Study of carbon fibres and carboncarbon ... - Wiley Online Library

JMI_974.fm Page 21 Monday, January 7, 2002 10:35 AM

Journal of Microscopy, Vol. 205, Pt 1 January 2002, pp. 21– 32 Received 2 May 2001; accepted 16 July 2001

Study of carbon fibres and carbon–carbon composites by scanning thermal microscopy Blackwell Science Ltd

C. BLANCO, S. P. APPLEYARD & B. RAND Department of Materials, University of Leeds, Leeds LS2 9JT, UK

Key words. Carbon–carbon composites, carbon fibres, scanning thermal microscopy, thermal conductivity.

Summary Scanning thermal microscopy (SThM) is a relatively new technique based on atomic force microscopy in which the tip is replaced by an ultra-miniature temperature probe. This paper reports on a preliminary investigation of the application of SThM in the characterization of the thermal properties of carbon fibres and carbon–carbon (CC) composites. The technique enabled a comparative study to be made of discrete fibre and matrix thermal properties in a series of model unidirectional composites. The thermal images revealed a marked increase in thermal conductivity of the matrix with increasing temperature of treatment and hence confirmed the development of a highly ordered carbon matrix. The results were in qualitative agreement with previously determined values of thermal conductivity from which the separate values of fibre and matrix thermal conductivity had been derived. The technique was also applied to the characterization of samples of unknown processing history, enabling an estimation to be made of the heat treatment and type of the fibres and matrix present in the composite. It was concluded that SThM promises to be a powerful technique for the study of the thermal properties of CC composites and carbon fibres, as it uniquely enables variations in local thermal conductivity to be detected and resolved. Absolute quantification of the technique remains the key to its future widespread acceptance in materials characterization. Introduction Carbon fibre reinforced carbon matrix (CC) composites form an important group of materials that can have high specific strength and stiffness at elevated temperatures, high thermal conductivity and excellent dimensional stability. They are widely used in high performance applications in the military and aerospace industry and, in considerable quantities, as Correspondence: Professor Brian Rand, Department of Materials, University of Leeds,

Leeds

LS2

9JT,

UK.

Tel.:

+44

(0)113 2422531; e-mail: [email protected] Received 2 May 2001; accepted 16 July 2001

© 2002 The Royal Microscopical Society

(0)113 2332536;

fax:

+44

friction elements in aircraft braking systems (Fitzer, 1987; Savage, 1993). Recently, they have attracted attention as candidate materials for thermal management in several new applications such as heat sinks, substrates for electronic elements and plasma facing components of fusion devices (Bertram et al., 1992; Bowers et al., 1994; Fitzer & Manocha, 1998). The success of CC composites in these new applications relies on a better understanding of the factors controlling their thermal and mechanical properties in order to be able to optimize processing and better predict their properties. The use of straightforward characterization techniques that enable properties to be directly related to structure should expedite this process. CC composites have a heterogeneous structure consisting of fibres, matrix and pores. Both carbon fibres and the matrix may exhibit a variety of microstructures, from amorphous to graphitic, depending on the precursor and processing conditions (Rand, 1993). These microstructures relate to the size, degree of order and orientation of the aromatic lamellae. The different microstructures will also possess varying thermal properties. The thermal conductivity in graphite is dominated by the layer plane conduction (Kelly, 1981). The absolute magnitude of the thermal conductivity is determined by the crystallite size and the orientation of the layer planes (Taylor et al., 1993). For highly ordered graphite, the thermal conductivities parallel to basal planes may be two orders of magnitude higher than those in the perpendicular direction (Fitzer & Manocha, 1998). The thermal conductivity of carbons increases significantly with heat treatment, due to the development of a higher degree of order and greater extension of the aromatic lamellae. In summary, the thermal properties of CC composites are controlled by a complex combination of many factors. A significant advantage of this, however, is the potential to produce materials with tailor-made properties by the proper choice of constituents, their configuration and processing conditions (Fitzer & Manocha, 1998). In the last decade, new physical characterization techniques have been developed that are based on the atomic force microscope and scanning tunnelling microscope. These


