Document RJL9aGqjm8roRoEnvM9agoJ3n

Antimony-Based Flame Retardant Systems in Plasticized Polyvinyl Chloride Compounds JAMES C. FURNIVALL, ALAN D. KUPFER, and JOHN L. IRVINE Conoco Inc. Ponca City, Oklahoma 74601 A method of choosing the most effective way to meet specific flame retardant requirements in polyvinyl chloride compounds, which uses flammability tests based on oxygen indexes, shows that antimonybased and non-antimony retardants perform as well as commonly used antimony oxide retardants, and at a reduced cost INTRODUCTION The flammability of flexible PVC compounds is an important property in many applications of this versatile plastic. In recent years, the cost of flame retardant additives and the number of addi tives available on the market have increased sharp ly. As a supplier of PVC compound to the wire and cable industry, we felt that it was important for us to understand the most effective way to provide mate rials with which our customers could make wire and cable products meeting various flammability requirements. This study was planned to provide a base from which additional, more detailed work could be done. In order to provide a general data base, the flame retardant additives tested were broken down into the following categories; grades of antimony oxide, antimony oxide replacements, and antimony oxide synergists. The flammabillity data were con verted into cost-effective data to provide a practical basis for comparison. EXPERIMENTAL The flame retardant additives were tested in a representative wire and cable compound formula tion. phr PVC resin Plasticizer Lead stabilizer Lubricant Calcium carbonate Flame retardant 100.00 50.0 and 60.0 5.0 0.3 20-y y Since this study was directed towards wire and cable applications, DIDP was used as the plas ticizer in the comparison of the flame retardant ad ditives. The flame retardant additives were tested at 50-phr and 60-phr DIDP. DIDP was also com pared against DOP, DTDP, and TOTM to deter mine the effect of plasticizer on flammability. The type and parts level of flame retardant addi tive used was the most important variable in this study. The various types of flame retardants tested and their costs are listed in Table 1. The prices listed in Table 1 are list prices based on truckload quantities as of October 1,1980. These prices were the basis ofthe calculations that show the cost effec tiveness of various additives. The samples were milled on a two roll mill for five minutes at 320*F. Seventy-two mil thick plaques were pressed at 320F and 40 tons of pressure. The test samples for oxygen index were cut from these plaques. The total calcium carbonate/flame retardant level was held constant at 20 phr. All samples were conditioned in accord ance with ASTM D618. Limiting oxygen index was chosen as the primary test criteria. The simplicity of the test and the high degree of reproducibility were considered impor tant qualities in the testing of varied formulations. The limiting oxygen index was measured according to the procedure outlined in ASTM D2S63 on an apparatus manufactured by General Electric. The oxygen index was measured to the nearest 0.5 per cent. The U.L. 94 vertical bum test was also done. This test was done according to the procedure out lined in paragraphs 3.6 through 3.15. The sample thickness was 40 mil. DISCUSSION The mechanisms of flame retardancy have been studied (1). In halogenated polymers such as polyvinyl chloride (PVC), antimony oxide has been Table 1. Gradaa ol Antimony Oxlda Cost/pound Ultra fins (<i micron particle size) Lowing tinting (2.5-3.5 micron particle size) $1.96 1.69 Antimony Oxide Substitutes Antimony-based inorganic complex (41.0% Sb) Antimony-silicate complex (25.0% SbiOi) Antimony oxide/(25.0% Sb,0,) ammonium tluoroborate (50% SbiOi) 1-39 1.07 1.80 2.05 Antimony Oxide Synergists Zinc borate Magnesium oxide/zinc oxide 0.79 1.10 JOURNAL OF VINYL TECHNOLOGY, 5EFTEMBER 1981, VOL. 3, NO. 3 179 Jama C. FummaJl, Alan D. Kupfer, and John L. Irvine shown to be an effective flame retardant. It is be lieved that antimony oxide is activated by reaction with halogens, forming, antimony trihalides and antimony oxyhalides. These two compounds work primarily as flame-phase flame retarders. The cost ofantimony oxide has steadily increased over the last couple of years to nearly $2 per lb. In this study, we investigated the use of less expen sive antimony-based flame retardant systems. Sys tems were studied that would completely replace the use ofpure antimony oxide. Other systems were studied that would give a synergistic effect when added to antimony oxide, thus reducing the level of the more costly antimony oxide. . The study of flame retardance is complicated by the many variables in compound formulating and their effect on the flame retardant. Different formu lations may yield different results. It is necessary to note that the results presented in this paper are based on limiting oxygen index values. The oxygen index value may not correlate to other types of flammability tests (2). However, it does provide an efficient means of accurately comparing various flame retardant additives on a small scale. The results from limiting oxygen index testing were converted to a monetary value. This allows for a comparison ofvarious flame retardant systems on a cost effectiveness basis. This paper assesses the cost of a 20-part flame retardant/calcium carbonate filler system per one hundred pounds of resin. Table 2 is a list of results for two grades of anti mony oxide and four antimony-based substitutes. The results in Table 2 show that the two grades of antimony oxide, the antimony-based inorganic complex, and the 1:1 antimony oxide/ammonium fluoroborate mixture had similar effects on oxygen index at 50-phr plasticizer level. These four flame retardant additives outperformed the silicate/ antimony complex and the 1:3 antimony oxide/ ammonium fluoroborate mixture. Figure 1 shows the relative cost effectiveness of the antimony oxides and the antimony-based sub stitutes. It is apparent that there is a reduced cost effectiveness, with these additives, as higher limit ing oxygen indexes are obtained. The decrease in cost effectiveness is a direct result of a leveling effect which occurs around 29.0 to 30.0 percent oxygen index. Much higher levels of flame re tardant additive are required to achieve a given increase in oxygen index. Fig. 1. Antimony oxide and antimony-based substitutes with SO DtDP as plasticizer; (l) antimony oxide, (2) antimony-based inorganic complex, (3) ultrafine antimony oxide, (4) 1:1 anti mony oxide/ammonium fluoroborate, (S) 1:3 antimony oxide/ ammonium fluoroborate, (6) silicate-antimony complex. It is important to notice that even though anti mony oxide is widely used as a flame retardant, it may not be the most cost efficient The antimonybased inorganic material can provide the same oxy gen index as antimony oxide at a reduced cost. The 1:1 combination ofantimony oxide and ammonium fluoroborate also appears to perform as well as pure antimony oxide at low ana moderate oxygen in dexes. The silicate-antimony complex seems to in crease cost and lower oxygen index. The advantage of increasing the surface area of antimony oxide is more than offset by the reduced amount of anti mony oxide obtained per dollar spent. At 60-phr plasticizer, the flame retardant addi tives performed the same relative to one another as at 50-phr plasticizer. All testing showed lower oxy gen indexes. However, at 60-phr plasticizer level, the antimony-based inorganic complex no longer performed as well as pure antimony oxide. phr ot Flame ratardant additive 0 1 3 5 7 9 Table 2. Oxygen Index for Antimony Oxide and Antimony-Baaed Substitutes in 50-phr DIDP Antimony oxide 23.5 24.5 26.5 28.5 29.5 29.5 Ultrafine antimony oxide 23.5 24.0 27.5 29.0 -- -- Antimony based inorganic 23.5 25.0 26.5 2B.0 29.0 29.5 Silicate antimony complex 23.5 24.0 24.0 25.0 -- -- NH.BF, 1:1 1:3 23.5 23.5 25.5 25.5 27.5 26.5 28.5 27.0 -- -- --- 180 *JCC JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 037450 Antimony-Based Flame Retardant Systems in Plasticized Polyvinyl Chloride Compounds Table 3 is a list of results of two antimony synergists-zinc borate and a magnesium oxide/zinc oxide mixture mixed with 1- and 3-phr antimony oxide. At low levels, zinc borate appears to be more effective than the magnesium oxide/zinc oxide mix ture. However, at higher part levels, there is a more pronounced leveling effect with zinc borate than with a magnesium oxide/zinc oxide mixture. Figure 2 shows the relative cost effectiveness of the zinc borate and magnesium oxide/zinc oxide mixture. Zinc borate provides a better cost-efficient means of increasing oxygen index at low and moderate oxygen index values. For oxygen index values above 28 percent, the magnesium oxide/zinc oxide mixture becomes the best cost-efficient means. At 60-phr plasticizer, the same pattern is seen as in Fig. 2, but all oxygen index values are lower. When tested for rating in the U.L. 94 vertical burn test, all flame retardant mixtures received a V-0 rating. The standards containing no flame re tardants were rated V-l. The final part ofthis study investigated the effect different plasticizers have on flammability. Table 4 is a list of oxygen index values when various levels of antimony oxide is added to 50-phr DOP, DIOP, DTDP, and TOTM. Table 4 shows antimony oxide has the least effect on DIDP. The effectiveness of Table 3. Oxygen Index for PVC Formulations Containing Antimony xide and a Synergist Synergist, phr Antimony oxide 1 phr 3 phr Zinc borate 0 24.5 1 25.5 3 27.0 5 27.0 Magnesium oxide/zinc oxide 0 24.5 1 24.5 3 25.0 5 26.5 26.5 28.0 28.0 28.5 26.5 27.0 28.5 30.0 Table 4. The Oxygen Index of Plasticizers Containing Various Levels of Antimony Oxide (based on 50-phr plasticizer) Antimony Oxide (phr) DOP DIDP DTDP TOTM 0 25.0 24.0 23.0 24.5 1 25.0 24.5 25.5 27.5 3 28.5 26.5 27.5 28.5 5 30.0 28.5 29.0 30.0 10 30.5 29.5 -- -- antimony oxide seems to begin to level off at 5 phr. When the part level was increased to 10 phr, there was only an increase of0.5 percent in oxygen index for DOP and 1.0 percent for DIDP. This study has attempted to provide a basis for choosing the most effective way to meet specific flame retardant requirements while keeping cost at a minimum. Specifically, it is important to consider antimony-based and non-antimony flame re tardants for completely or partially replacing anti mony oxide. Based on oxygen index values, manyof these flame retardants will perform as well as pure antimony oxide at a reduced cost. Additional work is planned to see if other flam mability test methods lead to the same conclusions. Also, additional antimony oxide synergists and re placements will be examined. ACKNOWLEDGMENTS We would like to acknowledge the assistance of Frank A. Suess Jr. and Walter L. Adkins in this work. We would like to thank Eugene R. Cox for technical assistance. Fig. 2. Antimony oxide tynergittt with SO phr DIDP at plat ticizer; (1) antimony oxide, (2) 3 phr antimony oxide and mag nesium oxidetzinc oxide, (3) 3 phr antimony oxide and sine borate, (4) J phr antimony oxide and zinc borate, (5) 1 phr antimony oxide and magnesium oxidetzinc oxide. REFERENCES 1. Kirk Othmer, " Encyclopedia ofChemical Technology," Vol. 10, p. 357, Third Edition, John Wiley and Sons Inc, (1980). 