immediately following the 15 min exposure. COHb levels ranged from 13-30%. Kishitani and Nakamura [39] made a more detailed study in 1974, but, again, quantitative evaluation of their results is difficult. They noted that all surviving animals died within 24 hours following the exposures. Rapid decay of the HCJl concentration was noted. However, they believed that at 350°C, the toxicity was mainly due to HCA and that at 500°C and above, it was due to CO and HCJl.
Kishitani and Yusa [40] attempted to determine the relative toxicities of the thermal decomposition products of various building materials by observing the exposure time necessary to cause the collapse of mice placed in rotating cages.
Post-exposure deaths were noted. Based on times-to-collapse , comparison of a series of materials, which included five different types of wood, two melamine-resin sheets, a polyacrylonitrile, a polyamide, a wool, a urea resin foam, a polystyrene foam, and a polyurethane foam, showed a rigid PVC foam to be of medium toxicity and a PVC flexible sheet to be the least toxic.
Boudene et al. [26] determined the effects of water solubility on the toxicity of PVC decomposition products. They used a traveling furnace in a flow system similar to the DIN system but with the furnace travel cocurrent with the air flow (Table 4). Rabbits were exposed for 30 minutes including a 4 hour post-exposure period. When the products of PVC decomposition were passed through a water bubbler before entering the exposure chamber, the toxicity was decreased 5 to 6  Alarie and Anderson [42] give results for a 92% PVC resin which showed an LL^q of 7.0 g for a 30 minute exposure plus a 10 minute post-exposure period.
A plasticized formulation containing 46% homopolymer showed a correspondingly lower toxicity with an LL^q of 15.2 g. A 92% PVC resin containing 5% zinc ferrocyanide produced an LL^q of 2.3 g. Assuming all the material decomposed and the combustion products are diluted by 600 I of air in this flow-through dynamic system, the nominal 30 minute LC^q values would be 11.7, 25.3, and 3.8 tag/ for the three formulations, respectively. In Alarie 's classification scheme, the first two PVC materials fall into the more toxic than wood class, whereas PVC with zinc ferrocyanide is considered much more toxic than wood.
The increased toxicity of the last material is presumably due to the high concentration of HCN that is generated.
In the process of comparing the University of Pittsburgh's test method and the NBS test method, Anderson et al. [43] examined a number of PVC materials. The Pittsburgh test method was modified to include a 14 day postexposure period. The LC^q values are given in Table 5.
Herpol [44] used the DIN method to compare a number of fire retarded and non-fire retarded materials including some containing PVC. Rats were exposed for 30 minutes to the decomposition products of a PVC floor covering and a wall covering heated at three different temperatures, 500, 600, and 700°C. In all cases, 10 grams of material was placed in the furnace. The results for the fire retarded and non-fire retarded PVC materials are shown in Table 6.
From these results, it appears that the fire retarded PVC's were more toxic than the non-fire retarded materials at 500°C (where the materials did not flame). At 600 and 700°C, the toxicities were about equal. In many cases, -17 -the CO and COHb levels seemed high enough to account for the deaths. Since there were no post-exposure observations, the toxicity due to the HCi from the PVC is probably underestimated.
Levin et al. [28] decomposed a PVC resin according to the NBS toxicity test method. They were not able to calculate within-exposure LC^q values since only one rat died during the 30   Toxicological studies performed at NBS on CO in air indicated that 4600 ppm of CO were necessary to cause 50% of the rats to die during the 30 minute exposures [53]. No deaths occurred following these CO exposures. In PVC experiments [28], the highest concentrations (averaged over 30  This implies that at only slightly increased sample masses (~1.5 times), within-exposure deaths due to CO and CO 2 alone could have occurred.
In the non-flaming mode, the levels of CO and CO 2 were not sufficient to produce within-exposure deaths. The sample loading would have had to be increased 4-fold before the concentrations of CO and CO 2 would be lethal.
HCil, on the other hand, produces post-exposure deaths and is probably largely responsible for those deaths. These data emphasize the importance of a post-test observation period for materials producing irritant gases.
These PVC results are compared to those from a variety of materials in should be tested to determine if they rapidly produce concentrations of combustion products that will cause death from very brief exposures (i.e., 10 min). Deaths that occur within a 14 day post-exposure period are counted.
This test was performed for both the PVC and the same formulation with zinc ferrocyanide. The results are given in Table 9 and indicate that the PVC plus zinc ferrocyanide is a very rapidly acting toxicant. All of the animals were incapacitated in the 10 minutes and all died either during or post-exposure.
The high concentrations of HCN that were generated from PVC with zinc ferrocyanide are also noted in Table 9.

