The Decomposition Behavior of Thermoset Carbon Fiber Epoxy Composites in the Fire Environment

Carbon fiber composite materials are increasingly being used in the design and fabrication of transportation vehicles. In particular, the aviation industry is increasing transitioning from metals to this class of composites due to the high strength and low weight of the materials. Most aviation structural composites are thermoset, meaning they require thermal processing to harden the epoxy. In the event of a fire, they will behave significantly different than the metals they replace. Because they are not homogeneous, they also differ significantly from homogeneous solid combustibles. Sandia National Laboratories is motivated to study burning composites because we maintain experimental and modeling capabilities for assessing transportation safety. Understanding the thermal environment created by transportation fires is therefore paramount. This type of focus is not typical of the general literature on these materials in the fire environment. A serious issue with the majority of fire performance data found in the open literature is that the length and mass scales are generally orders of magnitude below those used in vehicle design. With a non-traditional perspective on composite fires, Sandia has performed several test series. Together with a review of the work from other institutions as found in the literature, this report presents a phenomenological overview of the relevant work on the behavior of composite materials in a fire environment. Introduction Carbon fiber composites are being used extensively in the design and manufacture of modern aircraft. The low weight coupled with high strength and durability of the materials makes them desirable for extensive use throughout the aircraft. Some general information on composites is available in the literature [1,2]. These composites generally consist of a binder and a fiber. The binder is usually a resin or glue that will set once the part is formed. The binder provides rigidity and strength to the composite matrix. Two of the more common classes of binders used in airframes are thermoset epoxies and bismaleimide resins. Bismaleimides in particular exhibit improved resistance to thermal damage under normal aviation environments. Strength, weight, and thermal resistance are three of the most significant considerations in selection of these materials. From the perspective of a fire involving these materials, the chemical constituency of the binder is important, but often difficult to understand in detail because the exact formula of each grade of epoxy is a differentiating factor in the manufacturing process. These are proprietary, and not normally divulged. Epoxy resins often have a fundamental resin repeating unit of [C18H20O3]n Combustion Institute Joint US Sections Meeting May 19-22, 2013 Paper 070FR-0184 Page 2 of 18 (diglycidyl ether of Bisphenol A, or DGEBA), and hardeners often consist of compounds with amine groups that are the active unit that promotes cross-linking of the polymer. Aromatic polyamine hardeners do not set at room temperatures, and are therefore good candidates for thermoset epoxies. Hardeners come in a variety of forms, but may have constituencies like C6H8N2, which is representative of the common m-phenylenediamine hardener [3]. Catalytic curing without the use of a chemical hardener is also possible [1]. Bismaleimide resin may be similar in elemental constituency, with a repeating unit of [C21H162(NO2)]n. Both of these binder materials contain aromatic ring structures, which would be expected to enhance the soot formation pathway among the volatile products of pyrolysis when compared to more common fuels consisting primarily of aliphatic compounds. Often, common compounds as described above are used for the main formulation, but trace materials are included in the matrix to mitigate flammability or for other performance reasons. These may also be proprietary. Carbon fibers are chosen because they are strong compared to other common fibers, and are typically manufactured by processing polymer strands through a series of reactions consisting of oxidation and charring steps. The manufacturing process yields strands of high carbon content. They have high tensile strength (~3GPa) [2], and are pliable such that they can be formed in a variety of shapes. Fibers can be from various sources, and may be different thicknesses and strengths. They are marketed typically in two forms frequently used by the design industry. The first is as a tape, which is a sheet of binder impregnated unidirectional fibers. The second is an epoxy impregnated woven sheet. Weaves can come in a variety of patterns. Fibers can be acquired at varying thicknesses in coated laminate sheets termed „prepreg‟. Uncured impregnated sheets are typically sold with tightly controlled epoxy to carbon fractions, generally around 35% epoxy and 65% carbon fiber. Manufacture of parts from these raw materials is also an important consideration. Raw tacky single-layer sheets are often stacked with similar sheets to form layered composites of varying thickness. It is a common practice to rotate the fiber orientation at varying sub-layers, as this has been shown to enhance the general strength of the product. Thermoset resins are cured in a variety of ways, with two more common being in a bag-pressed autoclave, or in a thermal press. The