2. Bernard J. Hill, "How Predictive Are Small-Scale Flame Tests," SPE Antec, New York, New York (May 1980). i JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 ucc 037451 181 I Role of Poly(Vinylchloride) and Di-2-Ethyl Hexyl-Phthalate in the Smoke Formation horn Plasticized Poly(Vinylchloride) ALAIN MICHEL, ALAIN SAINRAT, and MICHEL BERT CNRS--Laboratoire des Matiriaux Organiques BP 24-69390 VERNA1SON, France The increase ofthe smoke level from PVC plasticized with di-2-ethyI hexyl phthalate with respect to the rigid PVC is caused mainly by interaction ofthe plasticizer with HCI evolved from the polymer when the temperature is higher than 200*C. This interaction causes the increase of the yield of phthalic anhydride probably through HCI as catalyst for OOP decomposition but phthalic anhydride formation parallels the formation ofproducts enabled to increase the smoke level such as phthalic acid. In the presence of metallic compounds (iron, zinc, aluminium) it is possible to favor the formation of phthalic anhy dride with respect to these products which are responsible for the smoke production. Then phthalic anhydride can Ire used as a tracer to estimate the efficiency of an additive or a combination of additives as smoke suppressant for DOP because its smoke level varies inversely with the yield ofphthalic anhydride. The best combination to reduce the smoke level from plasticized PVC is obtained with the binary systems based upon copper compound, mainly efficient as smoke suppressant for PVC and either zinc -or aluminium compounds, mainly efficient as smoke suppressant for OOP. INTRODUCTION The use of plastics in many applications is linked to their resistance to combustion and to smoke formation. In connection with the fire problem, the main advantage of the polyvinylchloride) is a high self-ignition temperature which is above 500*C. Further, owing to its chemical structure it is a self extinguishing polymer. Nevertheless PVC pro duces more smoke and toxic and corrosive gas than many other polymers upon combustion. Much of this is due to the presence of HCI. Furthermore PVC is used extensively in modem construction and most PVC uses require that the polymer be compounded with plasticizers. Plasticizer levels are usually specified in terms ofparts by weight per hundred parts of PVC (phr). The common hardness and flexibility requirements are met in the range from about 30 to about 100 phr. The most important of these plasticizers are high boiling esters of di basic aliphatic and aromatic acids which, ofcourse, are flammable. The most extensively used plas ticizer is the di-2-ethyl hexyl phthalate (DOP). Its flash point is 216C and when PVC is compounded with DOP the result is a reduction in flame retardancy and an increase in the smoke level. Then the primary fuel is the plasticizer and not the poly mer itself. Thus in compounded PVC fire retardant and smoke suppressant additives are often used. As a smoke suppressant an effective additive must re duce the smoke from plasticizer as well as from the resin. An improved knowledge ofthe mechanism of smoke formation from the polymer and from the plasticizer as well as from their mixture is impor tant. Studies on high temperature pyrolysis of PVC and plastisols either under nitrogen (1) or air (2) have shown that the nature of the main products of the decomposition of PVC, mainly HCI and ben zene are unchanged. The three main products of the DOP decomposition (3) (4) (5) are phthalic anhydride, 2-ethyl-hexene and 2-ethyl-hexanol. The presence of HCI or PVC prevents the forma tion of this last product (3). This fact suggests that the PVC causes a change in the mechanism of DOP decomposition through HCI elimination, primarily between 200 and 300*0. Dickens (6) has shown that among the three previous products only phthalic anhydride and 2-ethyl-hexene have important smoke levels but they do not give the smoke level found for pure PVC. The purpose of this paper is to determine the exact origin of the smoke from the pyrolysis products of the individual components, both under smoldering and in flaming conditions and to understand the cause of the increase of the smoke level from plasticized PVC as compared with rigid PVC and pure DOP. EXPERIMENTAL Smoke Measurements Measurements ofthe smoke density were carried out in dynamic conditions according to the AFNOR Test T 57073 already described (7, 8). The sample 182 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 Rote ofPoty(Vinytchloride) and Di-2-Ethyl Hexyl-Phtholate in the Smoke Formahon from Plasticized PotyfVinylchloride) placed into a quartz container is suddenly intro duced into a quartz tube, itselfinside a tubular oven previously heated at a fixed temperature. The ex periments were carried out under air flow and sometimes nitrogen flow of 120 liters/h at tempera tures ranging from 400-600C. The smoke evolution was followed by optical measurement ofthe obscu ration of a light beam across a measured path be tween a light source and a photo cell receiver. An additional air circulation prevents any redeposition on the windows of the light source and photocell receiver which would alter the light transmission. The attenuation of the light beam was measured by a device designed by us and which allows to record continuously the optical density variations (Log /,,//) between 0 and 4 vs time. The smoke level (S) is calculated according to Eq. 1: where A is the area of the surface under the obscu ration curve, u is the area unit limited by the surface defined by the optical density equal to 1 and the time equal to 1 min. P is the weight ofthe sample in grams. Then S is expressed in decade min g"1. For the common use of the AFNOR test the smoke density chamber is connected to the furnace through a 2 liter flask to ensure good homogeneity of smoke. Nevertheless the suppression ofthis flask permits separation of the smokes produced from each component of the plasticized PVC as shown either in smoldering (Fig. 1 a) or in flaming (Fig. 1 b) conditions. However the amount of the DOP must not be higher than 50 phr. Above 50 phr the separa tion of the two waves is not great enough. At 600"C (flaming conditions) the response ofthe opacimeter shows only one peak and the contribution of the smoke produced from each component cannot be evaluated whatever the amount of plasticizer. The DOP and PVC are mixed together as a pow der at room temperature using an electric mill. Then the powder is pressed as rods under 1 ton/cm1 using a mold of length 100 mm, width 6.5 mm and thickness 3 mm. At 400*C and 500C the weight of the samples is 40 mg for pure DOP and the weight of the plas ticized PVC is such that the weight of the plas ticizer is about 30-40 mg. At 600C the weight of pure DOP and the weight of plasticized PVC is 150 mg. The influence of HC1 on the smoke level from the DOP has been studied with a device designed by us. The polymer is put in the lower compartment of a parallelepipedic cell of quartz divided in two parts by means of a sintered glass disc. The plas ticizer is placed in the upper side. The cell is joined to a cylindrical glass tube to introduce it inside the quartz tube itself inside the furnace. Gas flow (60 1/h) nitrogen or air forces HC1 to diffuse across the sintered glass. For experiments carried out at 400C either under nitrogen or air the weight of DOP is 100 mg TIME (min) Fig. 1. Obscuration curves from PVC plasticized with di-2ethyl-hexul phthalate. a) Smoldering conditions (400"C)--DOP 16.6%; b) Flaming conditions (500'C)--DOP 33.3%; A--Response of the opacimeter for DOP; B--Response of the opacimeter for PVC. and the weight of PVC is 200 mg. For experiments, carried out at 600C under air a gas flow ofpure HC1 (1.66.10"' 1/s) diffuses across the sintered glass. Di-Ethyl-2-Hexyl Phthalate and Plasticized PVC Degradation The DOP and the plasticized PVC degradation has been carried out either in the same fumace used for the smoke measurements or in a flash pyrolyser. In the first case a glass wool filter is put to the exit of the fumace to avoid flame propagation and to retain the soot. The products of decomposi tion are trapped after the filter in acetone cooled at -78C. The furnace is connected to the trap through a heating jacket (300C) to avoid the con densation of high molecular weight products such as phthalic anhydride. Flash pyrolysis experiments were carried out using a Girdel pyrolyser connected to a gas chromatograph (Intersmat type IGC 15) equipped with a flame ionization detector. The DOP or plas ticized PVC was dissolved in tetrahydrofuran at 3 percent by weight and 5 microliters ofthis solution was deposited on the filament. The pyrolysis time was selected as 50 s. The separation and estimation of the products of degradation trapped in acetone or evolved during JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 UCC 037453 183 Alain Michel, Alem Sainrat, and Michel Bert flash pyrolysis was achieved through gas chromatography with temperature programming at heating rate of 10"C/min between 50 and 250C. The column packing is SE-30 (silicone oil on chromosorb). Nitrogen was used as carrier gas at 2.2 kg/cm*. Carbon dioxide and carbon monoxide mea surements have been carried out continuously with two infrared gas analyzers cosma apparatus type Rubis 3000) in line after the furnace and parallel to each other. The gas flow was 80 1/h. Then in smol dering and in flaming conditions it is always possi ble to separate the carbon dioxide resulting from each component (Fig. 2) but not carbon monoxide. As for smoke measurements, the samples are intro duced in the furnace previously heated to the de sired temperature. To avoid the saturation of the analyzers the amount of the samples (pure DOP or plasticized PVC) is 250 mg at 400C, 50 mg at 500*C and 25 mg at 600*0 for the experiments carried out under nitrogen. Under air the amount of the pure DOP and plasticized PVC is the same as previously at 400 and 500C. At 600*C, it is 25 mg for pure DOP and 15 mg for plasticized PVC. RESULTS AND DISCUSSION In our experimental conditions the self-ignition temperatures is between 400 and 500C according to the rate ofthe plasticizer because ofits flash point at 216*C. At 400 and 500C under nitrogen most of the white smoke is produced from the