Ablative properties of carbon black and MWNT/phenolic composites: A comparative study

Composites: Part A 43 (2012) 174–182

Contents lists available at SciVerse ScienceDirect

Composites: Part A

journal homepage: www.elsevier .com/locate /composi tesa

Ablative properties of carbon black and MWNT/phenolic composites:A comparative study

Maurizio Natali, Marco Monti, Debora Puglia, José Maria Kenny 1, Luigi Torre ⇑University of Perugia, Strada di Pentima 4, 05100 Terni, Italy

a r t i c l e i n f o a b s t r a c t

Article history:Received 26 May 2011Received in revised form 26 September2011Accepted 9 October 2011Available online 14 October 2011

Keywords:A. Polymer–matrix composites (PMCs)B. High-temperature propertiesD. Thermal analysisOxyacetylene torch test

1359-835X/$ - see front matter � 2011 Elsevier Ltd.doi:10.1016/j.compositesa.2011.10.006

⇑ Corresponding author. Tel.: +39 0744 492918.E-mail address: [email protected] (L. Torre).

1 Present address: Institute of Polymer Science andCierva 3, 28006 Madrid, Spain.

In this work, we investigated the ablative properties of two carbon nanofiller-based composites. In par-ticular, carbon black (CB) and multi-walled carbon nanotubes (MWNTs) were used to produce highlyloaded (50 wt%) phenolic composites. The thermal properties and the ablative response of the compositeswere studied through the pre and post-burning morphology of the burnt surfaces and an evaluation ofthe in-depth temperature profiles. When compared to the CB-based counterpart, the MWNT-basedcomposite exhibited a higher thermal diffusivity and an erosion rate that was exactly localized abovethe flame plume. The CB-based system showed a thin charred region whilst the MWNT-based was char-acterized by a thick and wide pyrolyzed zone.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Ablative materials play a strategic role in the aerospace indus-try. They are used to produce the Thermal Protection System(TPS) which protects the structures, the aerodynamic surfacesand the payload of vehicles and probes during hypersonic flightthrough a planetary atmosphere [1]. Ablative materials are alsoused to manufacture passively cooled rocket combustion chambersand to provide insulation of propulsion devices from very hightemperatures [2]. Such a wide range of heat shields is generallyproduced using reinforced polymers. The selection of the properablative material is strictly related to the properties of the hyper-thermal environment in which the TPS must work: no singlematerial can satisfy a broad series of operational conditions in ahighly efficient manner. Accordingly, during the last 50 years,depending on the application area, many types of polymeric abla-tors have been developed [3].

Among the various types of polymeric matrices used in abla-tives [1,2,4–8], the burnt matrix is relatively weak, even whenusing a high char retention resin such as a phenolic: the charredpolymer can be mechanically removed by the friction actioncaused by the atmospheric gases or to the rocket combustion prod-ucts. To assist the char retention, a wide range of reinforcementscan be added to the matrix. Fibers made of carbon, refractory

All rights reserved.

Technology, CSIC, Juan de la

oxides, mineral asbestos or glass are typically used. Micron-sizedpowdered fillers also play a very important role [1,2,9]. However,several limitations associated with the traditional ablativecomposites, which are structured on a micron scale, motivatedthe efforts to identify the next generation of ablatives [10]: in fact,even in the presence of fibrous reinforcements, the material cansuffer high mechanical erosion. Composite materials producedwith nanosized reinforcements have led to a new paradigm forpolymeric ablators. In contrast to conventional composites con-taining micron-scaled reinforcements, Nanocomposite RocketAblative Materials (NRAMs) [11] are nanocomposites particularlydesigned to work in severe hyperthermal environments. LayeredSilicates (LSs) [10,12] as well as nanosilica [13–15] have been suc-cessfully used in nanostructured ablatives leading to improvedablation resistance.

Among the nano-additives, nanosized carbon fillers play animportant role in both traditional as well as nanostructured abla-tive materials: in fact, well before the appearance of the termnanocomposite, carbon black (CB) was widely employed in theproduction of ablators. For example, Cytec Engineered MaterialsMX-4926, one of the NASA standard carbon/phenolic nozzle mate-rials, is composed by 50 wt% of carbon fibers, 15 wt% of carbonblack and 35 wt% of phenolic resin. Recently, in the class of nano-sized carbon fillers, nanofibers (CNFs) have been studied as a fillerfor polymeric TPSs. Koo et al. [16] compared the ablative propertiesof MX-4926 to laboratory produced CNFs/phenolic nanocompos-ites (CNF-NRAMs). Borden Chemical’s SC-1008 was selected as aphenolic matrix for the preparation of the CNF-based nanocompos-ites. Three CNF loadings, namely 20%, 24% and 28 wt% were

M. Natali et al. / Composites: Part A 43 (2012) 174–182 175

dispersed in the matrix without a rayon-carbon fiber reinforce-ment. A subscale rocket motor (heat flux 1140 W/cm2) was usedto study the ablation and insulation characteristics of the ablatives.According to the experimental results, when compared to the MX-4926, all the CNF-NRAMs exhibited lower ablation rates and lowermaximum backside heat-soaked temperature rose. Peak erosion ofthe higher loaded CNF-NRAM without the rayon fabric was 42%lower than the MX-4926. However, the residual mass of MX-4926 was higher than that of CNF-NRAMs. Additionally, the surfacetemperatures of the CNF-NRAM samples were higher than that ofMX-4926.