22

C. BLANCO ET AL.

techniques enable the measurement of physical properties with a lateral resolution ranging from micrometres to nanometres (Pool, 1990; Wickramasinghe, 1990). Scanning thermal microscopy (SThM) is one such technique, which is capable of detecting variations in the thermal properties of a sample, such as thermal diffusivity and thermal conductivity, with a high spatial resolution (Majumdar, 1999). In the present study, the measurements are carried out by using an ultraminiature temperature probe that both provides a source of heat and, together with associated electronic circuitry, detects the thermal response (Dinwiddie et al., 1994; Hammiche et al., 1996a). The ‘thermal images’ are generated from the difference in heat flux from the tip while scanning across different regions of the sample, which depends on the thermal conductivity/diffusivity of each region. SThM has been applied successfully in the characterization of polymeric and pharmaceutical materials, but its application in the wider field of materials science has hitherto been limited, especially in the case of materials with high thermal conductivity. A significant limitation of the current technique is the qualitative nature of the information obtained, absolute quantification remaining the key to the future widespread acceptance of the technique in materials characterization. Recent literature describes different ways of attempting to develop quantitative thermal microscopy (Fiege et al., 1999; Gorbunov et al., 2000a,b). With regard to the characterization of CC composites, the technique is expected to offer possibilities of the direct determination of relative contributions of the fibre and matrix components to the bulk composite thermal properties. This preliminary study focuses on the characterization of a series of model unidirectional CC composites made from high modulus PAN-based fibres and pitch-based matrix, after heat treatment at different temperatures, up to 2400 °C. The structure of these fibres will not be affected by the thermal treatment, thus facilitating the study of the changes occurring in the matrix. These composites are included in a broader study, the aim of which is the systematic correlation between structure and thermal properties of CC composites. The results obtained with SThM are compared with values obtained from bulk thermal conductivity. This paper also highlights several of the generic experimental problems and limitations that are characteristic of the technique. For example, the complicating effects of surface topography on the thermal image contrast are discussed. Experimental Materials examined The materials included in this study comprise a series of commercially available carbon fibres, a series of model unidirectional (1-D) CC composites that had been heat treated at different temperatures and a number of CC composites of unknown processing history. The carbon fibres included PANbased fibres (Toray: T300, T800 and M40J, www.toray.com)

Table 1. Main properties of commercial carbon fibres.

Fibre

Thermal conductivity (W m–1 K–1)

Tensile strength* (GPa)

Young Modulus* (GPa)

T300 T800 M40J P55 P100 K1100

7** 10** 65** 120* 520* 900–1000*

3.53 5.49 4.41 1.90 2.41 3.10

230 294 377 379 758 965

* Data sheet. **Measured by Taylor et al. (1993).

and mesophase pitch derived fibres (BP-Amoco: P55, P100 and K1100, www.bpamoco.com) and provide a wide range of thermal conductivity (Table 1) as a result of the nature of the precursor and the heat treatment temperature used in their processing. The model 1-D CC materials in this study comprise high modulus PAN-based fibres (Courtaulds: Grafil HM-370, www.courtaulds.com) with carbon matrix derived from a thermoplastic petroleum pitch (Ashland Aerocarb 75, www.ashland.com). The composites were fabricated by wetwinding (with a solution of pitch), followed by lay-up, moulding and carbonization at 1000 °C. Each carbonized composite was densified by means of three cycles of vacuum impregnation with melted pitch and carbonization at 1000 °C. Finally, samples of the composite were heat-treated at temperatures ranging up to 2400 °C (Appleyard et al., 1998). The resultant composites were named HMP1000, HMP1500 and HMP2400, where the numeral signifies the final heat treatment temperature. Due to the nature of the fibres used, only the matrix is expected to undergo structural changes during heat treatment. This will therefore facilitate the study. Sample preparation The flatness of the samples is a critical factor for the SThM studies. Polished transverse sections of the fibres and composites were therefore prepared following a procedure used for study by optical microscopy. Each sample was vacuumimpregnated with a Struers Epofix low viscosity epoxy resin (Struers, www.struers.com) and polished using a Buehler Motopol 12 automatic grinding/polishing machine (Buehler, www.buelher.com). The polishing route involved, first, planar grinding with 400 grit SiC grinding paper, followed by fine grinding with 9 µm diamonds on a Buehler Metlap 4 cloth (oil lubrication, 6 min, 2 kg load per sample, 60 rev min–1). The samples were then polished on Buehler Texmet 1000 cloths using Buehler Micropolish II 0.3 µm alumina suspension (5 min, 1 kg load per sample, 60 rev min–1), and then Buehler Masterpolish 0.02 µm alumina suspension (1 min, 1 kg load per sample, 60 rev min–1). © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21– 32


STHM OF CARBON FIBRES AND CARBON–CARBON COMPOSITES

23

Fig. 1. Schematic diagram of the scanning thermal microscope set-up.