sublimation of the phthalic anhydride evolved from the decom position of the plasticizer. As shown in Table 1 the percentage of char residue at 400 and 500C does not change when the DOP fraction increases and from 500'C it is independent of the tempera ture and represents about 10 percent of the initial weight of the polymer. Furthermore the amount of benzene evolved in flash pyrolysis experiments is constant at 600*C, whatever is die DOP concentra tion but it becomes proportional to the plasticizer percentage from 700C where the benzene is also a product of the DOP decomposition (Fig. 3). In smoldering (400C) or flaming (500C) condi tions the presence of DOP does not change the total amount of COt evolved from the polymer and does not influence its smoke level. All the previous results show that the degrada tion ofthe PVC is not influenced by the presence of the DOP and confirm the conclusions of O'Mara (1) and Boettner(2) who have shown that the pyrolysis of PVC and plastisols at high temperature under nitrogen or air does not modify the nature of the main products of the decomposition of the PVC. In return PVC enhances strongly the smoke level from the DOP either under nitrogen or under air (Table 2). Under nitrogen or in smoldering condi tions under air the response of the opacimeter is partly due to the sublimation of the phthalic anhy dride coming from the DOP decomposition. Under air, in smoldering conditions (400*0) the smoke level from pure DOP is about 12 times higher than the smoke level from PVC. When the two components are mixed, the smoke level from the plasticizer increases very much as well as the amount of phthalic anhydride and it increases when the molar ratio HCl/DOP increases. In flam ing conditions (600*C) there is self-ignition of each mixture whatever may be the DOP content The smoke level of each pure component is equivalent but when they are mixed the smoke level from plasticized PVC is higher by about 30 percent (TabU 3). At 600C it is not possible to differentiate ex perimentally the smoke levels from each compo- ucc 03745-4 TIME (mini Fig. 2. Carbon dioxide evolutionfrom plasticized PVC. a) Smol dering conditions <400'C)--DOP 33.3%; b) Flaming conditions i 500X1)--DOP 33.3%; A--Response of the IR gas analyzer for DOP; B--Response of the IR gas analyzer for PVC. Ot-2-ethyi-haxy! phthaiata Iweight "/*) Fig. 3. Influence of the di-2-ethyl-hexyl phthalate on the ben zeneformation duringflash pyrolysis of the plasticized PVC at 600X (*) and at 7OOX (L). 184 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 Role of Poly(Vmylchloride) and Di-2-Ethyl Hexyl-Phtkalnie in the Smoke Formation from Plasticoed PolyfVmykhloride) OOP X 0 0.166 0.333 0.44 0 0.166 0.333 0.44 PVC 1 -X 1 0.834 0.667 0.56 1 0.834 0.667 0.56 Table 1. Influence of OOP on PVC Pyrolysis and Combueilon Temperature PC) Nitrogen Char residue SfVC wtight, % dcad min * fl"' Char residue weight, % Air decade min g" 29 400 26.5 29.5 30.0 10 500 9 10 10.5 0 25 3.5 0 26 4 0 24.5 traces 0 25.5 1.6 2.9 traces 16 2.2 traces 15.3 2.8 traces 10.7 4.5 traces not defined mole ' g~' 10* 2.64 2.70 3.1 4.6 8 6.4 10.4 8.5 Table 2. Influence of PVC on DOP Pyrolyele and Combustion T(*C) DOP X PVC 1 -X Degraded DOP mole % Nitrogen *ID0* decade min g~` CO^mn mole 10** Phthalic anhydride mole 10** Degraded DOP mole % Air 3, DOT' decade min g_1 CO,, dot, mole * 10fc 0.166 0.834 95 56 5 1.7 91.5 94.2 0.144 400 0.333 0.667 89 56.5 4.1 2 90 56.4 1.2 0.44 0.56 89 39 3.1 2.6 82 82 0.39 10 43 13 0 traces 78 43 1.8 0.166 0.834 500 0.333 0.667 100 98 19 9 46 39 6.8 100 5.6 100 15.2 0.6 7 1.24 0.44 0.56 98 13.5 35 6.7 100 9 0.96 10 84 18 26 6.4 85 16.1 2.54 1 valUM *r* gtwi tor tit* moto of dojradod DO#. Phthalic anhydride mole 10** 7.4 13.8 9.8 4.4 4 traces ' traces traces Table 3. Smoke Level from Plasticized PVC (St) and from DOP Fraction (Sur) in Flaming Conditions (600*C) DOP X PVC 1 -X St deeede min g'1 Soot decade min g~' 0 0.166 0.333 0.444 1 1 0.834 0.667 0.556 0 9.8 12.1 12.0 11.0 8.3 9.8 24 16.5 12.8 8.3 * Cttowtoto# voIum ceer&ng to Sq 2. Table 4. Influence of HCI Evolved from PVC on the Smoke Level from Pure DOP Smoke level (decade ' min g*1) Nitrogen Air 400*C 400*C 600*C OOP DOP + HCI 7.2* 36.7 22* 9.4-, 31 13 * TltoM valmM caimof b* wwipirabto wtffi to* vtoi tor from 00# glvM In m* Tiblt 3 bMnH of m* *pariniaMM candHWni an aHtwwtt nent. Nevertheless as DOP does not influence the degradation of the PVC we can assume that it does not change the smoke level from the polymer (Spec) whatever may be the percentage of the plasticizer. Then with this assumption it is possible to calculate the smoke level from the plasticizer (Spop) (Table 3) with respect to its fraction (X) from the smoke level measurements of the plasticized PVC (ST) accord ing to Eq 2: ST - So,,,<X) + Sm<l - X) (2) for the thermal degradation of DOP (Fig. 4) and causes an increase in its smoke level (Table 4). The role ofHC1 as a catalyst for the degradation of the DOP appears in the following reactions: r In flaming conditions PVC yet enhances the smoke level from the DOP fraction. Besides phthalic anhydride and carbon dioxide undergo the influence of PVC from quantitative point of view (Table 2). These results suggest that the increase of smoke level from plasticized PVC is due mainly to the interaction of the PVC with the mechanisms of DOP degradation through HC1 which is the main product of PVC degradation evolved between 200 and 300C. Furthermore HC1 is an effective catalyst t ,- ucc 037455 > Tim* Fig. 4- OOP degradation at 244C under nitrogen either in absence (*) or in the presence (A) of HCi JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3. NO. 3 185 Alain Michel, Alain Samrat, and Michel Bert C.H, O H--C 4CH,),--CHS A--o--Ah, H*+Cl4"CjHs o\h^A 4CH,)*-CH, A-o^Ah, 0( ^ c--o--CHt A H--A 4CH,),--CHS A.H, 0( C--OR A o A--OH C,H, (I) + CH, - CH, C--OR A (A) with an excess of HC1 (low percentage of plas ticizer) the two ester groups can be decomposed simultaneously into acid according to the Reac tion II: phthalic acid. Furthermore whatever may be the temperature the smoke level from phthalic acid either under nitrogen or under air is always higher than the smoke level from phthalic anhydride \ or O Ha * C.H, <> + 2CH, - A^HrkCH, (II) \ OR \ OH (B) The phthalic anhydride can be formed either from the phthalic acid (B) according to the Reac tion III or from the monoester (A) according to the Reaction IV: (TableS). At temperature higher than 500C the smoke level ofthese two compounds are in a ratio of about 2. The previous observations suggest that any modification of the mechanisms of DOF decompo sition favoring the formation of phthalic anhydride may contribute to decrease the smoke level from the plasticizer. This purpose can be achieved (Table 6) using metallic chlorides as Lewis acid or metallic oxides enabled to be transformed into chlorides as Lewis acid by reaction with HC1 evolved from the polymer when the temperature is raised above 200C. From the previous results it appears as a general rule that the smoke level from DOP decreases when the phthalic anhydride yield increases. The At moderate temperature (200C) the Reac tion IV is quantitative but at higher temperature the yield of phthalic anhydride can be reduced because of spontaneous decarboxylation of Table S. Smeke Level from Phthalic Anhydride and Phthalic Acid T(*C) Smoke level (decade - min a-1) Nitrogen Air Phthalic anhydride Phthalic acid Phthalic anhydride Phthalic acid 400 2.92 515 4.38 600 645 16.4 14.25 3.6 4.23 5.75 5.32 14.3 10.35 8.76 6.00 186 ucc JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 037458 Role of-Poly(Vinykhioride) end Di-2-Ethyl Hexyl-Phthalate m the Smoke Formation from Plasticized PolyfVmylchloride) Table 8. Influence of Metallic Addltlvea on Phthalic Anhydride Formation and Smoka Level from 0 P and PVC In Smoldering Condition* (4S0C) for PVC Plasticized with 16.6% of OOP (PVC 100 mg--OOP 20 mg--Additives 1.5 mg In Weight with Respect to the Polymer) Additives Phthalic anhydride mels/mol* of degraded OOP Smoke level from PVC from DOP decade - min - g~' purs OOP PVC-DOP pvc-dop-f#ci; PVC-DOP-Fe.0, PVC-DOP-ZnCI, PVC-DOP-ZnO PVC-D0P-AI,0,3H,0 PVC-DOP-AICI,H,0 0.076 0.36 0.46 0.15 0.29 0.17 0.26 35 5.2 57.8 -- --v 4.4 19.8 0.96 30.5 4.8 31.8 1.1 46.8 2.7 40.4 * tn prfiiw rt HHiwf poMtfals to Mpon>>thw wppon-- at ttt opaclmrtw fr oweft cortpowtrt proboHy Mnw at It* atna allIdoncy aa oiibottvo *trtyrt at tamataiMr-- htghar than 300'C, Th# iiWHfli at aadi imuNc artdWva la 1.9 pa^aiH by wrtphi wHh raapart la bw PVC. most efficient additives are iron oxide and zinc compounds. These results suggest that phthalic anhydride can be used as a tracer to estimate the efficiency of any additive as smoke suppressant for the PVC plasticized with DOP. This statement is confirmed in Fig. 5 where it is shown that the smoke level from DOP is in reverse ratio with the amount of phthalic anhydride formed in smolder ing conditions (400C) in absence or in presence of ferrocene Fe55 supplied by ARAPAHOE chemical company. The ferrocene is well-known to reduce the smoke level from rigid PVC through its ability to lead to FeCls and FejOa which can act as an oxidative catalyst for the char residue after dehy drochlorination of the polymer. From the results (Fig. 5) the ferrocene also contributes to decrease the smoke level from DOP for plasticized PVC probably through the same previous by products because it has no influence on the nature and the yield ofthe products ofpure DOP decomposition as on the smoke level from pure DOP in smoldering or in flaming conditions. Moreover ferrocene causes the increase of the char residue during pyrolysis under nitrogen whatever may be the percentage of the plasticizer and the temperature (Fig. 6) prob ably because ofthe influence ofiron chlorides, well known as catalyst for PVC dehydrochlorination and crosslinking. Under air in smoldering conditions ferrocene reduces the char residue of about 75 per cent because of the incandescence initiated by FejOj. Nevertheless the ferrocene efficiency as smoke suppressant for PVC fractions is not very great in smoldering or in flaming conditions be cause ofa dilution effect which may favor its volatil ity and contribute to decrease its concentration in the char residue. These results suggest that it is necessary to combine at least two additives to ob tain the best efficiency as smoke suppressant for plasticized PVC. One additive must be efficient as smoke suppressant for the polymer and the other as smoke suppressant for the plasticizer. For this pur pose the combination of copper compound either with zinc oxide or hydrated alumina appears as the most efficient to reduce the smoke level from each component in smoldering conditions (Table 7). From these results the efficiency of the copper compound as smoke suppressant for PVC is not modified in the presence of an efficient additive as smoke suppressant for DOP such as zinc oxide. Furthermore there is a synergistic effect between copper compound and zinc or aluminium com pounds. The efficiency of these binary systems is PHTHALIC anhyORIDE tools per rots at atyrsatd DOR) Fig. 5. Influence of the ferrocene FeS5 on the phthalic anhy dride formation and the smoke level from DOPfor plasticized PVC in smoldering conditions (400C). (~) PVC--DOP (33.3%); (Z) PVC-DOP 133.3%) ferrocene 1.5%; (A) PVC--DOP (44.4%); <) PVC--DOP 144.4%) ferrocene 1.5%; (it) PVC--DOP 116.6%); (V) PVC--DOP (16.6%) ferrocene 1.5%. Fig. 6. Influence of the ferrocene FeS5 (it) (1.5%) on the char residue formed during the pyrolysis of plasticized PVC (DOP 33.3%) under nitrogen (A). JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 UCC Q37457 187 ' Alain Michel, Alain Sainrat, and Michel Bert Table 7. Influence of Metallic Compound* on the Smoke Level from OOP and PVC In Smoldering Condition* (450C) for Plaetlciied PVC (OOP: 16.6%) Additive* Soor decade min g*1 Save decade min g*1 PVC-DOP PVC-DOP-CuSO, PVC-DOP-ZnO PVC-OOP-ZnO-CuSO, PVC-DOP-ZnCI, PVC-DOP-ZnCl.