Patton et al. [17] produced a series of nanocomposites with aCNF load up to 40 wt% (in absence of woven cloth reinforcement).A plasma torch test bed was used for the ablation testing (1650W/cm2 heat flux). Under this test condition, in comparison to theMX-4926 baseline, the CNF-NRAMs experienced higher erosionrates and lower weight losses. According to the authors, the highererosion rates of the CNF-NRAM samples probably reflected theirlower carbon content. In fact, even when considering the higherloaded CNF nanocomposite, NRAMs had a significantly lower car-bon loading than the MX-4926 composite (65%). Thus, the erosionrate per weight percentage of carbon reinforcement present in thecomposites were very similar. However, the CNF nanocompositesappeared to be far better insulators than the MX-4926.

The aim of our investigation was to evaluate carbon nanotubes(CNTs) as a potential filler for ablatives. A series of interconnectedfactors gave rise to this research: first, CNTs, particularly MultiWall Carbon Nanotubes (MWNTs), experienced a dramatic dropin costs. Secondly, MWNTs proved to be very effective flame retar-dant additives for polymers [18]. Kashiwagi et al. [19–22] attributethe improved flame resistance to the formation of a continuousprotective nanotube network structure working as a heat shield.Such a protective barrier reduces the mass loss rate and materialflammability. Flame retardancy improved at higher CNT loadingswhich is consistent with the above-mentioned mechanism.Accordingly, in addition to the use of CNTs directly dispersed inthe matrix, the use of buckypapers (carbon nanotube membranes)[23] is a very effective alternative method for exploiting the barrierproperties of CNTs: dense nanotube networks and small pore sizewithin the buckypaper provide low gas and mass permeability,which means that a buckypaper acts as an inherent flame retardantshield. Moreover, CNTs embedded in the charred surface re-emitsmuch of the incident radiation into the gas phase from its hot sur-face, reducing the transmitted flux to the inner layers of the mate-rial and thereby reducing the polymer pyrolysis rate [24,25].

However, to date, considering the remarkable flame retardantproperties of MWNTs, there is a lack of research on this nanofillerfor ablatives. Therefore we investigated the ablative response ofMWNTs embedded in a polymeric matrix. Furthermore, as in thecase of the aforementioned research involving the use of CNFs inablatives, there is likewise the necessity to study the behavior ofMWNT-based ablators at very high filler loads. As pointed out bythe cited literature, MWNTs exhibited better flame retardancy athigh loads. In the previous paragraphs, the important role playedby carbon black in the production of many polymeric TPSs was alsohighlighted: as a consequence, a direct comparison of the ablativeproperties of CB and MWNT certainly represents valuable and use-ful research for both the scientific community and the industry.

In our study, the ablative properties of a carbon black/phenoliccomposite were compared with a MWNT/phenolic counterpart.These composites were produced with a very high filler load,namely 50%, using a compression molding technique. The pro-duced materials were firstly studied by means of thermogravimet-ric analysis (TGA), thermomechanical analysis (TMA) anddifferential scanning calorimetry (DSC). Then, the ablativeresponse of these composites was evaluated in a severe hyper

thermal environment provided by an oxy-acetylene torch testbed. The pre- and post-burning morphology of the studied materi-als were investigated by means of scanning electron microscopy(SEM) and optical analysis. The erosion rate and loss of mass ofthe tested composites was also investigated. Finally, in-depthtemperature profiles were acquired and discussed.

2. Materials and methods

In this study, a powder novolac resin (PR) containing 5.5–6.5%hexamine was employed (kindly supplied by Hexion and commer-cialized as Bakelite� PF GA T 6). Carbon black and carbon nano-tubes were utilized as fillers for the composite materials. Thecarbon black, supplied by Cabot Corporation (Vulcan� 7H), con-sisted of nano-sized spherical particles with an average diameterof 90 nm. MWNTs were supplied by Arkema (Graphistrength�

C100). The technical datasheet reports that these MWNTs have acarbon content above 90 wt%, an outer diameter of 10–15 nm(about 5–15 walls) and a length of 0.1–10 lm. In this study twodifferent mixtures were produced, both consisting of 50 wt% ofthe phenolic matrix and 50 wt% of the nanofiller: one with CB(hereinafter referred to as PR-CB), constituted by 50%-PR and50%-CB, the other with MWNTs, constituted by 50%-PR and50%-MWNTs (hereinafter referred to as PR-MWNT). The compositematerials were prepared in several steps. First, the exact amount ofresin powder and filler was weighed and mixed with a ball millingtechnique for 4 h. Once an even dispersion was obtained, a mount-ing press (Struers, Predopress model) was used to produce thesamples. The blended powder was put in the cylinder of a press(30 mm diameter) and then processed. Samples with two differentthickness values were produced in order to have samples for all thevarious types of measuring equipment. The thicker specimenswere needed for the flame test, whereas the thinner ones wereused for all the other thermal characterizations. Consequently,two amounts of mixtures, namely 9 g and 3 g, were used. As a re-sult, cylindrical-shaped specimens with a diameter of 30 mm wereobtained. The density of the produced composites was evaluated asabout 1.05 g/cm3 for both the systems.

The mounting pressure was set at 40 kN at 8 bar. During theapplication of the pressure, the press cylinder was heated up to180 �C for 6 min to allow the curing of the resin. Then, a coolingcycle of 7 min was applied. These parameters were chosen after around-robin series of experiments for the production of a defectfree sample. In order to obtain fully-cured materials, all the sam-ples were then post-cured at 180 �C for 2 h. The effectiveness ofthe curing process was verified by DSC analysis, not reported inthis paper.

The thermal stability of the produced materials was investi-gated by means of TGA, comparing the two mixtures (PR-CB andPR-MWNT) with the neat phenolic resin. The measurements wereperformed with a Seiko model Exstar 6300 thermogravimetric ana-lyzer, in both air and nitrogen and consisted of dynamic scans at aheating rate of 20 �C/min from 30 �C to 1200 �C. Bulk samples ofabout (10 ± 1) mg were tested. In order to have good repeatability,at least 10 tests for each material were carried out.