Comparison of the Post-Exposure Toxicity of PVC Decomposition Products With That of HCJl
If HCil were the principal toxic product responsible for the deaths following exposure to PVC thermal decomposition products and if that decomposition gave a nearly quantitative yield of HCJl, we would expect good agreement between the properly weighted toxicity of PVC decomposition products and that of pure HCJl gas.
To make this comparison, the toxicity data on HCJl need to be examined first.
Higgins et al. [45] give data on the effects of 5 minute exposures of rats and mice to HCJl gas followed by a seven day observation period. They also investigated the additional effect of exposures to CO (concentrations tested were calculated to produce 25% carboxyhemoglobin levels) on HCJl toxicity. Their results are given in Table 10. The difference in LC^q values between exposure to HCJl with and without CO at this sub-lethal level was not statistically significant, supporting the position that HCJl and CO react by independent physiological mechanisms.
In related work by Darmer et al. [46], rats and mice were exposed to HCJl vapors and water saturated aerosols for periods of 5 and 30 minutes and observed for 7 days. Rats were found to be considerably more resistant to HCJl vapors than mice. The five minute LC^q values for the rats and mice were 41000 and 13700 ppm and the 30 minute LC^q values were 4700 and 2600 ppm, respectively (Table 11). No rats died during the 5 minute exposures, although some mice died at the two highest concentrations (26000 and 30000 ppm). In the 30 minute exposures, deaths during the exposures were noted at 27000 ppm and above.
Exposures to the HCJl aerosols did not significantly affect the toxicity. Rabbits and guinea pigs showed a degree of susceptibility to HC£. similar to that of mice. In summary, the HCii LC^q value for 5 minute exposures range from~11000 to 41000 ppm and the 30 minute LC^q values range from 2100 to 5700 ppm. These include post-exposure deaths.
Since the toxicity of HCil occurs primarily following exposures, data from some of the better characterized experiments on the toxicity of PVC decomposition products and which include post-exposure observations are compared to those for HCil gas in Table 12

LARGE-SCALE FIRE EXPERIMENTS
The problem of relating results obtained in a small-scale laboratory apparatus to conditions in a real fire is a universal one in fire research and particularly critical in combustion toxicology. The problem is particularly difficult in the case of PVC for two reasons. First, PVC is among the more fire resistant of commonly encountered fuels. It does not burn readily by itself and usually requires a vigorous fire involving more combustible fuels to bring about its thermal decomposition. The products from these combustible fuels, usually including substantial amounts of CO, mix with the products from the PVC and may be responsible for much of the toxic effects observed.
Second, the HC£ in the smoke can behave not as an inert mixture when the smoke moves away from the fire source, showing, instead, condensation on and settling out with particulate matter, adsorption on surfaces, and dissolving in water droplets [47]. Consequently, the exposure to smoke that has traveled some distance in a building fire may be quite different from that received in a close-coupled laboratory system.
In an effort to obtain a better understanding of the problem, a number of investigators burned PVC in large-scale experiments simulating real fire scenarios. Stark et al. [48] were among the first to carry out such experiments at the Fire Research Station in 1969. They burned cellulosic fuels in a small compartment with PVC wall lining or with strips of rigid PVC. With complete combustion of the PVC, the HCZ was equivalent to the chlorine content. They concluded that the solid PVC would contribute a significant amount of HC^. to the fire gases but that PVC wall paper would not be likely to contribute much to the toxicity due to the small quantity of PVC material compared to the amount of other furnishings.
Packham and Crawford [49] thermally decomposed electrical nonmetallic tubing (ENT) made from rigid PVC and studied the toxicity in both small-scale and large-scale experimental conditions. In the small-scale tests, the NBS toxicity test method was utilized to decompose the materials in the nonflaming mode. The resultant hC^Q value was 28.5 mg/Z,; all lethalities were post-exposure.
In the large-scale tests, samples were decomposed in a closed -22 -\ 2 21.7 room, using either a 2.2 kg wood crib or a radiant flux of 25 kW/m to decompose the ENT. In the experiments with wood cribs, they concluded that high temperatures and CO from the cribs posed a greater threat than the HCJl.
In the radiant heating experiments, significant concentrations of PVC decomposition products were found, but animal fatalities were still considered to result from heat stress rather than smoke inhalation. HCi. recoveries ranged from 23 to 50% of the theoretical in the large-scale tests; whereas in the small-scale tests, 64 to 96% of the theoretical yield was generated.
In none of these tests was a quantitative full-scale/bench-scale correlation obtained. 6.