-CuSO, PVC-DO P-Al,0j3H0 PVC-D0P-AI,0,3Ht0-CuS0. 57.8 40 31.8 19.2 30.5 25.2 46.8 28.8 5.2 0 4.8 0.28 0.26 0 1,1 0 TIM aiMHiU o( MMMc sdOMn M1 hvmtaM oNM raapMI M *w PVC. conserved for higher concentration of DOP but also in flaming conditions (7008C), Table 8, where the most efficient combination is obtained with copper and aluminium compounds. In conclusion this study suggests that the in crease of the smoke level from PVC plasticized with di-2-ethyl-hexyl phthalate with respect to the rigid PVC is caused mainly by interaction of the plasticizer with HC1 evolved from the polymer when the temperature is higher than 200*C. This interaction causes the increase of the yield of phthalic anhydride probably through HC1 as catalyst for DOP decomposition but phthalic anhydride formation parallels the formation of products enabled to increase the smoke level such as phthalic acid. In the presence of metallic com pounds (iron, zinc, aluminium) it is possible to favor the formation of phthalic anhydride with respect to these products wtiich are responsible for the smoke production. Then phthalic anhydride can be used as a tracer to estimate the efficiency ofan additive or Table S. Smoke Level from Plasticized PVC (OOP 16.6%) In Flaming Conditions (700*C) In Presence of Metallic Compound* Additive* S decade min g"' PVC-DOP PVC-DOP-ZnO PVC-DOP-CuSO, PVC-D0P-AI,04H,0 PVC-DOP-AI,0,3H,0-CuSO, 8.3 4.6 7.3 8.5 5.2 Th iuyt *f #*h utototo tiliWIn to 1.1% toy wtogtot wMto yipMt to to* PVC. a combination ofadditives as smoke suppressant for DOP because its smoke level varies inversely with the yield of phthalic anhydride. The best combination to reduce the smoke level from plasticized PVC is obtained with the binary systems based upon copper compound, mainly efficient as smoke suppressant for PVC and either zinc or aluminium compounds, mainly efficient as smoke suppressant for DOP. REFERENCES 1. M. M. O'Mara, ]. Polym. Sci., Al, 8, 1887 (1970). 2. E. A. Boettner, G. Ball, and B. Weiss,/. Appl. Polym. Sci., 13, 377 (1989). 3. E. N. Zilberman,O. D. Strizhakov,and E. M. Perepletchivoa, Platt. Matty., 29 (1965); Chem. Abstracts, 64, 8392 (1966). 4. Von Hannes Von Harpc, E. A. Hammer, and H. H. Oelert, Chemilcer Zeitung, 98, 189(1974). 5. Katsuhiko Saido, Masakichi Satomi, Maknto Kirisawa, Takashi Kuroki, Tahashi Kubo, Takenori Watabe, and Tadashi Ikemura, Yakugaku Zatthi, 97, 479 (1977). 6. E. D. Dickens, Polymer Conference Series, "Advances in Flame and Smoke Retardance of Polymers," University of Detroit (May 1975) (conference not published). 7. L- Lecomte, M. Bert, A. Michel, and A. Guvot,/. Macromol. Sci.-Chem., All, 1467 (1977). 8. M. Bert, A. Michel, and A. Guyot, Fire Ret., 1, -301 (1977/78). ucc 0374 188 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981. VOL. 3, NO. 3 Basic Studies of Smoke Reduction in Rigid Poly(Vinylchloride) ALAIN GUYOT, ALAIN MICHEL, MICHEL BERT, and TRAN VAN HOANG CNRS--Laboratoire dee Materiavx Organiqttes BP 24--69390 VERNAISON-LYON, France A review of recent results in the literature and from our laboratory lead to drive a general picture ofthe combustion or rigid PVC as well as to the mechanism by which the most powerful additive can reduce the production of smoke. It is shown that, in smoldering condition the black smoke comes chiefly from direct volatilization of heavy tar molecules from the decomposing residue. The additives change the degradation process of the PVC by catalysis of the intermolecular crosslinking reactions, which compete with the intramolecular reac tions leading to formation of benzene. Whatever their initial nature, they are at least partly transformed into chlorides and then into oxides during the combustion process. The oxide of Cu, Fe and also Zn are catalysts of the oxidation of the char residue, which may be partially inhibited by phosphorous compounds. The catalytic effects seem less pronounced in Haming conditions, which cause the productions of soots from the initial tars. INTRODUCTION Because of its self-extinguishing character, PVC has became popular in many parts of the build ing and housing industry. However, it suffers for two major drawbacks in case of fire: first the evolu tion of a large amount of HC1 which is produced at rather low temperatures, and second the produc tion of large amount ofblack smoke as well in smol dering as in flaming conditions. The evolution of HC1 can be partly overcome by the addition oflarge amounts of finely divided magnesium or calcium oxide or carbonate; however HC1 is easily con densed with water so that its concentration de creases very rapidly at increasing distance from the fire location. Its more dangerous carrier is the black smoke, soot or tar particles, where it may be con densed as well. So, if we forget the corrosion prob lem caused by droplets of HC1 water solutions, the more serious problem is the one raised by the pro duction of smoke. A large variety ofcompounds have been patented as smoke suppressors for rigid than for plasticized PVC. They include ferrocene (1), iron oxides (2-3), alumina trihydrate (4,5), molybdenum oxide (6, 7), vanadium oxide and acetylacetonate (8), copper cyanide and thiocyanate (9), mixture of iron powder and copper oxide (10), zinc and alkaline metal ferrocyanide (11) various alkaline, alkalin-earth or magnesium oxide (12) and even nickolcene (13) and many other compounds or mixtures. The various mechanisms by which these various additives may be efficient are not clearly estab lished and remain the object of controversy in re cent discussions (14, 15). It is the purpose of this Paper to try to clarify some points by critically re viewing the work carried out in our laboratory (1618) as well as other works recently published (14, 15,19,20,21). We will limit our discussions here to rigid PVC, without any other additives than the smoke suppressors ora very simplified stabilization recipe. Another paper has dealt with PVC plas ticized with dioctyl phthalate (22). It is generally accepted that in the combustion of polymers, the development of heat due to the fire causes the pyrolysis of the polymer which leads to the formation of volatile fuel which bums to give the combustion gas as well as the tar or soot parti cles. So much attention has been paid to the com position of the volatile compounds upon pyrolysis. In the case of PVC, for instance, in addition to the non-combustible hydrochloric acid which is the major product, the main volatile product is ben zene, and Wooley (24) was the first to point out that the origin ofthe smoke is to be found in the produc tion of aromatic compounds during the pyrolysis. We will begin by a discussion of the pyrolysis ex periment and will discuss the combustion experi ments later. PYROLYSIS OF PVC Coupling thermogravimetric analysis (t.g.a.) and gas chromatography (g.c.) allows to show that the pyrolysis of pure PVC takes place in two steps (16, 17,24). The first one corresponds to dehydrochlori nation which is practically complete at 300*C, and is highly endothermic. The elimination of HC1 is accompanied by a significant loss of benzene, which may account for up to 3 percent ofthe weight loss and a very small amount of volatiles such as ethylene and propane. The second step begin at JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 UCC 037459 Iftq Alain Cuyot. Alain Michel. Michel Bert, aid Trwn Van Hoang 350*C and gives a very complex set of volatile com pounds including H,, light aliphatic* such as methane, ethylene, propylene, butene ... alkylaromatics as toluene and many others. The com bination of pyrolysis, g.c., and mass spectrometry give a complete analysis of all these products (23, 25-28). Because of the suspected influence of the benzene and other aromatic compounds on the production of smoke, the mechanism of benzene formation has been studied in detail. Using mix tures of PVC and perdeuterated PVC, 0*Mara (29) has shown that benzene is produced through an intramolecular mechanism. Such conclusion was confirmed recently and extended to pure conju gated products (benzene, styrene, naphthalene, biphenyl, anthracene) by Lattimer and Kroenke (20). These authors showed further that mixed aromatic-aliphatic compounds are formed at least partially via intermolecular mechanisms which may be crosslinking or hydrogen transfer. In our laboratory (30), another indirect proof of the intra molecular mechanism of benzene formation came from the study of the pyrolysis of vinyl-chloride (VC)/methyl methacrylate (MMA) copolymers. In these products the more striking reaction is a lactonization between adjacent co-monomer units which leads to the formation of methyl chloride, according to the reaction scheme; pendant on the polymerization temperature which governs the extent of svndiotacticity (31). Although the major part of the benzene is pro duced during the first step of the pyrolysis, its amount is dependent on kinetic factors. So in time resolved g.c. the total amount is dependent on the heating rate and tends to increase when the heating rate decreases (Table 1). In flash pyrolysis experi ments (16) the amount of benzene is lower (less than 1.5 percent), and further, the relative amount of benzene in the whole production of volatiles decreases from 70 to 37 percent when the nominal pyrolysis temperature increases from 500 to 700C (27). It means that there may be competition be tween the process of benzene formation with other processes like isomerization from cis to trans conformer in the polyene sequence, or crosslinking through intermolecular processes. Obviously the competition is in favor of the intramolecular "ben zene" process