A TMA analysis was carried out with a Perkin Elmer model TMA7 thermo-mechanical analyzer. Dynamic scans were performed at20�/min from 30 �C to 700 �C.

The heating capacity (CP) of the composites as a function of tem-perature were measured using a DSC (TA, model Q200) in the rangeof temperature from 30 to 450 �C. The measurements were carriedout as follows [26]. Firstly, the heat flow patterns of each materialwere acquired by means of a DSC dynamic scan (at a heating rate of20 �C/min). The heat flow curves obtained with the DSC were com-bined with TGA thermograms: in this way, the loss of mass of the

176 M. Natali et al. / Composites: Part A 43 (2012) 174–182

tested materials was taken into consideration in the analyticalequation used to extrapolate the heat capacity. The heat capacitywas calculated according to the following equation:

CPðTÞ ¼1

mðTÞ@H@T¼ 1

mðTÞ@H=@t@T=@T

¼ 1mðTÞ

DPb

where oH/ot is the heat flow, m(T) is the mass of the sampleprovided by the TGA, DP is the signal of the DSC (W) and b is theheating rate.

In order to appropriately test the ablative materials, the use of ahyperthermal environment with a very high heat flux is necessary.This cannot be reproduced by a TGA because of the elevated heat-ing rate needed (up to 50 000 �C/min) [11] and the fact that it isimpossible to reproduce any shear rate stress caused by combus-tion gases. In this work, an oxy-acetylene torch was used to simu-late the severe hyperthermal environment [27]. This device is ableto produce both a high temperature flame (up to 3000 �C) and ahigh heat flux. A more detailed description concerning the em-ployed device can be found in our previous work [14]. In this study,the distance between the sample surface and the flame was set at25 mm and the heat flux produced by the flame was set at 500W/cm2. The calibration of the power of the flame was carried outusing a copper slug calorimeter [28]. Due to its features, the scien-tific community has expressed vivid interest in this test bed andnumerous research groups have studied ablative materials usingthis device [14,15,29,30]. Recently, in order to fully exploit theexperimental data provided by the oxy-acetylene torch based facil-ities, the thermal field generated by this heat source was modeled[31].

In our test, the flame was applied to the lateral surface of thecylinder (on the mantel) and in depth temperatures were acquiredat 5 mm (referred as T1) and 10 mm (T2) from the mantel (Fig. 1).Thermocouples were arranged in two blind holes drilled perpen-dicularly at the bases of the cylindrical shaped samples. To ensureconsistent acquisition of temperatures, the depth of the holes waskept constant at 4 mm. Fully sheathed, ungrounded, K-type ther-mocouples were used. The external body of the thermocoupleswas protected using two alumina tubes. All the specimens wereexposed to the flame for 50 s. Ten specimens of each material weretested. The role played by the fillers on the thermal diffusivity ofthe materials was also studied analyzing the temperature profilesacquired through the samples. The loss of weight due to the abla-tion process was evaluated.

The most effective way to evaluate the erosion rate of an abla-tive material subjected to a hyperthermal environment is based onan analysis of the post-burnt surfaces by optical and SEM analysis.

Fig. 1. Oxy-acetylene torch test bed layout. (For interpretation of the references tocolour in this figure legend, the reader is referred to the web version of this article.)

An assessment of the morphological aspects of the produced mate-rial was carried out with a ZEISS, model Supra 25 field emissionscanning electron microscope (FESEM). This analysis was per-formed both on the pristine and on the burnt samples, allowingthe study of the initial morphology of the composites and of the ef-fects of the high-temperature exposure. Due to the high load andelectrical conductivity of MWNTs and CB, no sputter coating treat-ment was needed for the SEM analysis.

3. Results and discussion

3.1. Morphological characterization

The morphology of the produced materials was investigated bymeans of a SEM analysis. Fig. 2a and b refers to the fractured sur-face of the PR-CB sample. By increasing the magnification, it is pos-sible to clearly identify the carbon black particles (the sphericalstructures) embedded in the polymeric matrix. In all pictures, theCB is uniformly dispersed on the whole surface of the sample.Fig. 2c and d shows some representative pictures of the PR-MWNTsamples. The MWNTs are distinguishable as bright single andentangled filaments. In some cases the nanotubes are not visibledue to fact the filaments are embedded in the matrix. Sometimes,only the ends of the covered MWNTs are visible (lower right cornerof Fig. 2d). The phenolic matrix surrounded the MWNTs, highlight-ing the effective impregnation of the filler, even at the high concen-tration at which the PR-MWNTs were produced. Moreover,comparing the PR-CB system (Fig. 2a) with the PR-MWNT compos-ite (Fig. 2c) it is possible to see that the former system shows amore uniform surface. By virtue of the fact that the CB is an homo-geneous, isotropic filler, the fracture surface of the PR-CB appearedrelatively regular and flat. Conversely, the MWNT filaments havean high aspect ratio. Hence, as a consequence of the fracture, theregions in which the MWNTs experienced debonding or loweradhesion with the matrix, appeared with a different morphologyof the zones in which the nanotubes remained embedded in thematrix. In any case, for the PR-MWNT system, the distribution ofboth types of regions is uniform on the whole sample surface.