curing process is in a press or pressurized vessel to help induce the removal of air bubbles from within the composite layers to enhance the strength. As the uncured material heats, the binder becomes mobile, so cured composites may have slightly lower binder content than the source prepreg with small amounts of binder escaping along the edges of the manufactured part. Lay-up is typically a manual process, resulting in some potential variability. Aircraft use many shapes and forms of composite materials. Panels for skins are frequently used. Sandwich configurations with any of a number of light-weight secondary internal materials may also be found on designed parts. Structural members can be custom shapes to fit design requirements. Significant fire testing has been performed previously on these types of materials with various objectives in mind. There have also been efforts to describe the fire behavior of composites through modeling. If one considers all fiber/binder composite materials, the literature is extensive. Three fairly recent reviews detail some broad accomplishments in this regard [1,4,5]. Narrowing the scope to just carbon fiber epoxy materials, there is a much more modest amount of literature on fire behavior. Combustion Institute Joint US Sections Meeting May 19-22, 2013 Paper 070FR-0184 Page 3 of 18 Being organic in nature, the above described composites are expected to be flammable and to burn in a fire environment. This is not to say that under normal operation that there is an increased risk related to these materials. Rather, in an abnormal and/or accidental environment the aircraft material may become a contributor to the thermal output from a fire event. Sandia as an institution maintains two principal objectives relative to these materials in a fire. First, we need to understand the behavior of these materials in fires to understand the consequences of transportation fires for safety studies. Second, to further this goal we desire to be able to model the combustion and fire behavior of composites in fire environments. This paper is a review paper on the topic of thermoset carbon fiber epoxy fires, with an emphasis on recent findings by the author and colleagues, distinguishing features of carbon fiber epoxy materials, and technology gaps that have been identified in this area of study. Sections below organize reports containing some of the experimental observations and studies that contribute to the understanding of carbon fiber epoxy fire behavior from previous research by topic. There have been many historical tests, and there is a lot of information generally available. Quintiere et al. (2007) [6] present data on many of these sub-headings for one type of carbon fiber epoxy, and is not necessarily mentioned in each section as having relevant data unless the data are particularly unique or revealing. Other reports are narrower in scope, and are therefore specifically commented on in the appropriate section. Scope is generally limited to studies that focus on carbon fiber epoxy materials, or to papers that clearly exhibit significant findings on the behavior of this class of materials in a fire environment. Binder Reactions There are a lot of types of binders that fit the sub-category of epoxy, and there are many studies on the decomposition of binder materials in the polymer literature. In a fire, the pyrolysis process typically ensues resulting in yield of a combustible gas and a solid char. Several types of experiments are generally found in the context of binder reactions. Thermogravametric Analysis (TGA), Differential Thermal Analysis (DTA) and cone calorimetry are the most common methods for characterizing pyrolysis. TGA generally yields reaction rates which can be used to develop global decomposition behavior reaction mechanisms. DTA and similar types of experiments describe the heat absorption or emission process during decomposition. Pyrolysis reactions are typically considered endothermic [7], as deduced from calorimetry and DTA. Gaseous products of the reaction are addressed in a later section. Documented TGA and DTA experiments can be found on binders alone, as well as on binder/fiber combined materials. Examples of such include the work of Chen and Yeh (1996), Kandare et al. (2007), Regnier and Fontaine (2001), Rose et al. (1994), Schartel et al. (2008), and Trick et al. (1997) [8-13]. TGA may be conducted in an inert environment [8-10,12,13], or in an oxidizing environment (normally air) [10,11]. In TGA, one normally finds binder pyrolytic decomposition taking place (temperature range) between 350-600oC [8,9,11-13]. Cone calorimetry can provide reaction rates, but since samples are typically much larger than in TGA, the instrument is most often used to derive flaming heat release rates (HRR), which is a practical measure of the energy released by the sample material including both pyrolysis and any subsequent reactions such as flaming combustion and surface oxidation. Most of these tests are Combustion Institute Joint US Sections Meeting May 19-22, 2013 Paper 070FR-0184 Page 4 of 18 found just for laminates, not individual constituents. Extensive data can be found in the work of Brown et al. (1998) and Mouritz (2006) [7,14]. Notably, epoxy is found to be one of the most susceptible binders to fire when compared to others in the studies. Avila et al. (2008) [15] also have extensive data on the binder