Scanning thermal microscopy (SThM) The SThM apparatus is commercially available and comprises an atomic force microscope (Thermomicroscopes Explorer, www.thermomicroscopes.com), fitted with a Wollaston thermal probe (Wollaston, www.thermomicroscopes.com) (Fig. 1). The probe provides the heat source and, together with associated electronic circuitry, detects the thermal response. It consists of a 75 µm diameter silver wire with a 3 µm diameter core of platinum/10% rhodium. The probe is formed by etching the silver away over a length of about 200 µm to reveal the platinum core. This exposed platinum area is bent into a loop of suitable shape that forms the scanning tip (Hammiche et al., 1996a). The thermal probe can be used to obtain topographic information, in the mode of conventional contact atomic force microscopy. In this case, the tip is scanned across the sample using a pair of piezo-electric elements aligned in the x- and y-axes. As the height of the sample changes, the deflection of the tip in the z-axis is monitored by an optical lever formed by reflecting a laser beam from a mirror at the back of the cantilever onto a four-quadrant photodetector. The tip is then moved vertically up and down by a feedback loop connected to a z-axis piezo which maintains a constant contact force, thus providing the height of the sample at each x, y position (Majumdar, 1999). Simultaneously to the force feedback, a thermal feedback can be used. In this case, an electric current is passed through the wire and heats the tip to a constant temperature by Joule heating. The probe forms one of the arms of a Wheatstone bridge (Fig. 1), whose circuit uses a feedback loop to adjust the bridge voltage to maintain the bridge balance, thus keeping the temperature of the probe constant. As the probe contacts the sample surface, heat flows from the probe to the sample. In the absence of feedback, this flow of heat reduces the probe temperature, decreasing its resistance and causing bridge imbalance. The feedback senses this shift and increases the © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21–32

voltage applied to the bridge, so that the probe resistance returns to the set point (i.e. corresponding to the set temperature). While the probe is scanned across the sample, variations in the heat flow out of the probe are measured by monitoring the bridge voltage, which is then used to create contrast based on the apparent local thermal conductivity of the sample (Hammiche et al., 1996a; Gomès et al., 1999). The operating temperature of the probe can be selected by adjusting the value of a control resistance integrated in the Wheastone bridge. The thermal probe may also be used in other modes such as constant current and temperature modulation (Hammiche et al., 1996b,c), although these are not applied here. A relatively high tip temperature of 150 °C was used to obtain the thermal images presented in this paper, as lower temperatures did not yield good thermal contrast for some of the composite samples. The colour intensities in the thermal images were normally scaled according to the difference in response between the constituent parts of each image, lighter shades corresponding to areas of higher dissipation of heat than darker ones. In the case of the carbon fibres (Fig. 2), the intensity scale was set the same for the set of PAN-based fibres and again for the set of pitch-based fibres, in order to allow a direct comparison between the images in each set. Due to the large difference in thermal conductivity between the two groups of fibres (Table 1), a different intensity scale was used for each set of images. It should be mentioned that, in both cases, the images were obtained using the same tip with the aim of keeping the conditions as similar as possible. The colour intensities in the topographic images were similarly scaled according to the difference between the highest and lowest features in the image. A minimum of 10 images per sample was obtained in order to ensure that results were representative of the whole sample. The scanned areas ranged from (100 µm)2 down to (12 µm)2 and were scanned at rates of 100 µm s–1 to 12 µm s–1, respectively. The image resolution was 300 points per line.


24

C. BLANCO ET AL.

Fig. 2. Thermal images of commercial carbon fibres: (a) PAN-based (1) T300; (2) T800; (3) M40J; and (b) pitch-based (1) P55; (2) P100 and (3) K1100.

Results and discussion The results in this study are presented in four sections. These are (i) commercial carbon fibres with a wide range of thermal conductivity; (ii) model unidirectional CC composites with different heat-treatment temperatures; (iii) experimental problems and limitations to the technique; and finally (iv) CC composites of unknown processing history. Carbon fibres Figure 2 illustrates SThM images for the range of carbon fibres under investigation. As explained in the experimental section, the contrast in the images corresponding to the PAN-based fibres (Fig. 2a) was based on the same intensity scale (5.00– 6.00 mW), to enable a direct comparison between them. Similarly, the same colour scale was used to compare the pitch-based fibres (Fig. 2b), but in this case the scale range was 5.25–8.00 mW, due to the higher conductivity of these fibres. All fibres in Fig. 2 are lighter than the epoxy resin in which they are embedded, indicating the lower thermal conductivity of the resin. In carbon fibres, the aromatic lamellae are aligned parallel to the fibre axis, but the extension and degree of order varies