when the temperature is moderate. At high temperatures, the other processes are fa vored and during the second step of the pyrolysis a very' small amount of benzene is produced. Most of the additives tried, and chiefly those which show a positive action in the reduction of smoke, exert some influence on the pyrolysis pro cess (16, 17). Generally speaking, they do accelerate the dehydrochlorination reaction and CH, CHt CH, / \/ CH \/ C--CHr-- ii i /\ CHjO O CH, CH, CH, n NCH/ XC--CH^ + CH,C1 ii Between these adjacents diads, the degradation of the inner sequence of either VC or MMA takes place as in pure homopolymer: it has been shown that the amount of benzene produced is exactly proportional to the amount of VC units in se quences larger than 5 units. Taking into account the two extreme units engaged in the lactonization re action, it is clear that the benzene results from se quences of conjugated dienes with at least 3 units. The precise mechanism of benzene formation has been discussed in detail by Starnes (14) who concluded that it results from the intramolecular eyclization of conjugated polyene segments into 1,3-cyclohexadiene moieties, followed by the re moval of polymeric ring substituents through se quential C-C homolysis. It is interesting to notice here that this mechanism involves the scission of the main chain and thus may lead to the formation of higher volatile compounds. It must be pointed out also that the mechanism involves the cis con formation around at least two successive un saturated units; for that reason the formation of benzene is inhibited by a trans-trans conformation of the main-chain as in syndiotactic units and so the amount of benzene produced upon pyrolysis is de- thev reduce the benzene formation. A rather exten sive study of the effect of 33 metal oxides and 23 metal chlorides has been recently published by Iida and Goto (39). They have shown that the more acidic metal compounds, and specially FeCl, and FeCl, drastically increase the amount of aliphaticvolatiles and also mixed aromatics as well as chlori nated aromatics, and decrease in the same way the amount of pure aromatics. The reverse effect was observed for the basic metal compounds. These catalytic effects were interpreted in term of the Table 1. Reduction of Benzene (Percent from Pure PVC) by Selected Addttlvee (1.5% Weight) In PVC Pyrolysle Additive Heating rate 2*C/mln rC/mln CuO Fe.O, FeCl, Ferrocene ZnCI, Sb,0, MoO, non* 10.8 11.2 3.2 10.8 44 100 -- 2.5* PIWPBfmBg IM* BumsS ,e ioo IauIaI uBiaU t- h ---- 26.2 18.3 earn -- -- -- 41.8 3* JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 ucc 037460 Basic Studio of Smoke Reduction m Rigid Poly(Vim/ldtloridt) capacity of the chlorides (eventually formed from reaction of HC1 with the oxides) to promote the addition of HC1 and Cl* onto the polyene chain. The behavior ofcopper and some other compounds of Sn, Bi and Ga, was interpreted in term of the ability of their chloride to dissociate to give Cl*, lida and Goto stated that the mixed aromatics and the chlorinated aromatics come from partially satu rated or chlorinated polyene chains formed after readdition of HC1 or Cl*. We do not agree with that view, at least for the formation of mixed aromatics, which has been shown by Kroenke and Lattimer (20) to result from intermolecular process. We do consider that the effect ofthe acidic chloride and of CuCl* is more likely to be catalysts for these inter molecular processes through Friedel-Craft reac tion, for instance, owing to their Lewis acid charac ter. Nevertheless, the possibility of some readdi tion process must be considered which can be also catalyzed by a charge transfer complex between the polyene and the metal chlorides. Some typical results from our laboratory concern ing the reduction of benzene are reported in Table 1. Again, the influence of the kinetics are to be noted: the slower the heating rate, the smaller the benzene reduction is. However it seems difficult to find a direct correlation between the benzene re duction ability ofan additive with its catalytic effect of one of the several reaction during pyrolysis. De hydrochlorination (DHC) itself seems to be out of consideration if one compares the action of iron compounds which strongly accelerate the DHC process, and that of the copper compounds which do not vary much, both lead to about the same activity in benzene reduction. The zinc derivative, especially ZnCl*, which is known as a very power ful catalyst for both DHC and crosslinking, snows a rather moderate activity in benzene reduction. However it must be pointed out that it might be dangerous to make too much comparison between various additives because their mixing with PVC powder may be very dependent on its initial physi cal state (morphology, grain size ....). Special attention has been paid recently to MoO*. It has been reported to reduce the benzene forma tion (14) and also to accelerate the DHC and im prove the crosslinking during pyrolysis (15). The most recent results of Kroenke and Lattimer (15) show that it enhances the rate of formation of mixed aromatics formed upon intermolecular mechanism. For that reason, the authors suggest that a reductive-coupling mechanism is operative. This mechanism, well documented in the literature for copper compounds, involves first the formation of an organometallic compound upon reaction of an organic radical R8, followed by the reaction with a reactive organic chloride R'Cl. The resulting moiety RR'MeCl where Me is the metal atom (which may have other ligands) undergoes a reduc tive coupling which leads to the formation of a reduced MeCl moiety and a coupling of R and R' to give a R-R' molecule. This mechanism is believed to explain both crosslinkinig and dehydrochlorina tion as well as the reverse chlorination ofpolyenes. It is believed also to explain the synergistic effects between Mo and Cu compounds. As proof of its reality, the authors state that Cun may be reduced to Cu1 and CuO in the presence of PVC, and also that reduction of MoO* has occurred, owing to the f)resence of MoO* and MoC in the residue. Using a aser-microprobe analysis, Lum (19) was able to show that the products of reduction of MoO* by the carbon residue are able to chemisorb benzene and other aromatics and then concluded that Mo com pounds may act through the formation of ir-arene complexes, but Kroenke and Lattimer pointed that such a mechanism cannot explain the catalysis of intermolecular reaction by MoO*. On the other hand, Starnes and Edelson (14) concluded that the ability of MoO* to reduce the yield of benzene is likely to be connected with their ability to function as Lewis acid. By that way it can accelerate the heterolysis of a C-Cl bond by coordination the in termediate Cl" ion in an ionic DHC mechanism, to give a more stable anion, thus enhancing the ioniza tion ofthe system. Lewis acid may also catalyze the isomerization of diene from cis to trans so that the cyclization mechanism leading to benzene is less favorable. Finally they can catalyze intermolecular Diels-Alder cyclization as well as Fried 1-Crafts alkylations. In our opinion, it seems dangerous to extrapolate at high temperature, various mechanisms which have been well established at moderate tempera tures. It is dangerous also to try to explain eveiything with only one kind of mechanism. The only point which is clear is that all the additives which are good smoke suppressors do reduce the benzene formation most probably by catalyzing the various competitive processes which may take place during the first step of the pyrolysis of PVC. COMBUSTION OF RIGID PVC The first studies which dealt with the problem of smoke pointed out two major ideas: the first one was the correlation between the production ofben zene during pyrolysis and the production of smoke during combustion. After the initial suggestion of Wooley (23), it comes from the work by Liebman, et al. (32) comparing the behavior of PVC with chlori nated PVC and polyethylene in the production of benzene, chlorobenzene and smoke. We have found actually a fine correlation between the amount of benzene produced upon time resolved g.c. pyrolysis experiments and the combustion at oOtPC (flaming) with PVC plus various amount of ferrocene as smoke suppressor (16). The second idea, issued from the work of Kracklauer and Sparkes (1), was a correlation between the reduc tion of smoke and the increased char residue. Again we have found a very good correlation (Fig. 1) by comparing both results for pure PVC in smoldering JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 ucc 037461 191 Akm Guyot, Akin Michel. Michel Bert, and Tran Van Hoang CHAR RESIDUE (WEIGHT % I Fig.l. Smoke levelfrom pure PVC (250 mg) os charresidue after the end of smoke wave in smoldering conditions at various temperatures between 300 and 550X1. conditions at various temperatures. These two ideas have remained popular, as, for instance, Starnes and Edelson (14) stated that "benzene combustion is the major source ofsmoke during the burning of the polymer"; on the other hand, Kroenke (21) states that "a large variety of metals can form compounds which are smoke-retarder and act to change the thermal degradation pattern ofthe PVC and promote the formation of char". Other experiments seem to lead to inverse con clusions. So, in one of our experiments in smolder ing conditions (400C) introduction of benzene in the air stream did not increase the amount of smoke; clearly, higher temperatures are necessary to cause benzene dehydrogenation and condensa tion into soot particles as present in black smoke. On the other hand, we have observed (16) that ferrocene causes a decrease of the char residue in combustion experiments carried out at 400DC and higher temperatures, and the best smokesuppressor additives which were oxidation catalysts or precursors of oxidation catalysts cause the char residue to be totally consumed (17). A more careful examination of the literature shows that the reasons for these apparent dis crepancies are to be found in the different protocol of the experimentation. Most of the smoke mea surements have been carried out in the NBS smoke chamber where the smoke is accumulated in a lim ited volume. The thermal stress, or the flame, may be applied for different times. In the Goodrich Smoke-Char Test of Kroenke (21) quenching in ni trogen atmosphere and at room temperature after one minute is done: such conditions are well de signed to keep the amount of char residue high. It was shown to be the case for MoOj and other addi tives, but different conclusions were reported in a real fire test (33). In our laboratory we have chosen to work with a tubular reactor kept at a preselected temperature, under an air stream (in the range 200300 1/h). The sample, pelleted powder, between 15 and 500 mg, is introduced rapidly at a given time. Various devices may be placed following the tube: optical cell for smoke obscuration, infrared analyzer for CO or COf, heated glass-wool filter to retain the soot particles, cold glass-wool filter to condense the tars. A thermocouple placed near the sample allows detection of any endo or exothermic event Finally the experiment may be quenched after a given time by switching the air stream to a nitrogen stream and opening the furnace around the tubular reactor. Experiments have been carried out at various temperatures between 300 and 900C, with different amounts ofPVC (25 to 500 mg) or PVC plus 1.5 percent of different additives. A set of remarks must be made: (1) The quantitative as well as the qualitative results are very dependent upon the weight of the sample: the self-ignition temperature of pure PVC decreases from about 750C to about 500C when the weight increases from a few milligrams to 500 mg (18). The maximum amount of smoke is ob served to shift from 490 to 550C when the weight decreases from 250 to 175 mg. Moreover, the amount ofsmoke is not proportional to the weight of the sample in the whole range of weight As shown in Fig. 2 there are at least two ranges. WEIGHT OF THE SAMPLE (mg) Fig. 2. Smoke levelfrom pure PVC vs the weight ofthe sample. 