PVC IN FIRES
Because of the wide occurrence of PVC in buildings and residences, it is to be expected that PVC will be present in many accidental fires. Since the decomposition products are usually accompanied by large quantities of combustion products from other fuels, it is difficult to judge their contribution to the toxic threat. Post-exposure analytical studies of fire victims are of little help since chloride ion, the most characteristic PVC product, is a common constituent of all body fluids. Casualties attributed to "smoke inhalation" are frequently blamed on PVC although there may be little direct evidence of its having played a significant role. For example, the Beverly Hills Supper Club fire is frequently cited as a fire where PVC contributed significantly to injury and loss of life, yet the amount of PVC involved was small compared to the amount of other fuels capable of producing toxic combustion products [3 ] Lowry et al. [50] analyzed gas samples from 72 residential fires in the Dallas area. Hydrogen chloride was found in 9% of the samples, with an average concentration of 1.1 ppm and a maximum concentration of 40 ppm. Of the gases examined, CO was the principal toxic gas found.
A small number of fires have occurred where PVC decomposition products appear to have played a significant role. These are fires involving electrical insulation and apparatus, although, in some of these cases, the origin of the toxic gases is not clearly established. Dyer and Esch [5] describe a fire "confined to an office copying machine constructed of plastic and Teflon parts" in which fire fighters sustained respiratory injuries and one fire fighter died 24 hours later. The "plastic" was not identified and the involvement of PVC was not demonstrated, although the article was titled, "Polyvinyl Chloride Toxicity in Fires".
Einhorn and Grunnet [51] described a number of fires in the Salt Lake City area where PVC was specifically identified by analysis as a major fuel component. The fires resulted in respiratory and skin injuries to the fire fighters and other exposed personnel.
Perhaps the prime example of a PVC fire was the New York Telephone Company fire of 1975 [3] Fire burned through an eleven story switching center, fed largely by electrical insulation of wires and cables. A total of more than 700 fire fighters were involved and many received varying degrees of respiratory injury. Follow up studies [52] suggest long range health effects from exposure to PVC smoke. Other more limited studies suggest similar results [27].

CONCLUSIONS
When strongly heated or exposed to a fire environment, PVC-containing materials decompose at relatively low temperatures to give a nearly quantitative yield of HCZ, in addition to CO and a wide variety of minor organic products. At higher temperatures, the concentrations of C0£ and its potentiating effect on CO should also be considered when assessing the toxicity of the combustion products.
HCi appears to be the major toxicant which is responsible for the deaths that occur following acute inhalation exposures to PVC decomposition products.
HCJZ, is both a sensory irritant and a pulmonary irritant. In the latter role, it can cause delayed respiratory difficulties, permanent injury, or death. In one case, however, a laboratory formulation containing the additive zinc ferrocyanide produced significant quantities of HCN that were probably the cause of the animal deaths that occurred from 10 min exposures.
PVC decomposition products are usually accompanied in fires by large amounts of products from other, more combustible, fuels. This, together with the rapid decay of HC£ concentration presumed to occur in the fire environment, makes it difficult to determine the contribution of PVC to the toxicity of the atmosphere. 9. [6] Dufour, R.E., Survey of Available Information on the Toxicity of the Combustion and Thermal Decomposition Products of Certain Building Materials Under Fire Conditions, Underwriters Laboratories Bulletin of Research No. 53, Underwriters Laboratories, Inc., Chicago (1963). [7] Tewarson, A., The effects of fire exposed electrical wiring systems on escape potential from buildings. Part 1 -A literature review of pyrolysis/corabustion products and toxicities -poly(vinyl chloride  [9] Tewarson, A. and Pion, R.F., The effects of fire exposed electrical wiring systems on escape potential from buildings.  [11] Barrow, C., Toxicity of plastic combustion products. Toxicological methodologies to assess the relative hazards of thermal decomposition products from polymeric materials, NBS-GCR-77-85 , Nat. Bur. Stand., other products present in amounts less than 2% of total products not detected Blood taken from dead animals. Rats were exposed for 115-140 minutes, under ramped heating conditions. Lethalities include a 24 hr post-exposure period. In all cases, 100 £./hr of air passed through the furnace containing 5 g samples ^A n additional 100 il/hr of air was added prior to animal exposures resulting in 50 mg/£ specimen charges.
Specimen charge reduced to 20 mg/H by an additional 400 il/hr of air.
Specimen charge reduced to 10 mg/il by an additional 900 il/hr of air.   COHb is the mean value of the dead animals unless there were no deaths, in which case, it is the mean of the live animals.
Contains phosphate plasticizers and antimony trioxide.
Contains phosphate and chlorine-containing plasticizers and antimony trioxide.
Short bursts of flames.