3.2. TGA results

3.2.1. Results in nitrogenThe TGA patterns of the pristine fillers are shown in Fig. 3. Fig. 4

shows the thermogravimetric patterns of the studied materialstested in nitrogen. Thermal degradation of the neat novolac andof the filled systems occurred in three temperature dependentstages: a low temperature stage (below 200 �C), an intermediatetemperature stage (between 200 and 500 �C) and a high tempera-ture stage (above 500 �C) [32]. At temperatures below 200 �C, thephenolic networks were relatively stable in inert atmospheres.However, a small amount (1–2 wt%) of gaseous products werereleased: this loss of weight can be attributed to the evolution ofgases trapped in the matrix during the curing reaction. These com-ponents include water, formaldehyde and phenol [33]. Concerningthe data on the neat novolac, the system could be considered freeof unreacted monomers and that no moisture had been absorbedby the bulk neat resin. If compared to the neat resin, the PR-CBand PR-MWNT composites exhibited a higher loss of weight.Depending on the type of filler, the different amounts of moistureabsorbed by the composites from the environment could probablybe attributed to the specific porosity of the materials.

For the neat novolac, a striking increase in volatilization was ob-served in the region 200–350 �C which can be directly related tothe evolution of water, CO and CO2. This large increase in waterevolution observed for the cured resin sample can be related to

Fig. 2. Fracture surfaces of the produced materials: (a and b) PR-CB composite, (c and d) PR-MWNT composite.

Fig. 3. TGA patterns of the used fillers. (For interpretation of the references tocolour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4. TGA patterns of the studied materials (nitrogen). (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version ofthis article.)

M. Natali et al. / Composites: Part A 43 (2012) 174–182 177

the post cure reactions in which water is liberated while dihy-droxybenzophenone linkages are formed in the polymer structure[33]. The increased evolution of CO and CO2 from the cured novolacresin is related to the decarboxylation and decarbonylation of theoxidation products formed during the curing process in air. Onthe other hand, the evolution of these gases is less evident forthe PR-CB and PR-MWNT cured composites.

In the range 350–500 �C, the most significant structural changescould be observed in the networks. According to the literature onthe thermal degradation of phenolic systems [34], gaseous compo-nents such as water, carbon dioxide, methane, phenols and cresolsare released during this stage. The presence of CB and MWNTs sig-nificantly affected the observed phenomena. In particular, the

carbon based fillers hindered the formation of water [35]. Themodified decomposition of the composite systems could be ex-plained considering the possible interaction of volatile molecularspecies with the functional groups containing oxygen, which arelocalized on the surface of the carbon fillers. Since these groupsare capable of condensation reactions with volatile compounds,these reactions could partially account for the lower level of vola-tilization observed in the pyrolysis of PR-CB and PR-MWNTcomposites.

The degradation mechanism at temperatures above 400 �C wasessentially the same for the three different materials and is due to aradical bond rupture and the subsequent formation of aromatic

Fig. 5. TGA patterns of the studied materials (air). (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version ofthis article.)

178 M. Natali et al. / Composites: Part A 43 (2012) 174–182

products [36]. Above 400 �C, the presence of the nanofillers did notintroduce any shifting in the DTG peaks. In fact, for all the threesystems, the main DTG peaks displayed their maximum values atnearly the same temperature: about 425 �C for the first peak andnearly 550 �C for the second one. The highest DTG peak can beattributed to the rearrangement of the cross-linked system and isassociated with the evolution of light molecules (methane, ethane,hydrogen, carbon monoxide and carbon dioxide). For temperaturesabove 650 �C, dehydration occurred and a carbon-like structure(char) was gradually formed. Beyond this temperature, the curveis approximately flat. The neat novolac exhibited a typical charyield of about 60 wt%, whilst for the filled systems, the residualmass was 76.1 wt% for the PR-MWNT composite and 75.6 wt% forthe PR-CB system.

3.2.2. Results in airFig. 5 shows the thermogravimetric patterns of the studied

materials tested in an oxidizing environment. In air, the oxidativedegradation began at lower temperatures than in an inert environ-ment. In particular, for the neat phenolic, the oxidation started at atemperature below 300 �C: oxidative branching and crosslinkingwere the prevalent degradation mechanisms in air [37]. As forthe systems studied in nitrogen, the neat novolac exhibited a strik-ing increase in volatilization in the region 200–350 �C, whilst thefilled systems did not display a corresponding behavior.

The fragmentation of the polymer chains began at approxi-mately 350 �C, while the network collapsed as polyaromatic do-mains nearly above 500 �C [38]. As reported in literature,oxidation occurred most readily on benzylic methylene carbonssince they are the most vulnerable sites, leading to the formationof dihydroxybenzohydrole and dihydroxybenzophenone deriva-tives. Upon further heating in air, quinones and carboxylic acidswere formed. Dihydroxybenzophenone may cleave and furtheroxidize to form carboxylic acid before decomposing to form cre-sols, CO and CO2 [39]. The presence of carbon based fillers loweredthe temperature of the maximum DTG peak (about 700 �C for PR-CB and 740 �C for PR-MWNT) while the neat novolac exhibitedthe maximum degradation rate at about 770 �C. The neat matrixcompletely degraded at approximately 900 �C as the weight lossapproached 100%, whilst the PR-CB and PR-MWNT degraded atnearly 800 �C.

For the PR-MWNT system, the residual mass at 1200 �C wasroughly equivalent to the content of the metal catalysts used for

the synthesis of the MWNT (6.3 wt% on the initial weight of thecomposite), while in the case of the PR-CB composite, a residualmass of about 3 wt% was probably due to the incomplete oxidationof the charred material. To confirm this hypothesis, bulk PR-CB andPR-MWNT composite samples were burnt in a muffle at 1000 �C for2 h. For the PR-CB system, no residual mass was detected after thetest, whilst for the PR-MWNT the presence of metal precursorsquantitatively and qualitatively confirmed the TGA results.