significantly for the different fibres, and with these, their thermal conductivity. PAN-based fibres have low crystallinity, and this only improves slightly with heat treatment. This is reflected in their relatively low thermal conductivity (Table 1). The thermal images in Fig. 2(a) show a change in the colour intensity of the fibres that follows the same trend as their thermal conductivity, the fibres becoming progressively brighter. In pitch-based fibres, by contrast, the extension and degree of order of the aromatic lamellae can be significantly improved with heat treatment. These result in a great increase in their thermal conductivity, which can vary from 120 to 900– 1000 W m–1 K–1 (Table 1). The thermal images in Fig. 2(b) show a more dramatic change in the colour intensity of these fibres than for the PAN-based fibres (in Fig. 2a), in agreement with their thermal conductivity. From these results, the information obtained from SThM seems to be consistent with the absolute values of thermal conductivity. However, this information is restricted to comparative purposes, due to several limitations in the technique that are explained below. It is worth noting here that the resin used to embed the fibres is the same in all cases, and therefore it should be seen with the same colour contrast. However, Fig. 2(a) shows a variation in the resin colour for image 3, which could only be attributed to drifts in the equipment. © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21– 32



Model unidirectional CC composites SThM enables direct comparison between the thermal properties of the matrix component and the fibres. Figure 3 shows the thermal and topographic images obtained for the composites under study. Consider the composite heat treated to 1000 °C (HMP1000) (Fig. 3a-1). The conductivity of the fibres is apparently higher than that of the matrix, as the fibres appear brighter in the thermal images. Relatively good contrast between components is observed. On their conversion into carbons, pitches go through a nematic discotic liquid crystal phase (mesophase) in which the disc-like aromatic molecules in the pitch stack, forming domains with preferred molecular orientation. Mesophase is the first step towards the formation of graphitic structures (Rand, 1993). In the presence of carbon fibres, the mesogenic molecules tend to align with the fibre surfaces, resulting in the development of a sheath-like structure around the fibres, with lamellae parallel to the fibre axis (White & Zimmer, 1983). The degree of order in the pitch-based matrix is expected to increase on increasing the heat treatment temperature, resulting in more perfectly aligned sheath structure in the matrix with basal planes parallel to the fibre axis (Fig. 4). This further development of order in the matrix in turn results in an increase in the matrix thermal conductivity in the direction parallel to the fibre axis. No such changes are expected for the fibres in this series of composites, as they had been previously heat-treated at 2500 °C during their manufacturing. This is in agreement with the resultant thermal image of the sample heat-treated at 1500 °C (Fig. 3a-2), which shows that, although the matrix is still slightly darker than the fibres, the contrast between them is much less than for composite HMP1000. The thermal images obtained for the composite heat-treated at higher temperature (HMP2400) show the matrix now lighter than the fibres (Fig. 3a-3), as a result of the higher degree of order developed in the matrix with the heat treatment (Fig. 4b). The topographic images (Fig. 3b), corresponding to the same areas as the thermal images, illustrate the effects of relief created during the polishing. Note, however, that the extent of the relief (i.e. peak to trough) is under 200 nm. There are several cases in which the information obtained from the topographic images is essential in order to understand the thermal images, as will be discussed in the next section. The information obtained from the thermal images of the CC composites was compared to values of fibre and matrix thermal conductivity reported by Appleyard et al. (1999). These values were derived from bulk measurements of thermal diffusivity using the laser flash method. As it is not possible to perform direct measurements on the matrix, the thermal conductivity of the matrix composites was estimated by the rule of mixtures, using the measured values for the composite and the fibres, taking into account the volume fraction of fibres in the composite (Table 2). The conductivity of the fibres (90 W m–1 K–1) remained constant with increasing © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21–32

25

Table 2. Measured and estimated thermal conductivity of the composites and components at room temperature, determined from the laser flash method for thermal diffusivity (from Appleyard et al., 1999).