192 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 f JOC 037462 Basic Studies of Smoke Reduction m Rigid PolyiVinyidilande) These observations may be tentatively explained as follows. A bigger sample gives more volatile fuel so that the ratio Q,/fuel is more favorable for igni tion at lower temperature. Concerning the second observation one may speculate that the tempera ture inside the sample increases more rapidly to nominal temperature in a smaller sample so that the competition between the various processes during pyrolysis is in favor of those occurring at higher temperatures, such as the ratio of aliphatic to aromatic compounds. This first remark may give a first explanation of the fact that it is quite impossible to extrapolate to a real fire on large scale fire experiment the results obtained at a laboratory scale. (2) The dynamic nature of the test chosen allows us to separate a set of successive events, at least if the temperature is low enough, i.e., in smoldering conditions. A convenient temperature is 450*0. The first event is the endothermic dehydrochlorination which causes a delay of about 30 s for the nominal temperature to be reached. A first wave of smoke is evolved during and after that event. The major part of the smoke is produced after, in a second wave with a maximum. The two waves are the more separated when the amount of sample is the smaller. Typical results are shown in Fig. 3. Tentatively it may be suggested to associate each of the smoke waves to each step of the pyrolysis process quoted above. Under 400C the amount of smoke is quite negligible. When the temperature increases the two waves become less and less separated. Above 500*C only one wave is observed. In smoldering conditions, CO and CO* are not detected before 450*C. At that temperature, the production of CO and CO, begins before the second smoke wave, but is observed fora very long time after the end of the smoke production. Both gases are produced in approximately equal molar amount, not dependent on the temperature up to 700C. They both account for between 60 and 70 percent of the carbon. It seems that the major source of these gases is the slow combustion of the char residue. A part may come from the combustion ofvolatile hydrocarbons but we have not checked if that combustion is completed or not. Pyrolysis in air does not change the weight loss significantly after the first step of pyrolysis; the same is true for the benzene production; however the production of aliphatic volatile compounds is significantly de creased by about60percent(16). Recently Lattimer and Kroenke (20) have shown that pyrolysis of PVC in air does not change the nature and the distribu tion of the primary pyrolysis products which are only partly oxidized to a great degree. Most prob ably the same situation is valid for the smoldering conditions and changes only if the temperature is high enough for the self-ignition to take place. (3) The char residue which remains after the end of evolution of smoke (10-12 min), accounts for the major part of the carbon in the PVC at 300C or lower, and decreases at higher temperature to dis appear totally above 550*C (18). Obviously this is because enough time has been given for a slow combustion of the residue in smoldering condi tions. If the experiment is stopped at shorter times, the char residue is more important So, at 450C, quenching after the first smoke wave (4 min) leaves a char residue which accounts for 35 percent of the initial weight of 100 mg. After the second wave (10 min) the value drop to 28 percent At the same time the production ofCO and CO, accounts for 3 and 14 percent of the weight of carbon, respectiv ly. For the same conditions, when full time is given (100 min) the residue disappears totally and CO + CO, account for more than 80 percent of the carbon. As shown in Fig. 1, there in a good correlation between the amount of char residue left by 250 mg of PVC when the smoke emission ceased and the total amount of smoke produced in a range of tem perature between 300 and 550*C. It does not mean that the amount of material evolved as smoke does' correspond to the weight of char missing. The maximum amount of material obtained by both hot and cold filters, observed at 670*C in flaming condi tions (Fig. 4) accounts for 10.7 percent of the initial weight (18). In smoldering conditions, this amount does correspond to about 7 percent between 450 and 550C and the smoke is made exclusively oftars condensed at room temperature. Soot appears only in flaming conditions and tends also to level at 7 percent around 700*C. Some tars remain at the lower temperatures and disappear at about 700C. Fig. 3. Obscuration curve (-------- ) CO,(l)andCO(2) evolved vs time at 450*0 (smoldering conditions) for FVC (100 mg). temperature ra Fig. 4. Weight ofmaterial retained by thefilteras soot(D) or tart (A) for FVC (500 mg) at various temperatures. JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 193 ijCC U'37483 Alem Guyot, Alain Michel, Michel Bert, end Tren Ven Haeng It may be concluded that the tars lead to the soots upon flaming. The exact nature of the tars is not known; they are soluble in many solvents, and from GPC experiments in tetrahydrofuran, their molecular weight is of the order of magnitude of 1000*2000. The pyrolysis process involves not only crosslinking reactions, but also chain scission reac tions so that it may be suggested that the tars may be formed, at least partly, dirough a direct volatiliza tion of the residue after dehydrochlorination. In fact, some heavy products, accounting for about 3.5 percent of the initial polymer weight can be con densed by a cold ring system at the exit of the oven in pyrolysis experiments (16). Of course direct polycondensation of aromatic volatile compounds may be another process, more generally accepted, but its contribution is probably important mostly at high temperatures. It is known that the combustion of aromatics to produce soot needs temperature of the order of 1000*0 (34). Further, it may be noted that the amount of tars condensed on the cold filter represents about the double of all of the hydrocar bons identified. So, at least a large part of the tars comes from larger volatile molecules. INFLUENCE OF THE SMOKE REDUCERS ON THE COMBUSTION OF PVC In most of the work already published, the influence of the smoke suppressor additives has been discussed in terms of their action on the char residue. The various mechanisms by which they are able to modify the amount and the distribution of hydrocarbon pyrolysis products has been dis cussed too, as shown earlier in the present paper. The reason for that was the belief that the smoke comes exclusively from the combustion products of the pyrolyzate. We have just shown that this belief is not correct to describe the smoldering conditions and is only partly correct in flaming conditions. On the other hand, after observation of some incandes cence within the char residue, exothermicity and also increased amount of CO* produced, we have suggested that the major way of action of the best smoke-suppressors additives was through an oxida tion catalysis (17). It has been shown that a list of compounds, or combination of compounds known as catalysts for total oxidation (35), were efficient smoke suppressors at 400C (i.e. smoldering condi tions). Most of them also cause a decrease of char residue to a level sometimes lower than the amount of additive introduced (1.5 percent). However, it might be argued that in our experiments enough time is given to allow a slow combustion ofthe char residue to take place so that the oxidation catalysis effect is mostly a post effect not really connected with the mechanism ofsmoke reduction. So, further experiments have been carried out with a more limited number of additives in order to clarify a number of points. The first observation does concern the observa tion of ignition: the additive seems not to change significantly the ignition temperature, so with a sample of 50 mg the same temperature, 680*C is observed for pure PVC as well as for PVC plus 2.5 percent of FeCl,, FeCIj, CuO although it is 705#C with FejOj and 72(PC with ferrocene; on the other hand, CrjOj, although it is neither very efficient for the reduction of benzene during pyrolysis nor for the reduction of smoke in smoldering or in flaming conditions, tends to cause the self-ignition at a lower temperature (see Table 2). Most of the additives cause the char residue to disappear at rather low temperatures: for instance, with 500 mg of product with 1.5 percent of addi tives, the residue which was 26 percent in the case of pure PVC is quasi null at 400C for all the addi tives except for CrjOs where it is 20 percent and for ZnO (5 percent). For all the others, CuO, Fe*Oj, ferrocene, MoO,, VtO... it is less than the amount of additive so that apart of the additive has been evolved as aerosol. Inis was clearly visible in the case ofCuO where incandescence can be observed on the filter so that the soots are very much reduced (18). Some results concerning the distribution of the aerosols between tars and soot are reported in Table 2. In these experiments with a rather large sample weight, two glass wool filters have been placed after the reactor. It is clear drat CuO does reduce the amount of tars by about 50 percent' whatever the temperature in smoldering condi tions and causes the soot to practically disappear probably because of its catalytic oxidation power: incandescence is observed