1. Introduction
The need of high temperature insulation polymers in the aerospace industry has led to the development of high-performance polyimide foam systems for various applications depending on its structure [
1]. The PolyuMAC
TM polyimide foam insulation is a material developed in a joint effort between the NASA Langley Research Center and PolyuMAC TechnoCore, Inc. [
2]. Its success is due to the flexibility in production and low manufacturing cost. The fabrication of this low-density and high-performance polyimide foam evolved into another generation of FPF-44 foam. This underwent ice mitigation tests on the insulation of liquid oxygen (LOX) feedline on the space shuttle external tank into the commercial production of the joint NASA LaRC-PolyuMAC™ foam [
3]. This polyimide foam emerged from failed attempts to fabricate a composite material for a supersonic aircraft project [
4]. They considered fabricating a new generation of insulation polyimide foams which resulted on the production of the TEEK (an acronym denoting the inventors’ names) technology. TEEK is a term used by NASA given to a polyimide insulator that can be found in powder form (precursor), friable balloon (intermediate material), and a foam [
5]. First generation TEEK polyimide foams had a rigid structure manufactured using a friable balloon format, which allowed molding and curing it to the desired shape depending on the application. The chemical structure of this foam developed by NASA was improved into a new product FPF-44, also known as PolyuMAC
TM. Unlike the former polyimide foam manufacturing technique, the new technology consists of a polyimide foam that rises at room temperature and cures via microwaving [
4]. This TEEK technology derivation has acoustic and thermal insulation capabilities, self-extinguishing properties, and fire resistance and retardation. The foam is prepared from an aromatic of the polyimide precursor’s solid residuum and is composed by an expandable powder imidized upon the curing cycle [
6]. This foam provided an effective insulation at cryogenic temperatures while maintaining a flexible structure. Moreover, this material is the first polyimide foam to rise at room temperature during the foaming phase. For the curing process, a mold with the desired shape (with the foam material) is exposed to microwave radiation of low intensity. This process reduced the manufacturing cost and increased the production rate [
1,
7].
Polymeric foams can be classified into two types, thermoplastic and thermoset. Polyimide foam is a highly cross-linked thermoset material. Their mechanical properties will depend on the glass transition temperature value of the polymeric material [
8]. Knowing that polyimides have thermosetting matrices, one can expect a higher thermal resistance and higher thermal insulation [
9]. Although, polyimide foams have been characterized in many ways, tortuosity is rarely mentioned. The concept of tortuosity can be simplified as the transport of a fluid through a complex configuration. This perspective allows a basic design of possible tortuosity models, depending on the investigation. C. Carman [
10] introduced a tortuosity index intended to match experimental observations with a permeability model on granular beds influenced by the dimensionless analysis of parallel tubes [
11]. For porous media, empirical tortuosity models have been proposed as a function of molecular diffusivity and electrical resistivity. For instance, an ideal tortuosity model for characterization of sandstone rocks was developed by Garrouch and Ali through diffusion and electrical measurements [
12]. In the case of the polyimide foam, the acoustic insulation performance of the material on an aircraft double panel application was studied matching inverse characterization data with finite element predictions of transmission loss [
2]. However, a model of tortuosity was not presented; instead, an optimization procedure on the finite element software COMET/Trim estimates the tortuosity.
Recent studies of foams for thermoacoustic applications demonstrated how tortuosity is very relevant to characterizing the material [
13]. In this study, Napolitano et al. proposed an empirical approach. Additionally, this type of evaluation of acoustic properties is not restricted to polymeric foams. For instance, tortuosity is estimated through a diffusion technique applied to carbon foams [
14]; in this case, the tortuosity value was assumed based on previous studies [
15], but results in a methodology lacking practicality. Tortuosity becomes relevant in mass transfer applications due to the interaction of two mediums (gas/solid), and, by experimental measurements and numerical simulations, some authors proposed correlations for mass transfer in open-cell foams [
16,
17]. These types of materials (i.e., foams) possess great acoustic properties, that depend on their manufacturing methods and the characterization techniques used and that deal with the concept of tortuosity [
18].