Sample

Composite (W m–1 K–1)

Fibre (W m–1 K–1)

Matrix (W m–1 K–1)

HMP1000 HMP1500 HMP2400

55 60 145

90 90 90

40 50 270

heat treatment temperature, whereas the calculated conductivity for the matrix increased from being lower than the fibres in composites HMP1000 and HMP1500 (40 W m–1 K–1 and 50 W m–1 K–1, respectively), to being significantly higher for composite HMP2400 (270 W m–1 K–1). The information provided by SThM was thus proved to be consistent with that obtained from calculations based on thermal diffusivity measurements. Experimental problems and artefacts The examples in the previous section show how SThM can be used to detect the presence of regions of different thermal conductivity in CC composites. It must be noted, however, that the technique has some major limitations. The flow of heat is affected by the contact area between the tip and the sample, which is determined by several factors (Hammiche et al., 1996a; Majumdar, 1999), such as the size of the thermal tip. Wollaston probes are made manually and so, inevitably, the contact area varies from tip to tip. In addition, the shape of a given tip may be altered with usage, resulting in a change in the contact area and, hence, heat flow. This is a significant limitation as, in general terms, the information obtained from the different images cannot be compared directly. Comparisons can only be established when the same tip is used, provided it is not altered during the scanning. The shape of the tip determines the lateral resolution of both the thermal and topographic images (Pollock & Hammiche, 2001). The resolution of the tips is reportedly in the region of 1 µm (Hammiche et al., 1996c; Sano et al., 1997). We have found that significant differences in the quality of the images are obtained with nominally identical tips. Some authors have also described how contamination from the material under study can be accidentally attached to the tip itself during the scanning, in some cases resulting in a significant improvement of the image resolution (Sano et al., 1997). New batch-microfabricated thermal tips are currently being developed and studied (Hammiche et al., 2000; Pollock & Hammiche, 2001). They promise not only to significantly increase the image resolution but also to provide a more systematic response. The other factor that affects the contact area between the tip and the sample is the topography of the sample surface (Hammiche et al., 1996a; Moon et al., 2000). In CC composites, the main topographic features are voids and interfaces


26

C. BLANCO ET AL.

Fig. 3. (a) Thermal images and (b) topographic images of model unidirectional CC composites: (1) HMP1000; (2) HMP1500 and (3) HMP2400.

© 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21– 32



27

Fig. 4. SEM images of a CC composite showing the effect of heat treatment on the degree of order in the matrix: (a) 1500 °C and (b) 2400 °C.

that exist between each phase. Pores and cracks are a characteristic feature of CC composites that arise from the combined effects of matrix shrinkage, incomplete densification and thermal expansion mismatch between the fibre and matrix components. In samples that had been vacuum-impregnated in resin most of the pores will be filled with resin. Some closed pores, however, will be opened during the polishing process. The infiltrated resin will appear as an additional element in the image and may lead to difficulties in the interpretation. The effect of polishing relief was minimized to the extent that © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21–32

its influence on the thermal images was considered to be minimal. Some examples of these features mentioned above can be observed in the topographic images of the composite HMP1500 (Fig. 5a). The dark regions observed correspond to debonded areas between matrix and fibres (position 1) and to cracks in the matrix (position 2). These regions appear in bright colours in the thermal image (Fig. 5b). At these regions, the contact area between the probe and the sample is likely to be larger, as the tip moves down the side of the pore/fissure. Now not only the apex of the tip is in contact with the sample,


28

C. BLANCO ET AL.

Fig. 5. (a) Topographic and (b) thermal image of a composite showing debonded areas between matrix and fibre (position 1) and matrix cracks (position 2).

but also there is a lateral contact with heat dissipating surfaces and the flow of heat from the tip may increase. These regions may therefore be wrongly deduced to be areas of apparent high thermal conductivity, when in fact they are voids, if care is not exercised in the interpretation of data. The extent to which these effects contribute to alter the thermal image depends on the size and shape of tip in relation to the feature. Clearly, the interpretation of the thermal images must be done with care and always using the corresponding topographic image to correctly account for differences in image contrast. Application to the study of samples of unknown processing history SThM was used to study four CC composites of unknown processing history. Figure 6(a) shows thermal images of three of these composites. The topographic images corresponding to the same areas were also studied in order to check for any possible topographic artefact that could lead to a misinterpretation of the thermal image. Based on the studies of carbon fibres and model CC composites discussed in previous sections, the images in Fig. 6(a) suggest that the samples have all been heat-treated above 2000 °C as, in all three cases, the matrix shows lighter colours than the fibres. Fibre contours are welldefined in samples 1 and 3. These are typical for fibres that do not graphitize and therefore they show good contrast with the apparently graphitized matrix. It is interesting to note that the fibres have a core of apparently lower conductivity that may reflect differences in texture and crystallinity across the crosssection. The edges of the fibres in sample 2 apparently have a similar thermal conductivity to the matrix and there is not a clear contrast between them. It is likely, therefore, that both fibre and matrix are derived from pitch. The matrix in sample 3 shows areas of lower conductivity, similar to that of the