on the soot filt r in the presence of CuO. On the other hand, Fe,0* and Cr,Oj do not reduce the amount of tars in smolder- Table 2. Influanca of ttia Additive (1.5%) on tba Distribution of Tara and Soot at Different Tamparaturai--Sampla Weight: 500 mg--Air Stream: 214 1/h T(*C) 470 530 505 670 720 800 AddMva norm CuO Fa,Oi Cr,0* none CuO F#,Os Cr.O, non# CuO Fv.O, CfiOi no CuO Fa.O, non# CuO FoOj non# CuO F#fOi Flame no no no no no no no yaa yaa yaa no yaa yaa yaa yaa yaa yaa yaa yaa yaa yaa Char raaidua, i%> 26 1.2 tracaa 1.2 tracaa 1.1 tracaa 1.1 tracaa 0.8 tracaa 0.B Tars, weight % 7.2 4.4 72 5.S 6.8 4.2 6.6 4.7 4.2 3.0 6.1 4.4 3.B 1.4 2.0 0.1 0 1.2 0.1 0 -- Soot, weight % 0 0.S 0.4 0 0 0 0 1.6 2.9 1.1 0 1.5 6.8 0.8 2.2 6.6 oxid. 4.2 6.6 oxid. -- 194 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 Basic Studios of Smoko Reduction m Rigid PolyiVinyldiloride) ing conditions and Fe,0, only is active in flaming conditions. Finally it is confirmed that the presence of soot is to be associated with the presence of flame. Corresponding smoke data are reported in Table 3. A filter for soot has been placed between the reactor and the optical device for measuring the obscuration so that the contribution of the smoke from tars might be estimated. Surprisingly here, although there is no flame, a part of the smoke seems to have been stopped by the heated filter. Possibly a few larger particles of tar have been retained on the filter. The weight of these particles may be negligible but their contribution to the obs curation measurements should have been rather large, thus explaining the apparent reduction of smoke. Nevertheless, all the additive cause a rather large reduction in smoke, and again, CuO is the most active. Some other results are given in Table 4 to compare the effect of CuO with MoOj and Fe,Os. Again, CuO is the most active and it may be con cluded that the efficiency of Mo03 Is comparable to that of FejOj. The last series ofexperiments with 50 mg of PVC was carried out to compare the effect of ferrocene, Fe,0,, FeClj, FeCl, and CuO at various tempera tures between 300 and 710*0 (Table 5). Surpris ingly, although they reduce the amount of smoke above 550*0 as compared with PVC, FeCl, and moreover FeCl, cause a significant amount of smoke to be observed at 300*C. These additives and also CuO and even Fe,0, cause the production of significant amount of CO and CO, at such low temperatures, with a ratio of CO,/CO always larger than 1. (Table 6) Table 3. Smoke Production at 570*C (Non-Flaming) from 50 mg of PVC with 1.5% of Various Additives Smoke total Smoke from tar* Additive None CuO F*,0, V.O, ZnO CriOj 18 4.75 9.5 13.6 17.8 20.5 13.4 4.9 6.8 9 5.2 6.7 Table 4. Smoke Reduction (% of Smoke from Pure PVC) by CuO, MoO, and Fe,0, (14%) at Different Temperature* Temperature, *C 450 471 508 583 593 840 680 MoO, 68 78 59.8 87 87 65.5 65 Fe,0, 57 100.2 64.S 75 84 69.4 40 CuO 12.4 44.7 48.2 58 52.6 49.3 53.6 Table 5. Influence of Vartoua Additives on the Whole Production of CO and CO,, aa Compared with Pure PVC Additive 520%, 175 mg CO + co; coyc 720%, 50 mg CO 4 CO,* CO^CO non* 1 0.84 1.54 1.02 F*tOs 1.52 1.50 1.69 1.19 Ferrocene 1.47 0 0 0 CuO 1.54 2.49 1.38 1.89 v,o. 1.25 2.68 0 0 Cr,0, 1.71 1.20 1.84 1.02 ZnO 1.61 0.90 1.62 1.28 * WiIimw fiimi m bam ** IN nm iswhh ICO + CO pwjmd wm pot* PVC M tire <179 ). Table 7. Molar Ratio COa/CO for Selected Addltlvea at Various Temperatures (Smoldering Conditions) Temperature <*C) 450 471 508 553 593 640 680 Weight (mg) 150 75 75 50 25 25 25 CuO 1.93 2.63 2.34 1.97 1.83 1.15 0.94 Fe.O, 1.78 2.11 1.73 1.62 1.17 0.85 0.79 MoO, 1.87 1.78 1.78 1.47 1.25 1.28 1.22 PVC 0.78 0 0 0 0 0 0 Other results are reported in Table 7 f r CuO, Fe,0, and MoOj at various temperatures. The qual ity of the combustion, as measured by the ratio CO*/CO, is the best for CuO but is optimum at moderate temperature. Surprisingly it decreases significantly by increasing the temperature. Th variation is less important for MoO, which is better, in this respect, at the highest temperatures. How ever, the results may be influenced by the weight of sample which is varied in order to avoid at any time the saturation of the apparatus. Another possible explanation lies in the fact that a part ofthe additive may be swept out as aerosol ana do not work com pletely at the highest temperatures. The kinetic aspects of the process have been studied for the smoke (obscuration) and also for the production of CO or CO,. With pure PVC, two waves ofsmoke have been observed at 450*C with a small sample (Fig. 3). Some results concerning the effect ofa few additives are shown in (Figs. 5 and 6). Figure 5 shows that CuO does reduce drastically both waves, while ZnO totally suppress the first one and reduce significantly the second one. With iron compounds (Fig. 6), the two waves are not separatedand FeCl, is the most efficient additive. A difference can be observed between Fe,Os and Table 5. Amount of Smoko (Decade Min - g") Produced upon Combustion of PVC (50 mg) at Various Temperature* (*C) with i.s% of Additives Additive none FeClj FeCl, Ferrocene CuO 300*C 0 0 3.6 0 0 0 414*C 0 6.75 4.95 0 0 0 470% 0 8.4 6.25 2.9 0 traces Smoldering 530% 6.9 10 6.85 5.6 7.8 2 574% 18.8 11.1 7.01 5.48 9.35 2.7 607% 18 11.1 7.62 8.32 9.35 7.2 628% 22 12.2 7.31 7.92 12.2 8.2 Flaming 671% 710% 12 0 1.52 4.56 5.7 2.84 8.6 4.16 1 3.96 4.98 4.58 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 UCC 037465 104 Alain Cuyot, Alain Michel, Midi*/ Bert, and Tran Van Hoang e i Tint lain) Fig, 5. Obscuration curves at 450*C for pure FVC (200 mg) (-------- ) or in presence of 1.5 wt percent ofeither CuO (----) or ZnO Fig. 7. COt (l) and CO (2) evolvedfrom FVC (100 mg) at 450"C with 1.5 wt percent of CuO. Fig. 6. Obscuration curves at 460*Cfor FVC (50 mg) with IS u>t percent of FeCli (--), FeCt, (..... ), FedO, or iron acetylacetonate (-------). either FeCl,, FeCl,, or the acetlyacetonate; an in duction period is observed with the former com pound. The effect ofthe additive on the production of CO and CO, is even more important Although the total amount of carbon evolved as CO + CO, was increased by a rather small proportion (actually it may approach 100% of the initial carbon con tents), the rate at which it is produced is much more increased. Typical results are shown in Figs. 7 and 8. Most of the gas is produced in a big wave which began at about the same time that the obscuration curve and the process came to the end by about 20 to 30 min at 450C instead of 100 to 150 min for pure PVC. This wave is obviously caused by the catalyzed combustion of the char residue. A part of it, as well as some secondary waves which modulate the major one, may be due to the combus tion of volatiles or of some tars including a part of oxidation catalyst. Fig. 8. CO, (1) and CO (2) evolvedfrom FVC (100 mg) at 45fTC with 1JS wt percent of ZnO. Additional results obtained at 500*C with 100 mg ofPVC show that the catalytic action ofCuO may be partially inhibited by phosphorous compounds which are often used in fire-retardancy of many polymers, including plasticized PVC. The results concerning CO, are shown in Fig. 9. Here, P,0, has been used to represent the phosphor compounds, of which it is the more oxidized and dehydrated form. It does not inhibit totally the catalytic charac ter of CuO, for the kinetics of the CO, production, and also because the CO^CO ratio which is then 0.95 for pure PVC, reaches 2.33 with CuO and is reduced to 1.36 if P,Ot is added to CuO. In Table 8 a set of results dealt with the data obtained after the first wave (4 min) and the s cond wave (10 min) are reported. In the absence of any volatilization except for complete dehydrochlori nation, the residue should have been 41.6 percent for pure PVC or 43.1 percent for PVC plus additive (44.6 percent in the case of CuO + P,0,). During the first wave, with pure PVC a rather large amount ofmaterial has been volatilized and a small part ofit 196 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 ucc 037466 Basic Studies of Smokt Reduction m Rigid PolyCVmylehioride) Fig. 9. COt evolvedfrom FVC (100 mgjatSOO'C without (--- )or in presence of either 1.5 percent CuO (-------- ) or 1.5 percent CuO plus 1.5 percent P>i is fully oxidized. The amount ofvolatilization is not increased by CuO but the volatile material is rather fully oxidized so that the reduction ofsmoke is very large. PjOs causes the volatilization to be quite suppressed, but the amount which reamins is enough to give a comparatively high level ofobscu ration. Ferric oxide causes a high level ofvolatiliza tion, but moderates oxidation so that the obscura tion is not reduced that much. Zinc oxide causes a strong reduction of volatilization and then a strong reduction ofobscuration. About the same reduction is observed with zinc chloride, although it causes much more volatilization and moderates oxidation. Finally, alumina causes only a small change in both volatilization and obscuration. The results are rather different for the second wave. The amount of volatilization is higher and is reduced chiefly by P*0 and also Al,Os, but en hanced by Fe,Oj. Although it causes a very high oxidation, the latter does not reduce the smoke. High levels of oxidation are always observed, tak ing into account the amount of volatilization. Cop per and zinc compounds lead both to strong reduc tion of the smoke, without big difference for oxida tion activity. Alumina again shows a moderate ac tivityjust reducing the volatilization and the obscu ration to the same extent It is interesting also to try to follow the additive themselves. We nave already noted that a part of it may be evolved as an aerosol or even might give a volatile product: that is the case of ferrocene for which only about 20 percent of the iron remain in the solid residue. This compound has been shown to give a monochloro-monocyclopentadienyl vol atile compound (36) and may further react with HC1 to give FeCl, at about 200*C. We have already re ported that, in the presence of air, it leads finally to FetOj which causes incandescence and is a good oxidation catalyst (16). We have carried out a more recent study, the details ofwhich will be published elsewhere (37), to compare the effect ofvarious iron compounds at different times. In that study, larger amount of additive, up to 6%, have been used, and the experiments have been quenched at various times, such as indicated by letters in Fig. 10. The char residues have been studied through Mossbauer spectroscopy. In addition they have been extracted with hot water (80C) for a long time and the chloride ions titrated by coulometry. Whatever is the initial iron additive, FeClt, FeCls, FetOs or acetylacetonate (FeAcAc), the whole iron is finally transformed into FetO*, which is already Table 8. Char