Fire and Polymers II - American Chemical Society

Chapter 24

Smoke Production from Advanced Composite Materials Downloaded via UNIV OF CALIFORNIA SANTA BARBARA on July 11, 2018 at 12:23:07 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

1

2

3

D. J. Caldwell , K. J. Kuhlmann , and J. A. Roop 1

U.S. Army Medical Research Detachment, Wright-Patterson Air Force Base, OH 45433-7400 ManTech Environment, Inc., Dayton, OH 45431-0009 Air Force Institute of Technology, Wright-Patterson Air Force Base, OH 45433-7765 2

3

The use of advanced composite materials (ACM) in the B-2 bomber, composite armored vehicle, and F-22 advanced tactical fighter has rekindled interest concerning the health risks of burned or burning ACM. The objective of this work was to determine smoke production from burning ACM. A commercial version of the UPITT II combustion toxicity method developed at the University of Pittsburgh, and refined through a US Army-funded basic research project, was used to establish controlled combustion conditions which were selected to evaluate real-world exposure scenarios. Production and yield of toxic species varied with the combustion conditions. Previous work with this method showed that the combustion conditions directly influenced the toxicity of the decomposition products from a variety of materials. Introduced in the 1960s, advanced composite materials (ACM) are expected to compose 40-60 percent of future airframes. Figure 1 illustrates the increased use of ACM in US Air Force aircraft. During the 1990s, several events focused attention on the human and environmental consequences resulting from fabrication and incidental combustion of ACM. In addition, although the fibers and epoxy resins of advanced composites appear to be safe in their original state, the chemical transformation to a hazardous substance during combustion is not well characterized. These resins, such as epoxies, polymides, phenolics, thermosets, and thermoplastics, may release potentially lethal gases, vapors, or particles into the atmosphere when burned. As the uses of composites increase, so do the potential risks to the environment and those exposed to the smoke and combustion gases during aircraft mishaps. The objective of this work was to determine smoke production from burning ACM and predict its toxicity.

0097-6156/95/0599-0366$12.00/0 © 1995 American Chemical Society Nelson; Fire and Polymers II ACS Symposium Series; American Chemical Society: Washington, DC, 1995.

24. CALDWELL ET AL.

Smoke Production from ACM

367

Figure 1. Percent of ACM (by weight) for selected U.S. Air Force aircraft.

The apparatus used to establish controlled combustion conditions is a commercially available version of the cone heater combustion module of the UPITT Π method developed at the University of Pittsburgh (1). Previous work with this method showed that the combustion conditions directly influenced the toxicity of the smoke from a variety of materials (2). The toxic potency of the ACM smoke can be estimated from experiments conducted over a variety of combustion conditions selected to enable evaluation of real-world exposure scenarios. The smoke concentration can be calculated and used to predict toxicity (2). This prediction can be subsequently validated by bioassay to detect the presence not only of unusual or uncommon toxicants but also of biological interactions between common gases. If a further refinement is desired, the timeto-toxic effect can also be determined (2,3). Experimental Materials. A graphite fiber/modified bismaleimide resin ACM (approximate 2:1 ratio by weight) was used in these studies. Coupons were fabricated with epoxy/graphite skins; woven graphite/epoxy, MS-240; Hercules AG380-5H/8552 Prepreg; Chemmat 4011, 8-ply; thickness: 0.125 inches. The ACM coupons were 110 mm square (approximately 0.01m ) by 2.5 mm thick. The average mass of sixteen coupons was 54.60 + 0.18 g. 2

Combustion Module. A commercial version of the UPITT Π combustion toxicity apparatus (1) was used to establish controlled combustion conditions. This apparatus consists of a truncated cone heater, as used in the cone calorimeter, to irradiate specimens at selected, controlled heat flux levels; a balance to determine the mass loss rate of the burning specimen, and an enclosure so the ventilation (airflow) can be accurately measured and controlled. This is the first combustion system to combine the three essential elements necessary to describe the burning conditions to which a specimen is submitted: the heatflux,ventilation level, and mass loss rate. With this apparatus, the radiant heat flux level and

Nelson; Fire and Polymers II ACS Symposium Series; American Chemical Society: Washington, DC, 1995.