reactions, but in conjunction with a glass fiber. A recent report by Eibl (2012) [16] employing mostly cone calorimetry suggests that the fiber lay-up influences the substrate burn velocity, heat release rate, and ignition in calorimetry tests. It is therefore concluded that the details of the fiber lay-up are consequential to the way binders decompose and burn. We have some limited data in the TGA environment, the publication of which is in progress. An example of the type of data we have obtained that are generally consistent with others in the literature is found in Brown et al. (2012) [17]. We also have unpublished cone calorimetry data for several materials. Much of the historical data adequately describe the behavior of these materials. Documentation of these data is in progress. Fiber Decomposition Carbon fibers can vary in diameter and in constituency based on methods of manufacture. If the range of fibers described in Jiang et al. (2008) [18] can be taken as representative, they are typically around 80% carbon, 15% oxygen, with the balance composed of hydrogen and minor species. A fiber therefore reacts much like other highly carbonaceous materials such as coke, soot, and graphite. High carbon fraction material reaction studies often borrow from each other and use interchangeably reaction rates, etc. Generally speaking, carbon does not pyrolyze, making it a good surface material for very high-temperature applications. It will decompose, but reactions are typically negligible until the temperature exceeds around 700 oC. At these temperatures, the solid carbon will react with gaseous oxidative molecules including OH, O, H2O, CO2, as per Acharya and Kuo (2007) [19]. The rocket design community uses carbon materials for thermally resistive component design, and therefore have studied the reaction rates in the interest of being able to determine lifetime in a severe environment. Examples of these include Acharya and Kuo (2007), Bianchi et al. (2011), Klager (1977), and Kuo and Keswani (1985) [19-22]. Jiang et al (2008) evaluating recyclability of fibers noticed increasing carbon ratios, but similar strengths for fibers put through a fluidized bed to remove the binder. They find that the fibers can be recovered through an intermediate intensity pyrolysis process that will consume the binder, but leave the fibers intact. This is evidence that suggests the char formed by the binder may be expected to preferentially oxidize before fibers oxidize in a fire environment. Further relevant work on carbon oxidation is found in the body of work relating to the oxidation of carbon particles. These studies mostly attempt to uncover the burn rate for the particles. Examples include the work of Blake and Libby (1991), Blake (2002), Chelliah et al (1996), Kassoy and Libby (1982), Libby and Blake (1979), Libby and Blake (1981), and Makino and Law (2009) [23-29]. In a closely related paper with a geometric variation, Makino et al. (2003) evaluate the combustion rate for graphite rods in a high-temperature air flow. Brown et al. (2011) [30] documents a series of recent tests done at Sandia National Labs where 25-40 kg of varying types of carbon fiber epoxy materials were combusted in an insulated Combustion Institute Joint US Sections Meeting May 19-22, 2013 Paper 070FR-0184 Page 5 of 18 enclosure. Shape of the composites was also significantly varied. The intent of the tests was to evaluate the capacity for the fibers to burn in an extreme environment, and to explore the peak heat fluxes that could be generated by the burning of carbon fiber epoxy composites. The tests were notable in the duration and in the intensity. Peak fluxes of 200 kW/m 2 were obtained during flaming and glowing (char and fiber oxidation) phases. The environment exhibited two peaks in intensity (as deduced by the measured heat flux), the first corresponding to flaming combustion, and the second to oxidative reactions (See Figure 1 for an example result). Medtherm radiometers and a bulk metal calorimeter were used to monitor the heat flux during the test. The drop in intensity between the two peaks is presently believed to be due to the lower energy release in the char oxidation phase immediately following flaming combustion when compared to that of the fiber oxidation phase. The duration of the tests was remarkably long, 5-8 hours. Comparable (on a mass basis) wood fire tests were just 1-1.5 hours long. Figure 2 shows some images from one of the tests that illustrate the dynamic behavior of the panel decomposition with time. Notice in particular one panel that fell off the rack and was in good view of the camera as it was decomposing in the top-left part of the opening in the images. This test series achieved 90-98% mass consumption of the initial material, which is suggestive that under ideal conditions it is possible to consume very close to 100% of the carbon fiber epoxy material used in the aircraft design. Pickett et al. (2011) [31] note in contrast that 25-50% mass consumption is typically expected. Figure 1. Measured heat fluxes from the combustion of a carbon fiber epoxy in an insulated enclosure. The reports on carbon oxidation (including many listed above) suggest minimal reaction until a sustained temperature in the vicinity of 700oC is attained in air. It is therefore not expected that Time [min.] 0 60 120 180 240 300 360 R a d ia tiv e F lu x [k W /m 2 ] 0 50 100 150 200 250