Residua Volatilization, Carbon Transformation into CO, and CO and Smoke Laval Variation during Heating Tima at 450*C (Smoldering Conditions) for 100 mg of PVC Tima Additive Residua wt% Total AR wIV Partial AR w%* <WCM C(CO + COJ wt % =AS % Total Partial* dacada - min - g" 4 min (at the and of the first wave of smokes) non* CuO CuO + P,0, Fa.O, ZnO ZnCI, Al,0, 35 36.4 42.9 31.5 41.1 36.1 37.1 -13.4 -12.5 +3.1 -24.3 -1.2 -13.2 -10.6 0.45 1.38 0.30 5.95 0.80 9.85 0.38 2.9 13.3 1.4 9.4 3.1 6.7 5.1 0 -84 -45 -30 -73 -67 -14 10 min (at the end of the second wave of smokes) none CuO CuO FftfOj ZnO ZnCI, Al,0, PjO, 28.2 23.6 37.5 10 23.7 23.2 32 -32.2 -43.2 -9.8 -76 -43.7 -44.2 -23.1 18.8 30.7 12.9 51.7 42.5 31.0 12.3 0.71 14 11.1 1.44 47.7 34.4 0.19 19.5 18.1 1.45 89.9 80.5 3.18 46.9 45.6 1.27 44.7 38.0 0.71 19.5 14.4 0 -96 -63 +9 -71 -92 -37 * AR calculator from tho thoorotfeol dkr vooMuo n*r total MhydrpcHtartoottoo * Portol Aft roprooofiu tho dWtarcim Nowi tho total ehor rooMwo oltar tho oocoo# wavo (10 mta) on4 tho (trot woo* ot omoho (4 mta|* c(4Tmhoinp),artial carbon trowitotmoo ho CO oita CO oomspoodo to tho dtfforooco bohioon tho ootaoo froitoformoO Iota CO, oita CO off*tho SMowd wovo(1> utai)and <ho firm wo*o * AS % npmvmt in* pwwitii, m variattan o( th* tout mat* me *n*f m* imt wav* ( min) a-- altar SM a-- wav* (is --a). JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3, NO. 3 ucc 037487 197 Akin Guyot, Akin Michel, Michel Bert, and Tran Van Hoang Fig. 10. Obicuration curve (--------j, COtll) and CO (2) evolved from PVC (750 mg) at 450V in the presence of6 wt percent of FciO>. Quenching of the experiments have been carried out at time indicated by the letters A, B, C, D, E, F for Mossbauer analysis. detected through its characteristic six peaks spec trum as shown in Fig. 11, after about 60 s in the case of FeAcAc for instance. Intermediate production of a chloride is proved also by the Mossbauer spectmm, and specific coloration test, such as with phenol, potassium thiocyanate, or potassium ferrocyanide, show the presence ofFe1+. Except for FeCl* which may be easily oxidized to a mixture of FeCl* and Fe*Os, the amount ofFeCl,, as titrated by coulometry, is limited; with FeAcAc the maximum is observed as 32 percent of Fe after about 60 s, and with Fe*0* it remains less than 20 percent of Fe reacted to give FeCl*. Of these additives, Fe,0* is the less efficient to reduce the smoke although it is the only one initially in the state of an oxidation catalyst; it may be noted a long delay before the evolution of smoke as well as for the production of CO and CO*. Such a delay is not observed with the other compounds which lead to FeCl3 sooner and in a larger amount The delay is not observed also in the case ofCuO but it is known (38) that CuO reacts readily with HC1 to give CuCl*. So it is suggested that these chlorides of transition metals play an important part in the reduction ofsmoke. According to the recent work of Iida and Goto (39), FeCl*, FeCl* and CuCl* are among the strongest catalyst for the degradation of PVC. They cause a rather large increase of the production of aliphatic prod ucts during the pyrolysis. On the other hand it is mm(s-l) Fig. 11. Mossbauer spectrum taken from the residue of the experiment quenched at the point F of Fig. 10. clear that the production of smoke level drops offas soon as Fe*0* does appear in the residue in the case of (FeAcAc) or of the chlorides. So both mecha nisms of catalytic modification of the degradation process, and catalytic oxidation of the residue are operative in the reduction of smoke; the former mechanism causes enhancement of the production of aliphatic volatiles (instead of aromatic com pounds) which are oxidized to give CO and CO*. If this mechanism is less efficient, as in the case of Fe*Oj, a larger amount of material is evolved with out total oxidation and thus gives the tars which produce obscuration. The oxidation catalysis does operate in the solid phase where it causes a direct oxidation of the char to CO* and CO. The oxidation also involves the volatile compounds at the very time of their forma tion. It is interesting to note that the initial sample of FejOj is not very active as an oxidation catalyst, possibly because ofthe presence ofa part of Fe*0; it becomes more active later and also, as shown by the Mossbauer analysis, it gives a better crystalline organization. CONCLUSION A general picture of the combustion process of rigid PVC might be as follows: the thermal degrada tion of PVC causes the production ofvolatile mate rial in two successive steps more or less separated according to the temperature and amount of mate rial. The first step does correspond to the en dothermic dehydrochlorination (DHC) and is accompanied with the production of most of the benzene and also some aliphatic light material. The DHC process may be accelerated by various addi tives, mostly chlorides, which are able to catalyze the formation of aliphatic volatiles and also the crosslinking process, so that the amount ofbenzene is reduced drastically. The second step of the deg radation gives more aliphatic volatiles and also produces heavier mixed aromatic compounds which are condensed as tars. Some additives are again able to enhance the production of aliphatic volatiles and reduce the production of tars during that second step. In the presence of air, the alipha tic volatiles are easily oxidized to CO and CO* but the main pyrolysis process is not affected other wise. In smoldering conditions, the black smok is mainly formed from these tars coming from the decomposition of the residue left after dehydrochlortnation with, possibly, a small contribution of the combustion products of the volatiles. After the production of smoke has ceased, a char residu remains which is slowly oxidized to give CO and CO*. Most of the smoke suppressor additives are oxidation catalysts which accelerate this combus tion process ofthe residue which is some cases may be as rapid as the production of smoke. They also increase the ratio CO* and CO. If the temperature and and amount of material are high enough, self ignition takes place. Then the total amount of 198 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1981, VOL. 3. NO. 3 ucc 037468 8ask Studies of Smote Reduction in Rigid PolyfVmylchlohde) smoke is reduced, the tars disappear, but soot is formed and more CO and CO* is produced; the additives do not change the conditions of self ignition. They again accelerate all the processes but their effect is not so dramatic as in smoldering conditions. Whatever their initial nature, the additives react with the HC1 formed upon DHC to give chlorides which are strong Lewis acid catalysts and are able to modify the degradation process. They favor the decomposition ofthe residue into light volatile eas ily oxidizable compounds, instead of the heavy tars which do constitute the actual smoke, either as such in smoldering conditions or after transformation into soot in flaming conditions. Further they are oxidized to oxides which are the oxidation catalyst. Depending on the nature of the metal, these processes are more or less complete and rapid. For such reason the additive may show the optimum of their activity at different tempera ture. So copper is specially active at low tempera ture and also zinc and some iron compounds as FeClj. But other compounds of iron (FetOa) and mostly molybdenum and vanadium seem to be use ful only at higher temperature. Moreover, some compounds or iron may cause the smoke to appear a lower temperature than for pure PVC and then lead to an increase of the smoke up to rather high tem perature. They become actually smoke suppressors only near the ignition temperature or in flaming conditions. REFERENCES 1. J. Kracklauer and C. J. Sparkes, SPE Tech. Papers, 20,616 (1974). 2. D- M. Florence (Armstrong Cork Co) French Fat. 2.209.793 (1973). 3. S- Hosokawa. T- Honda, and H. Matsui (Sekisiu Chem.) Japan Kokai, 74-77968 (1974). 4. P. V. Bonsignore and J. H. Manhart, 29th SPI Reinforced Plastics Composites Sect. 23, p. 1 (Feb. 1974). 5. J. Sobolev and E. A. Woycheshin, /, Fire Flam. Fire Betardant Chem., 1, 13(1974). 6. Mitchell and C. Lawrence (Ethyl Co.) US Patent 3.845.001 (1974). 7. A. W, Armour and E. R. Braithwaite, SCI Conf., Loughborough (Jan. 1976). 8. Mitchell and C. Lawrence (Ethyl Co.) US Patent 3.846.372 (1974). 9.M. Crowe and W. Arthur (B.F. Goodrich) US Patent 3.819577(1974). 10. Mitchell and C. Lawrence (Ethyl Co) US Patent 3.621.151 (1974). 11. M. Crowe and W. Arthur (B.F. Coodrich) US Patent 3.822.234 (1974). 12. T. C. Mathis and A. W. Morgan (Monsanto) Cer. Pat. 2.363311 (1974). 13. Y. Toyosaki, J. Natsume, and F. lida (Nippon Zeon) Japan Kokai 74.07349(1974). 14. W. H. Starnes and D. Edelion, Macromolecules, 12, 797 (1979). 15. W. J. Kroenke and R. F- Lattimer, J. Appl. Polym. Sci. (in press). 16. L. Lecomte, M. Bert, A. Michel, and A. Cuyot,/. Macromol. Sci.-Chem., 11, 1467 (1977). 17. M. Bert. A. Michel, and A. Cuyot, Fire Bee., 1,301 (1977). 18. A. Michel, M. Bert, A. Cuyot, and J. Lahaye, /. Fire Flam mability, 12.72(1981). 19. R. M. Lum. ]. Appl. Polym. Set.. 23. 1247 (1979). 20. R. P. Lattimer and W. J. Kroenke, /. Appl. Polym. Sci., 25, 101 (1980). 21. W. J. Kroenke, J. Appl. Polym. Sci. (in press). 22. A. Michel, M. Bert, and A. Sainmt, Vinyl Technal. 3, 182 (1981). 23. W. D. Wooley, Brit. Polym. /., 3, 186 (1971). 24. P. Burille, M. Bert, A. Michel, and A. Cuyot,/. Polym. Sci., Polymer Lett., 16, 181 (1978). 25. R. R. Stromberg. J. Polym. Sci,, 28. 537 (1958). 26. D. NoSx, W. Benz, and W. P&b, Z. Anal. Chem, 235,121 (1968). 27. M. O'Man, /. Polym. Sci. Al, 8. 1887 (1970). 28. T. lida, M. Nakanishi, and K. Goto. /, Polym. Sci., Polym, Chem. Ed,, 12, 737(1974). 29. M. O'Mara, Pure Appl. Chem., 49, 649 (1977). 30. J. Cuillot, M. Bert, J. Visile and A. Cuyot, Eur. Polym. /, 8, 641 (1972). 31. E. P. Chang and R. Salovey, /. Polym. Sci., Polym. Chem. Ed., 12.2927(1974). 32. E. J. Quinn, D. H. Ahlstrom, and S. A. Liebman, Polymer Preprints, p. 1022, ACS Meeting Chicago (1973). 33. D. Edelson, V. J. Kuck, R. M. Lum; E. Scalco, W. H. Starnes, and S. Kaufman, Combusion Flame (in press); presented at the Technical Meeting of the Combusion Institute, Kings- tone (1979). 34. G. Prado and L. Lahaye,/. Chim. Phys,, 11-12,1678 (1973). 35- J. E- Geramin "Catalytic Conversion ofHydrocarbons," Ch. V. p. 231, Academic Press. New York (1969). 36. Lawson (private communication) 37. Tran Van Hoang, A. Michel, M. Bert, A. Cuyot, and P. Bussiere, /. Appl. Polym. Sci. (in press). 38- P. Pascal "Traiti de Chimie Minirale," Tome III, p. 225- 255 Ed. Masson, Paris (1957). 39. T. lida and K. Goto, /. Polym. Sci., Polym. Chem. Ed., 15, 2427-2435 (1977). JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1961, VOL. 3, NO. 3 ucc 037469 199 -