3.3. Thermo-mechanical analysis

The pyrolysis of the phenolic based systems is generally cou-pled with volume shrinkage due to the degradation of the resin,followed by escaping gases and cracking of the matrix. Fig. 6 showsthe TMA results of the tested materials: the ratio between thethickness of the sample during heating and its initial value (l/l0)is displayed as a function of the temperature. It is possible to seethat in the whole range of temperatures, the introduction of bothnanofillers significantly reduced the volume changes. The TMApattern of the neat novolac showed a more marked dependenceon the temperature associated to the major volatilization of theproducts.

Concerning the neat resin, the TMA profile perfectly matchedthe DTG curve (inset picture of Fig. 6). All previously identified deg-radation steps could be confirmed in the thermo-mechanical anal-ysis. The results of the TMA investigation on PR-CB and PR-MWNTcomposites also confirmed the DTG results: overlapping the DTGcurves in air and the TMA patterns, it is possible to confirm thatno degradation occurs up to 300 �C. Only the volatilization of lowmolecular weight species and of the absorbed moisture could bedetected. After this temperature, PR-CB and PR-MWNT systemsstarted to shrink in a similar way due to the emission of volatilecompounds, until they began to degrade.

In the absence of any filler, above a temperature of about 550 �Cthe neat matrix shrunk at a higher rate than the loaded systems.The final shrinkage values calculated at 700 �C were 0.903 for neatnovolac, 0.970 for the PR-MWNT and 0.958 for the PR-CB. Accord-ingly, at the higher temperatures, the slope of the l/l0 can beconsidered nearly equal for both the composites.

3.4. Heat capacity

Fig. 7 shows the CP patterns of the tested materials. The study ofthe heat capacity as a function of the temperature can be veryimportant for the comprehension of the ablation mechanism andit can also contribute to the modeling of the process.

In the whole range of temperatures, both composites exhibiteda higher CP than the pristine matrix, with the PR-MWNT systemhaving the highest values. At a temperature below 150 �C, as high-lighted by the TGA analysis, the CP patterns confirmed the releaseof moisture in the tested composites: in fact, in this range oftemperature, they exhibited a peak in the CP patterns which clearlyrefers to the absorbed water humidity. On the other hand, the ab-sence of this peak in the neat matrix confirmed that the pristinematrix did not absorb any moisture. The broad release of volatilesin the neat resin (200–300 �C region) observed in the TGA profilecan be related to the corresponding peak in the CP pattern. TheCP trends of the studied composites can be considered qualitativelyin line with the values of traditional carbon fibers/phenolic com-posites [40].

3.5. Oxy-acetylene torch test results

Fig. 8 shows the surfaces of the produced composites after theexposure to the flame. Fig. 8a and c refer to the PR-MWNT system,whereas Fig. 8b and d are related to the PR-CB composite. At a first

Fig. 6. TMA patterns of the produced materials. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7. Heat capacity patterns of the produced materials. (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version ofthis article.)

Fig. 8. Post-burning surfaces of the PR-MWNT system (a and c) and PR-CBcomposite (b and d). (For interpretation of the references to colour in this figurelegend, the reader is referred to the web version of this article.)

M. Natali et al. / Composites: Part A 43 (2012) 174–182 179

glance, it is clearly visible how the burnt surface of the differentmaterials shows a different color. The brown color of the burnt sur-face of the PR-MWNT sample is strictly related to the presence ofthe metal precursors used in the production of MWNTs. This wasalso the color of the residual mass of the PR-MWNT compositeand of the pure MWNTs tested both in TGA (in air) and in a muffle.

These pictures clearly shows that in terms of the recession rate,the PR-MWNT composite exhibited higher erosion than the PR-CBsystem. The PR-MWNT samples underwent a visible removal ofmaterial – a crater – on a zone strongly confined under the plumeof the impinging flame. During the oxy-acetylene tests, from thesurface of the PR-MWNT samples, small flakes of the burning

material were peeled off. Such behavior was also found by Pattonet al. [17] for CNF-NRAMs. The PR-MWNT samples exhibited a verythin charred layer above the virgin material. Conversely, in the caseof the PR-CB composites, the flame produced a wide charred sub-strate having relatively good integrity: the flame did not produceany crater. Also, during the test, this material did not display anymacroscopic ejection of particles.

The bases of the PR-MWNT samples showed a very limitednumber of in-depth cracks propagating from the mantel to the cen-ter of the sample (Fig. 8a and c). On the contrary, the PR-CB com-position exhibited a remarkable cracking phenomena on a widesurface of the mantel as well as on the bases (Fig. 8b and d).According to this result, it can be concluded that the PR-CB systemsuffered a char penetration extended on a higher surface of themantel. Fig. 9a summarizes the proposed ablation mechanism forthe PR-MWNT recipe while Fig. 9b refers to the PR-CB composite.

The morphology of the burnt surfaces was also investigated bySEM analysis. At low magnification (Fig. 10a and b), the PR-CB

Fig. 9. Mechanism of the ablation process. (For interpretation of the references tocolour in this figure legend, the reader is referred to the web version of this article.)