The present research focuses on the analysis of an open cell polyimide foam with flexible structure. If tortuosity can be defined as the ratio of the real-to-apparent distances that air flow has to travel through the foam, it can be correlated to sound absorption coefficients using microphones [
19]. If the porous media (in our case the PolyuMAC foam) is intended for sound absorption applications (aircrafts, for instance), then the assessment of tortuosity in the porous media becomes very relevant. Since there is no standard method to measure tortuosity for porous media, we propose a methodology to define a tortuosity index. In addition, we provide a full characterization of PolyuMAC™ polyimide foam intended for aerospace applications. The material is evaluated as fluid is transported through the porous media.
3. Discussion
The characterization methods employed corroborated the chemical and thermal properties of the PolyuMAC
TM polyimide foam. In effect, this material possesses a high heat resistance even at temperatures higher than 500 °C, which gives it a wide range of potential applications. The FTIR spectra displayed a characteristic result for aromatic polyimides with high
. The thermal stability and resistance to oxidation degradation were successfully validated by a
of 294.2 ± 9.9 °C. As aforementioned, the new TEEK technology allows producing light polyimide foams with a density ranging between 3.2 and 16 kg/m
3 [
2]. With a density of 5.922 kg/m
3, the material under study falls within such range with a standard deviation of 0.325 kg/m
3. At this point, one should note that density is not a measure of the foam structure rigidity. Instead, density is one factor reflecting the quality and performance of the material. For instance, in a study of polyimide foam used as a double panel for sidewall applications by Silcox et al. [
2], a sample with a density of 6.4 kg/m
3 compressed at different percentages showed an average Young’s modulus of approximately 75, 95, and 130 kPa for 30%, 60%, and 90% compression levels, respectively. These values are comparable with the outcomes of the present research, where the mean Young’s moduli were 78, 93, and 180 kPa for 30%, 60%, and 90% compression levels, respectively. In that same study of double panels at the Gulfstream Acoustic Test Facility, a 9.6 kg/m
3 density sample compressed at 10% of its original length had a tortuosity of 3.11 and for samples of 5.4 kg/m
3 compressed at 20% of its original length presented a tortuosity of 1.02. The samples in this study, compressed at 30% from its original length had an approximate tortuosity of 3.05. The method to determine the tortuosity is not specified by Silcox et al., but we can infer that density and compression level affected such tortuosity.
At this point, one must acknowledge that a more extensive assessment of the polyimide oxidation sensitivity would have been suitable. Unfortunately, that experimentation would need a thermogravimetric analyzer operated with air atmosphere and interfaced with a Fourier transform infrared spectrophotometer for the simultaneous characterization of evolved volatiles [
29]. Our TGA unit lacks such capability and only produces a differential weight or weight loss versus temperature curve.
In our research, our empirical model used the pressure difference as an analogous approach for a resistivity factor. In effect, one can observe in the surface plot of
Figure 10 that the pressure drop changes with the sample length and compression percent. According to Darcy’s Law, the parameters involving pressure drop can describe the effect of the fluid velocity passing through the interconnected open cells. A lower value of permeability can then be considered as a lower flow rate of the fluid phase through the solid phase. The tortuosity barely fluctuates between dynamic compressions because of the almost constant porosity parameter, as explained before. On the other hand, the permeability increased with the sample length. Hence, the slope obtained in the linear fit per unit length resulted in an inverse relationship between
and
. The slope is affected by the permeability. Samples with 76 mm (3 in.) length had more volume than samples with 25 mm (1 in.) length; hence, the pressure drop was higher in the 76 mm samples, justifying the hypothesis of a velocity attenuation as the fluid flows through the porous media at a constant rate.