368

FIRE AND P O L Y M E R S II

ventilation level can be varied, and thus, well defined burning conditions can be investigated over a wide scale. A post-crash conflagration would have a maximum external heat flux, from a JP-4 fuel fire, of approximately 84 kW/m (i.e., 873.5°C) (4). For these experiments the sample was horizontal, heatflux(Q) was set at 38, 57, or 84 kW/m and the airflow (V) was 19.6, 28.7, 35.6, 41.5, or 51.3 L/min. The time to ignition (T ), percent mass lost, and mass loss rate (m ) were determined as previously described (1) except that a 10-minute period was used instead of a 30-minute period. The resulting smoke concentration (SC) was calculated by dividing the mass loss rate by the airflow through the apparatus. 2

2

ign

Thermogravimetric Analysis. A Perkin-Elmer TGA7 was used to conduct thermogravimetric analysis of ACM specimens isothermally at 650°C, 770°C and 950°C. Combustion Product Identification. A Perkin-Elmer Model 1650 FT-IR spectrometer was used to obtain transmission spectra of the filtered smoke produced by the burning ACM coupon. A Perkin-Elmer Q-Mass 910 GC/MS was used to analyze extracts prepared from cold-trapped vapor and the soot residue. Screening Electron Microscopy. Burned material was allowed to directly settle on SEM stubs that were prepared with adhesive tabs. SEM stubs were dried over night in a vacuum desiccator, coated with a 100-A layer of gold and examined in an Amray 1000B scanning the electron microscope equipped with a standard tungsten filament at 30 kV accelerating voltage. The working distance was 8 to 18mm. Results Time To Ignition And Mass Loss Rate. Results from sixteen experiments conducted under flaming conditions are presented in Tables I through HI. The Tign decreased as Q increased, while the mass loss rate increased with increasing Q. Table I. Time to ignition, percent mass lost, mass loss rate, and smoke concentration for 0.01m ACM specimens irradiated for 10 minutes at 38.5 kW/m and indicated airflow 2

2

Sample No. 22 10 14 17 27 Mean Std. Dev.

V (L/min) 19.6 28.7 35.6 41.5 51.3 -

T

(sec) 120 110 94 69 97 98.0 23.16

ign

Mass Lost % 21.5 24.2 23.4 23.2 23.5 23.16 1.00

m (g/min) SC (g/L) 0.060 1.180 0.046 1.323 0.036 1.273 0.030 1.262 0.025 1.290 1.266 0.053


24.

369


C A L D W E L L ET AL.

Table Π. Time to ignition, percent mass lost, mass loss rate, and smoke concentration for 0.01m ACM specimens irradiated for 10 minutes at 57.2 kW/m and indicated airflow 2

2

Sample No. 9 5 12 15 18 28 Mean Std. Dev.

V (L/min) 19.6 28.7 28.7 35.6 41.5 51.3 -

T „(sec) 45 50 47 39 34 35 41.7 6.6 iB

Mass Lost % 30.0 28.1 27.7 26.6 28.3 29.0 28.28 1.15

m (g/min) SC (g/L) 0.083 1.632 0.054 1.536 0.053 1.514 0.041 1.459 1.541 0.037 0.031 1.595 1.546 0.061

Table ΙΠ. Time to ignition, percent mass lost, mass loss rate, and smoke concentration for 0.01m ACM specimens irradiated for 10 minutes at 84.2 kW/m and indicated airflow 2

2

Sample No. 3 13 16 19 29 Mean Std. Dev.

V (L/min) 19.6 28.7 35.6 41.5 51.3 -

T „(sec) 30 34 32 18 16 26.0 8.4 ie

Mass Lost % 30.8 29.6 30.5 31.7 32.0 30.92 0.96

m (g/min) SC (g/L) 0.086 1.676 0.056 1.609 0.047 1.666 0.042 1.728 0.034 1.751 1.686 0.056

Consolidated results found the average mass loss of the ACM coupons to be 27.5 ± 3.4%. Further review of the results from these experiments under controlled conditions demonstrated that the mass loss rate increased with heat flux but was independent of airflow. The graphical presentation of these data are found in Figures 2 and 3. TGA Data. The TGA has the capability to heat a sample in a nitrogen atmosphere or in air. The atmosphere made a significant difference in the mass loss characteristics of the ACM. TGA In Nitrogen: After the initial mass loss due to the polymer pyrolysis, the mass stabilized at a little over 75% of the initial mass, and stayed there for the rest of the thirty minute test run. There was no significant change with extended time.