180 M. Natali et al. / Composites: Part A 43 (2012) 174–182

revealed the macro cracks also identified with the optical images.Increasing the magnification, the top surface of the PR-CB evi-denced a morphology of the char characterized by micron(Fig. 10b) and sub-micron (Fig. 10c) sized bubble-like structures:both the closed and the open bubbles are clearly visible. Also, thecarbon black particles were not distinguishable from the charredlayer: the CB seemed completely integrated in the pyrolyzed ma-trix. A possible explanation for the formation of such structuresis the following. During the exposure to the hyper-thermal envi-ronment, the matrix degraded and produced gaseous products.Part of these gaseous were not able to reach the top surface. Whenthe internal pressure became sufficiently high, the gases diffusedthrough the char structures thus forming the bubble-like struc-tures visible on the surface. When the trapped gases reached a crit-ical pressure, the closed bubbles experienced a burst. Such abehavior is very common in many charring ablators [41]. The mor-phology of the zones beneath the surface was also investigated.Particularly, since the top surface char was compact and hard,the presence of open cracks offered the possibility to investigatethe nature of the inner layers. In general, the morphology of theseszones was very close to that one of the surface but, in some cases, asecond type of char was found. Fig. 10d shows a detail of this otherkind of charred substrate: the CB particles were clustered one each

Fig. 10. SEM analysis of the post-burning surfaces of the PR-CB. (For interpretation of thethis article.)

other to form a uniform and compact medium with no voids orbubbles. In this case, it could be speculated that the gases foundalternative ways to reach the surface, thus leaving these portionsof the char free of voids or defects.

The SEM study was also carried out on the post burning PR-MWNT system. At a low magnification and on the same lengthscale, if compared to the PR-CB system, the surface of the PR-MWNT samples showed a homogeneous morphology free of anymacro cracks (Fig. 11a): the ablation was a surface phenomenonwith a very thin transition zone (char) between the virgin andburnt material. Increasing the magnification (Fig. 11b and c), theSEM evidenced a char constituted by many irregular aggregatesmainly composed of entangled MWNTs. In this dense network ofcarbon nanotubes, apart from the undamaged and unburntMWNTs, the SEM analysis also revealed a series of bright spots thatcould not be found in the pre-burning specimens (Fig. 11c). Thesedots are the residue of the burnt MWNTs after exposure to theflame: they are clearly related to the metal precursors used to pro-duce the nanotubes. These residues are responsible for the browncolor of the burnt surface on the PR-MWNT samples. Moreover, aSEM analysis was performed in the empty spaces between themacroscopic aggregates visible in Fig. 11b: the morphology ofthese relatively inner zones of the char was the same of the com-pact regions of the char (Fig. 11d).

Fig. 12 shows the representative in-depth temperature profilesacquired at 5 (T1) and 10 mm (T2) from the mantel of the burntsamples. As it is possible to observe from the recorded tempera-tures, both the PR-MWNT temperatures showed a faster increasethan the PR-CB composite. This clearly suggests that the MWNTbased system exhibited higher thermal diffusivity a than the PR-CB composite. Moreover, it is noticeable that after about 40 s, thePR-MWNT T1 became lower than that of the PR-CB, whereas forthe T2 profiles no crossover can be observed. This can be explainedconsidering that the PR-MWNT system tended to stabilize its finaltemperature at a lower value. This represents a further confirma-tion of the fact that the thermal diffusivity of the PR-MWNT ishigher than that of the PR-CB.

The in-depth temperature profiles help one to understand theablation mechanism highlighted by the analysis of the post burntsurfaces. It is useful to consider the definition of the thermal

references to colour in this figure legend, the reader is referred to the web version of

Fig. 11. SEM analysis of the post-burning surfaces of PR-MWNT composite. (For interpretation of the references to colour in this figure legend, the reader is referred to theweb version of this article.)

Fig. 12. In-depth temperature profiles acquired at 5 mm (T1) and 10 mm (T2) fromthe surface of the samples. (For interpretation of the references to colour in thisfigure legend, the reader is referred to the web version of this article.)

M. Natali et al. / Composites: Part A 43 (2012) 174–182 181

diffusivity a ¼ k=q CP , where k is the thermal conductivity and q isthe density of the considered material. Even if it is not completelyproper to compare the data obtained at different heating rates, it isreasonable to exploit the CP results for the interpretation of theoxy-acetylene torch test. In the range of temperatures 30–450 �C,the CP of the PR-MWNT composite was higher than the CP of thePR-CB system. On the other hand, the density of both systemswas nearly the same. Accordingly, the denominator of the equationfor a of the PR-MWNT system has a higher value than the PR-CBsystem. As a consequence, since the temperature profiles showedthat the PR-MWNT system had a marked higher thermal diffusivitythan the PR-CB composite, it can be concluded that this differenceis directly related to the thermal conductivity: the PR-MWNT com-posites possess a higher thermal conductivity than the PR-CB sys-tem. At the beginning of this section, it was also pointed out thatfor the PR-MWNT sample, the torch produced a hole with a sizeapproximately equal to the diameter of the flame whilst for thePR-CB system there was no crater. For the latter system, the heatwas diffused on a wide charred surface, extended also beyondthe external corona of the flame. Accordingly, it can be hypothe-sized that the PR-MWNT system had a higher out of plane k thanthe PR-CB composite. For the PR-CB system, the heat was dissi-pated on a broader surface of the sample, producing a wide charredlayer.

Finally, it is possible to observe that the difference in tempera-ture between the T1 and the T2 profiles was smaller for the PR-MWNT recipe. This represents further evidence of higher thermalconductivity of this material with respect to the PR-CB.

It is also worth remarking that for the PR-MWNT material,during the test, the flame removed a significant amount of mate-rial. Although the thickness of the region between the flame andthermocouple decreased, the value of T1 did not exhibit any corre-sponding increase in temperature. This experimental evidence isrelated to the high k of the PR-MWNT recipe, but it could alsosupport the hypothesis that for the PR-MWNT system, a part ofthe heat was dissipated by re-radiation. Literature on the flameretardant properties of CNT reinforced matrices provides supportto this conclusion [24,25,42].