A regression model for the dynamic compressions was obtained using an analysis of variance. The residuals of this lineal regression met the assumptions of normality, equal variances, and independence. Therefore, a regression equation was produced for the estimation of tortuosity index at different dynamic compressions. The R2adj of the regression equation increased and the standard deviation decreased with the number of dynamic compressions. These results indicate that the 3 DC model was statistically more significant than those obtained for less dynamic compressions. Still, further research is required to thoroughly study the tendency for more than 3 DC. These models were simplified by reducing the number of regressor variables without diverting from the original purpose of the analysis using a best subsets regression strategy. Since the technique selects the best combination of regressor variables at different levels, reasonable and simpler models are developed so as to explain more effectively the tortuosity at different dynamic compressions.
The values of R2 and R2adj, as well as the standard deviation were the principal criteria for the selection of the variable combination. The two variables that explained the variability of tortuosity the most they were and . For 1 DC and 2 DC, the R2adj for the new models were 91.3% and 94.3%, respectively. However, the model for 3 DC had a better correlation with an R2adj of 95.7%. In addition, these two variables were able to explain the interaction between the fluid and the solid phase. The reason why was eliminated by the best subsets’ regression is because the parameter is a geometry factor of the sample. Intrinsically each sample carries the effect after deformation at a compression percent when it is characterized on the flow resistivity test.
The method presented in this study considers the pressure coefficient as well as the Reynolds number as the necessary factors to determine the tortuosity index. This low-cost procedure allowed establishing that other factors do not have enough relevance in determining the tortuosity of the PolyuMAC™ foam. Thus, by an indirect yet economical means, the tortuosity index can be determined via the flow through the porous media. This presents an advantageous alternative to test this material, as flow properties can be employed to indirectly obtain the foam tortuosity. In addition, we think that this method can be readily applied to other foam materials because the factors used to determine the tortuosity index were independent of the foam nature, as long as sample is highly porous.
4. Concluding Remarks
This work presents a multifarious assessment of the PolyuMACTM polyimide foam used for insulation and acoustic absorption in aerospace components. The material was provided by NASA Langley Research Center “as is,” i.e., without prior compressions.
Dynamic compressions of 30%, 60%, and 90% of the foam’s original length allowed the mechanical response of the foam to be evaluated. As the compression percent of the specimens increased, so did their densification. This led to higher strength on the more deformed specimens. Although this flexible material was expected to recover its full length after 24 h, most samples recovered 95% of their initial length. Scanning electron microscopy confirmed the effect of the dynamic compressions on the structure and the damage of some ligaments.
Chemical and thermal characterization of the PolyuMACTM foam demonstrate the importance of this material type as an insulator, since it has softening and decomposition temperatures over 400 and 560 °C respectively. In addition, this foam possesses a very low density, causing also a high absorption capacity (of a low surface tension fluid) of over 97%. As an aromatic polyimide, the foam is apt for flexible structures that withstand high temperatures without losing fire retardation capacity. As expected, the collected FTIR spectra were characteristic of a highly aromatic polyimide foam and confirmed a thermoset material where mechanical compression produced no apparent effect on the polyimide structure (architecture).
Tortuosity, a complex parameter that describes the energy absorption capacity of the material, cannot be measured directly. Therefore, the present research proposes an alternative methodology for the determination of the tortuosity index via indirect characterization of the porous structure of a PolyuMACTM polyimide foam. This empirical model depends only on the pressure coefficient and Reynolds number of a fluid passing through the open cell structure of the foam. The methodology based on an empirical model uses the Biot parameters describing the interaction between a fluid and a solid phase for flow transport through porous media. Quantitative image analysis permitted to measure the porosity of the open cell structure of the foam. Straightforward analysis methods, such as Biot poroelasticity model and Darcy’s law, allowed registering macroscopic parameters defined on Pi parameters. An analysis of variance was the statistical analysis used to correlate the Pi terms. The factors represented by (a) the number of dynamic compressions and (b) the length of the samples required an independent evaluation for better understanding. The recovery capacity of the material was a component to help explain the lack of variance between the dynamic compressions.