370

FIRE AND POLYMERS II

Figure 3. Coupon mass loss rate vs. airflow


24.

CALDWELL ET AL.

371


TGA In Air: The specimen mass did not stabilized after the initial pyrolysis mass loss (due to the polymer resin loss). The mass loss curve changed with temperature, the slope of which increased as temperature increased. Given enough time, the graphite fibers completely disappeared, i.e., at 950°C all mass was lost within 25 minutes, while at 650°C the time required increased to over 60 minutes. Mass versus time curves for data presented in Table IV are illustrated in Figure 4. Smoke And Aerosol Characterization. The composition of the smoke and properties of the aerosol particles were evaluated. Initial evaluation indicated the smoke was composed of phenol groups, aniline groups, carbon monoxide, and carbon dioxide. Major spectrum peaks from an FT-IR spectrum of smoke from an experiment conducted at 57 kW/m and 28.7L/min are identified in Table V. 2

Table IV. Isothermal TGA of ACM specimens under nitrogen or air atmosphere at indicated temperature Sample Number

Atmosphere

9 11 20 28

N N Air Air

Temperature (°C) 770° 770° 950° 770°

2

2

Initial Mass (g) 0.0422 0.0632 0.0811 0.0765

770°C

(N )

770°C

(N )

2

2

7 7 0 ° C (Air)

9 5 0 ° C (Air)

0

5

10

15

20

25

30

35

Time (min)

Figure 4. Percent mass remaining vs time for TGA in air and nitrogen at indicated temperature


40

372

FIRE AND POLYMERS II

Table V. Peak Identification From Representative FT-IR Spectrum cm-1 3708 3628 3596 3566 3324 2510 2174 2116 1526 1304 1164 1138 730

Identity Aniline Aniline Aniline Aniline Phenol Carbon Dioxide Carbon Monoxide Carbon Monoxide Aniline Aniline Phenol Phenol Phenol

Height 17.64 19.23 25.35 53.31 53.36 81.09 50.68 56.36 10.45 29.53 47.28 53.65 10.25

The particle density was determined using standard laboratory practice and found to be 0.2879 mg/mL. Air samples analyzed by electron microscopy identified a range of aerosol diameters from 0.5 to 1.5μιη. Given this density and the observed range of particle diameters the gravitational settling velocity was calculated to be from 6.5 χ 10" to 3.8 χ 10" m/sec. 6

4

GC/MS Data. Two samples were analyzed by GC/MS; the first being combustion vapor condensed in a cold trap, and the second being a solvent extract of the soot. The quantitation amounts were obtained by using the Response Factor = 1 approximation as specified in the USEPA Contract Laboratory Program Statement of Work for Organics Analysis; Multimedia, Multiconcentration (5). The normal method of quantification used in GC/MS analysis is based on comparison of the area of the most prominent ion in a known concentration of the material to that of the most prominent ion of an internal standard and developing a ratio known as the response factor. As over 90 compounds were tentatively identified in the extract from the soot, it can be seen that preparation of this many standards is impractical for several reasons, including time and expense. An alternative method of quantification, namely that specified in the EPA Contract Laboratory Program Statement of Work, was chosen. This method of quantification assumes that the response factor is 1 and the total areas, not just those for the most prominent ions, are used. While not extremely accurate, this quantitation will give a general feel for the amount of material present in the sample. Based on over eight years of environmental laboratory experience in a production setting, the numbers given by this technique are probably low (K.J.K., personal observation). Most of the internal standards


24.

C A L D W E L L E T AL.