Loss of mass data - in terms of the difference between the preand post-test mass values of the samples – gave further evidenceof the above reasoning. In order to better understand the experi-mental results, the loss of mass data (DW) were normalized to

the initial weight of the samples, precisely DW ¼ Winitial�Wfinal

Winitial� 100.

According to the experimental results, if compared to the PR-CBrecipe (DW = 4.57 ± 0.08), the PR-MWNT (DW = 4.77 ± 0.30) sys-tem exhibited a slightly higher loss of mass. However, the mostinteresting consideration is the following. As previously high-lighted, the surfaces of PR-MWNT samples clearly experienced amore severe erosion than the PR-CB specimens. Since the amountof loss of mass is qualitatively of the same order of magnitudefor both materials, it means that the ablation mechanisms of thetwo materials is deeply different. For the PR-MWNT, the loss of

182 M. Natali et al. / Composites: Part A 43 (2012) 174–182

mass was merely due to the removal of material while, since thePR-CB did not erode appreciably, the loss of mass can be relatedto the higher degree of pyrolysis of the composite. Compared tothe PR-MWNT recipe, the PR-CB had a wider charred region inwhich the organic matrix was converted in char: the degradationof the phenolic matrix was accountable for the loss of weight inthe PR-CB composites.

4. Conclusions

In this work, the thermal and the ablative properties of highlyloaded CB and MWNT-based phenolic composites were studied.The TGA results obtained in a nitrogen atmosphere showed thatboth composites had superior thermal stability than the pristinematrix. The TGA patterns of the two composites were nearly thesame, indicating that at the evaluated filler load, the type of addi-tive only slightly affects the thermal stability of the matrix. In anair environment, for both composites, the presence of carbon fillersdid not lead to significantly improved resistance to oxidation. Theintroduction of CB and MWNTs in the phenolic matrix increaseddimensional stability. In the considered range of temperatures,the PR-MWNT recipe showed a higher CP than the PR-CB system.After exposure to the oxy-acetylene torch, the deep differences interms of an ablative response as a function of the filler were clearlyvisible. The PR-MWNT exhibited a higher erosion rate localizedabove the flame plume and higher thermal diffusivity than thePR-CB. The PR-MWNT system showed a very thin charred regioncovering the zone touched by the flame whilst the PR-CB was char-acterized by a thick pyrolyzed zone extended on a wider area,which effectively shielded the virgin material.

Acknowledgement

The authors wish to thank Marco Rallini for his contribution onthe SEM analysis.

References

[1] D’Aelio GF, Parker JA. Ablative plastics. New York: Marcel Dekker; 1971.[2] Schmidt D. Chemical propulsion exhaust. In: Rosato DV, Schwartz RT, editors.

Environmental effects on polymeric materials, environments, vol. 1. NewYork: Interscience Publishers; 1968. p. 413–86.

[3] Natali M, Torre L. Composite, ablatives. In: Encyclopedia of composites. NewYork: Wiley, in press.

[4] Donskoy AA. Elastomeric heat shielding materials for internal surfaces ofmissile engines. Int J Polym Mater 1996;31(1):215–36.

[5] Torre L, Kenny JM, Maffezzoli AM. Degradation behavior of a compositematerial for thermal protection systems. Part I – experimentalcharacterization. J Mater Sci 1998;33(12):3137–43.

[6] Torre L, Kenny JM, Maffezzoli AM. Degradation behavior of a compositematerial for thermal protection systems. Part II – process simulation. J MaterSci 1998;33(12):3145–9.

[7] Torre L, Kenny JM, Boghetich G, Maffezzoli AM. Degradation behaviour of acomposite material for thermal protection systems Part III Charcharacterization. J Mater Sci 2000;35(18):4563–6.

[8] Pilato L. Phenolic resins: a century of progress. Springer; 2010.[9] Miller GH, Robinson Jr GC. Investigation of fiber systems of ablative materials.

NASA CR-54722 TEI - TP 25; September 1965.[10] Vaia RA, Price G, Ruth PN, Nguyen HT, Lichtenhan J. Polymer/layered silicate

nanocomposites as high performance ablative materials. Appl Clay Sci1999;15:67–92.

[11] Koo JH. Polymer nanocomposites: processing, characterization, andapplications. New York: McGraw-Hill Professional; 2006.

[12] Bahramian AR, Kokabi M. Ablation mechanism of polymer layered silicatenanocomposite heat shield. J Hazard Mater 2009;166(1):445–54.

[13] Natali M, Monti M, Kenny JM, Torre L. Synthesis and thermal characterizationof phenolic resin/silica nanocomposites prepared with high shear rate-mixingtechnique. J Appl Polym Sci 2011;120(5):2632–40.

[14] Natali M, Monti M, Kenny JM, Torre L. A Nanostructured ablative bulkmoulding compound: development and characterization. Compos Part A–ApplS 2011;42(9):1197–204.

[15] Srikantha I, Daniel A, Kumar S, Padmavathib N, Singhb V, Ghosal P, et al. Nanosilica modified carbon-phenolic composites for enhanced ablation resistance.Scripta Mater 2010;63(2):200–3.

[16] Koo JH, Stretz H, Weispfenning JT, Luo ZP, Wootan W. Nanocomposite RocketAblative Materials: Processing, Microstructure, and Performance. Reston: AIAAPaper No.AIAA-2004-1996; 2004.

[17] Patton RD, Pittman Jr CU, Wang L, Hill JR. Ablation, mechanical and thermalconductivity properties of vapor grown carbon fiber/phenolic matrixcomposites. Compos Part A-Appl S 1999;30(9):1081–91.

[18] Monti M, Natali M, Petrucci R, Puglia D, Terenzi A, Valentini L, Kenny JM.Advanced fiber reinforced composites based on nanocomposite matrices. In:Nicolais L, Borzacchiello A, editors. Wiley Encyclopedia of composites. 2nd ed.2011. doi:10.1002/9781118097298.weoc025.

[19] Kashiwagi T, Du F, Winey KI, Groth KM, Shields JR, Bellayer SP, et al.Flammability properties of polymer nanocomposites with single-walledcarbon nanotubes: effects of nanotube dispersion and concentration.Polymer 2005;46:471–81.

[20] Kashiwagi T, Du F, Douglas JF, Winey KI, Harris RH, Shields JR. Nanoparticlenetworks reduced the flammability of polymer nanocomposites. Nat Mater2005;4:928–33.

[21] Cipiriano BH, Kashiwagi T, Raghavan SR, Yang Y, Grulke EA, Yamamoto K, et al.Effects of aspect ratio of MWNT on the flammability properties of polymernanocomposites. Polymer 2007;48:6086–96.

[22] Kashiwagi T, Mu M, Winey K, Cipriano B, Raghavan SR, Pack S, et al. Relationbetween the viscoelastic and flammability properties of polymernanocomposites. Polymer 2008;49:4358–68.

[23] Liu J, Rinzler AG, Dai HJ, Hafner JH, Bradley RK, Boul PJ, et al. Fullerene pipes.Science 1998;280:1253–6.

[24] Zhao Z, Gou J. Improved fire retardancy of thermoset composites modifiedwith carbon nanofibers. Sci Technol Adv Mater 2009;10:015005.

[25] Rahatekara SS, Zammarano M, Matko S, Koziol KK, Windle AH, Nyden M, et al.Effect of carbon nanotubes and montmorillonite on the flammability of epoxynanocomposites. Polym Degrad Stabil 2010;95:870–9.

[26] Simon P, Cibulková Z. Measurement of heat capacity by differential scanningcalorimetry. Thermophysic 2002, 6th meeting of the thermophysical society.Working Group of the Slovak Physical Society. Kocovce, West Slovakia.

[27] ASTM E-285-80. Standard test method for oxyacetylene ablation testing ofthermal insulation materials. Annual Book of ASTM Standards 2008, vol. 15.03,Space Simulation; Aerospace and Aircraft; Composite Materials.

[28] ASTM Standard E457-08. Standard Test Method for Measuring Heat-TransferRate Using a Thermal Capacitance (Slug) Calorimeter. Annual Book of ASTMStandards 2008, vol. 15.03, Space Simulation; Aerospace and Aircraft;Composite Materials.

[29] Pulci G, Tirillò J, Marra F, Fossati F, Bartuli C, Valente T. Carbon-phenolicablative materials for re-entry space vehicles: manufacturing and properties.Compos Part A-Appl S 2010;41:1483–90.

[30] Kim ES, Kim EJ, Shim JH, Yoon J-S. Thermal stability and ablation properties ofsilicone rubber composites. J Appl Polym Sci 2008;110:1263–70.

[31] Marra F, Pulci G, Tirillò J, Bartuli C, Valente T. Numerical simulation of oxy-acetylene testing procedure of ablative materials for re-entry Space vehicles.In: Proceedings of the Institution of Mechanical Engineers, Part L: J Mater:Design Appl 2011; 225(1): 32–40.

[32] Knop A, Pilato LA. Phenolic resins: chemistry, applications andperformance. Berlin Heidelberg (NY): Springer-Verlag; 1985.

[33] Fyfe CA, McKinnon MS, Rudin A, Tchir WJ. Investigation of the mechanisms ofthe thermal decomposition of cured phenolic resins by high resolution 13C CP/MAS solid-state NMR spectroscopy. Macromolecules 1983;16:1216–9.

[34] Morterra C, Low MJD. I.R. studies of carbons-VII. The pyrolysis of a phenol-formaldehyde resin. Carbon 1985;23:525–30.

[35] Lum R, Wilkins CW, Robbins M, Lyons AM, Jobes RP. Thermal analysis ofgraphite and carbon-phenolic composites by pyrolysis-mass spectrometry.Carbon 1983;21(2):111–6.

[36] Blazso M, Toth T. Sequence of phenolic units in phenol-formaldehydespolycondensates studied by pyrolysis-gas chromatography/massspectrometry. J Anal Appl Pyrol 1991;19:251–63.

[37] Conley R, Bieron J. A study of the oxidative degradation of phenol-formaldehyde polycondensates using infrared spectroscopy. J Appl Polym Sci1963;7:103–17.

[38] Morterra C, Low MJD. Infrared studies of carbon. 8. The oxidation of phenol-formaldehyde chars. Langmuir 1985;1(3):320–6.

[39] Rogers ME Long TE, Lin-Gibson S, Riffle JS. Chemistry and Properties ofPhenolic Resins and Networks. In: Synthetic methods in step-growthpolymers. Wiley, 2003. p. 418-425 (Chapter 7).

[40] Ohlhorst CW, Vaughn WL, Ransone PO, Tsou H-T. Thermal conductivitydatabase of various structural carbon-carbon composite materials. NASAtechnical memorandum 4787, Langley Research Center. Hampton, 1997.

[41] Ho WK, Koo JH, Ezekoye OA, Thermoplastic Polyurethane ElastomerNanocomposites: Morphology, Thermophysical, and Flammability Properties,Journal of Nanomaterials, 2010, Article ID 583234, 11 pages, doi: 10.1155/2010/583234.

[42] Kuan CF, Chen WJ, Li YL, Chen CH, Kuan HC, Chiang CL. Flame retardance andthermal stability of carbon nanotube epoxy composite prepared from sol–gelmethod. J Phys Chem Solids 2010;71:539–43.