373

are deuterated PAHs with strong, sharp responses. In general, most other compounds do not respond as well, which would give a lower concentration by this method than if one were basing calculations on actual, measured response factors. In the first sample a vapor aliquot was collected using a cold trap with the results summarized below in Table VI. The experimental conditions were heat flux of 57 kW/m and airflow of 28.7L/min. 2

Table VI. Quantitation of Identified Compounds by GC/MS Compound Aniline Phenol 4-Methylphenol (o-cresol) 2-Methylphenol (p-cresol) 3-Methy 1-1 -isocyanobenzene Quinoline Biphenyl Diphenyl Ether (diphenyl oxide)

Air Cone μβ/ηι3 0.57 1.60 0.11 0.12 0.01 0.04 0.01 0.19

In the second sample approximately 1.4g of the soot was extracted with 50:50 Methylene Chloride:Acetone solution. It was apparent, upon examination of the injection port liner, that many of the extracted compounds were not suitable for analysis by GC/MS as there was obvious evidence of pyrolysis and deposition in the injection port liner. We identified over 90 different compounds in the extract. The major compounds (i.e., >lg/kg) that were identified are shown in Table VII below. The compounds were identified by GC/MS, using the most recent version of the NIST mass spectral library. All matches were inspected by an experienced analyst, and the best matches were selected by the analyst rather than only relying on the computer algorithm. GC/IR was performed on the same samples, and the data was used in a confirmatory way. There are very few vapors phase libraries for IR currently available, and most of the compounds were not locatable. Several other comments need to be made. As these identifications were made by mass spectroscopy, it is quite probable that some of the compounds identified may be a different isomer. That is one of the weaknesses of the technique. There were a number of PAH peaks in the soot extract which were of too low an intensity to characterize properly, and were not included in Table VII. Additionally, as mentioned above, there are probably many compounds which either did not extract in the first place or did not make it out of the GC injection port. The key point is that several of these fourteen compounds, those annotated with (C) in Table V u , are known to be carcinogens (6). Cancer biassays have not been performed for most of the remaining identified compounds.


374

FIRE A N D P O L Y M E R S II

Table V u . Identification and Quantitation by GS/MS of Major Compounds Extracted from Soot Cone, in Soot g/kg 2.99 2.18 1.20 3.48 1.20 1.05 1.66 2.21 2.21 1.36 1.66 1.70 1.29 2.13

Compound Aniline (C) Phenol 2- and 3-Methylaniline Quinoline 5-Methylquinoline Diphenylether 2-Methoxyethoxybenzene l,2-Dihydro-2,2,4-trimethylquinoline 1 -Isocy anonaphthalene Dibenzofuran (C) 1 -Isocyanonaphthalene Anthracene (C) N-Hydroxymethylcarbazole Fluoranthene (C)

Discussion A C M Mass Loss Rate. The primary objective of these combustion experiments was to obtain a mass loss rate for the ACM to be used to model atmospheric dispersion and deposition of smoke from a burning aircraft. However, one significant limitation to this study was the lack of research on the heat transfer properties of this ACM. Therefore, we assumed that the flame spread characteristics demonstrated by this bench-scale combustion equipment accurately simulated those of a full-scale aircraft. To test this hypothesis, experiments were repeated with ACM coupons of different surface area. The results, under controlled heat flux and airflow conditions, identified a linear relationship between the mass loss rate and the area of the burning composite (7). (The results from experiments conducted with ACM coupons of different surface areas are not presented here.) Regression of the mass loss rate data with the sample coupon area was performed. The equation for the regression line is found below in Equation 1. m = β, (Area) + β (HeatFlux) - 0.01 where: β, =1.98 2

β =1.86x10^ 2


(1)

24.

CALDWELL ET AL.

375

Smoke Production from ACM 2

The regression results provided a linear equation (R = 0.999) that allowed accurate prediction of an emission rate for a full-scale aircraft. These findings enable regression analysis of a linear equation for the emission rate given constant heat flux, airflow, and area conditions. Aerosol properties were identified which enabled calculation of the gravitational settling velocity. This, in turn, will serve to better estimate the downwind plume characteristics. The combined results allow for accurately modeling the smoke and aerosol smoke plum generated during the combustion of composite material aircraft. Comparison of UPITT II with TGA. The manner in which energy was applied to the samples in the TGA was different than that in the UPITT Π apparatus. The TGA uses a cup design with the sample in the center of a small furnace. It wasn't possible to shield the sample from the heat during the heatup cycle, as was possible in the UPITT Π apparatus. Despite this difference, and the much smaller sample size used in the TGA, the mass loss measured by the two units was remarkably similar. During the first two to three minutes of TGA, the sample lost approximately 25% of its mass. What happened in the TGA after this time period was a direct result of the differences between the TGA and the UPITT Π combustion paradigms. We suspect that the graphite fiber was being "eroded" by the oxygen in the air. Unpublished work on diamond showed a molecular surface effect (R. E. Langford, personal communication). Apparently, when the material absorbs enough energy (heated), the impact of an oxygen molecule on the surface is enough to pull off a carbon atom and form C 0 or CO. This hypothesis is supported by infrared spectroscopy data, which show continuous evolution of these gases until the mass goes to zero. 2

Other Observations From The TGA Experiments. Once the resin was pyrolysed off the graphite fiber matrix, the fibers separated and puffed to several times the original volume and lost any cohesion or tendency to group together. After an experiment where the graphite was not completely consumed, there was considerable difficulty in cleaning the instrument. This phenomenon happened whether the experiment was conducted in nitrogen or air. The fibers are extremely fluffy, and potentially electrically conductive. Therefore, they could travel a significant distance in a mild breeze, and have the potential to short out electrical equipment from computers to power lines. In general, even after a long experiment, the fibers remained visible and were therefore not respirable. It is, however, possible that some fibers were being eroded to the point where they could be respirable. At the present time, the answer is unknown, however, there are clear hazards associated with what is known to be contained in the soot particles and that these present the greater hazard to life and property than does the physical shape (i.e., particle or fiber). It is critical that measures be taken in fighting a fire involving these materials to reduce dust and aerosols.


376

FIRE AND P O L Y M E R S II

Predicting Smoke Toxicity. As described above, the smoke concentration (in g/L) can be calculated by dividing the mass loss rate (in g/min) by the airflow through UPITT Π apparatus (in L/min). For the experiment conducted at a heat flux of 57 kW/m with an airflow of 28.7 L/min, taking the average mass loss rate at that Vof 1.525g/min from Table Π, the resulting smoke concentration was calculated to be 0.053 g/L (53 mg/L). Exposure to this concentration of smoke is expected to be lethal since it is nearly twice the LC50 previously determined for a variety of synthetic materials ( e.g., rigid polyurethane foam, plasticized polyvinyl chloride, etc.) (4). This statement is supported by the FT-IR identification in smoke and the GC/MS quantifications of toxic species in the cold trapped vapor from this experiment as identified in Tables VI and VII. 2

Conclusions The mass loss rate increased with heat flux, but changes in airflow had no effect on mass loss rate. The total mass loss determined by TGA was comparable to that determined by the UPITT Π apparatus. The particle density and aerosol diameters of the soot particles were measured. Combustion products quantitated by GC/MS included oxygenated compounds. Huffy graphite fibers were observed in TGA studies, but none were collected on filter media during testing with the UPITT Π apparatus. Small specimen size may have contributed to this phenomenon for the TGA findings. Smoke production and yield of toxic species from ACM varied with the combustion conditions. This finding is consistent with previous work with this method which showed that the combustion conditions directly influenced the yield and toxicity of smoke produced by a variety of materials (2). Future work will incorporate animal exposures to determine the toxic potency of the smoke and evaluate alternate non-lethal endpoints such as incapacitation. In addition to toxic potency in terms of smoke concentration, we can also determine the time to effect, i.e. lethality or incapacitation. This approach will result in the selection of safer advanced composite materials for new and existing weapons systems.

References 1. D. J. Caldwell and Y. C. Alarie, J. Fire Science, 8, 23-62 (1990). 2. 3. 4. 5.

D. J. Caldwell and Y. C. Alarie, J. Fire Science, 9, 470-518 (1991). D. J. Caldwell, Toxicology Letters, 68, 241-249 (1993). D. D. Drysdale, An Introduction to Fire Dynamics (1985). USEPA, Document No. OLMO1.8, (Aug, 1991).

6. Ν. I. Sax, Dangerous Properties of Industrial Materials (1989). 7. J. A. Roop, Modeling Aerosol DispersionfromCombustion of Advanced Composite Materials During an Aircraft Mishap, Master's Degree Thesis (AFIT/GEEM/ENV/94S-21 ), Air Force Institute of Technology (1994). RECEIVED November 28, 1994


Fire and Polymers II - American Chemical Society

Recommend Documents