fibres, and areas of higher conductivity. This suggests that this is a resin-based matrix that has been graphitized. Although in principle resins are non-graphitisable materials, it is well known that narrow regions of highly orientated structure can be developed at the fibre–matrix interface due to strains in these regions (Hishiyama et al., 1974; Manocha, 1994). The higher degree of order in these regions makes them more thermally conductive than the non-graphitic, amorphous regions. Optical microscopy (Fig. 6b) was used to corroborate the information deduced from the thermal images. Samples 1 and 3 in fact correspond to composites made with isotropic fibres, probably PAN-based fibres, and sample 2 is made from pitchbased fibres, anisotropic under the polarized light microscope. It is also clear from the images that both samples 1 and 2 have graphitisable matrices and, therefore, after heat-treatment at high temperatures, they show light colours in the thermal image. However, the thermal scanning microscopy fails to distinguish whether it is a chemical vapour deposition (CVD) matrix (sample 1) or a pitch-based matrix (sample 2). Differences between the two types of matrix can be seen clearly by optical microscopy, sample 1 showing the typical microstructure of pyrolytic carbon highly orientated around the fibres. In the optical micrograph corresponding to sample 3, it is possible to identify isotropic and anisotropic regions in the matrix, which is typical of a graphitized resin-based matrix composite. Figure 7 shows an example of a CC composite which contains two types of matrix. The optical micrographs (Fig. 7a) show an edge of the material, where large areas of resin-based matrix can be observed (positions A), that has been coated by a CVD layer (positions B), clearly identified by the anisotropic pyrolitic structures. The fibres, isotropic, are likely to be PANbased. The thermal image in Fig. 7(b-1) shows an area of high fibre density, the fibres having higher apparent conductivity than the matrix. The kidney shape of the fibres suggests that © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21– 32



Fig. 6. (a) Thermal images and (b) optical micrographs of CC composites of unknown process history. © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21–32

29


30

C. BLANCO ET AL.

Fig. 7. (a) Optical micrographs and (b) thermal images of a CC composite with resin-based and CVD matrix. Position A: resin matrix; B: CVD layer; C: CVD layer and D: resin matrix.

they are PAN-based fibres. Close to a porous region, fibres surrounded by a CVD layer can be observed (position C), together with an area of different texture (position D) that could be related to the resin-based matrix mentioned before. This is more evident in the following image (Fig. 7b-2), where the resin-based and CVD regions can be distinguished by their

different texture (resin-based regions appear more plain, whereas CVD regions seem to have more texture, possibly due to their non-graphitisable and graphitisable nature, respectively). The shape of each of these regions can easily be identified from the optical micrographs. The fact that the fibres are much lighter than both the resin-based and the CVD matrix © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21– 32



suggests that this sample has not been graphitized. Graphitization would be expected to cause two effects. The thermal conductivity of the CVD matrix should increase significantly and it would therefore appear much brighter, whereas stress graphitization would probably lead to some brighter regions appearing. Conclusions Scanning thermal microscopy using Wollaston thermal probes has been used successfully to study the thermal properties of carbon fibres and CC composites. The technique allows the study of small samples with easy preparation procedures, as samples polished following standard routes for optical microscopy were found to be adequate. The technique has demonstrated that the relative contributions of different components to bulk thermal properties can be evaluated. The results obtained were in qualitative agreement with measured thermal conductivity in the bulk materials. The technique has also been found to be useful for the characterization of composites of unknown processing history, as it was possible to assess whether samples contain graphitic components and also to obtain information about the nature of the matrix and fibres used. Some limitations to the technique have been described, which mainly relate to the effect that the topographic features have on the thermal image and the resolution of the images. The latter is determined by the size and shape of the tip, which varies from tip to tip and with usage. The amount of heat lost from the tip to the sample also changes with the variation in the contact area between them. In addition, it is not possible at the moment to know the volume of sample that is being heated and the real temperature profile within the sample. SThM has proven to be a useful tool in the study of the thermal properties of composite materials. However, the qualitative nature of the information obtained from the thermal images is a significant limitation. A better understanding of the mechanisms involved in the heat transfer from the tip to the sample, the development of new methods of sensing and measuring the thermal properties and the development of new thermal probes of smaller size and a more consistent response are some of the areas in which several research groups are making efforts in order to be able to develop scanning thermal microscopy as a quantitative technique. Acknowledgements The authors would like to thank R. Taylor and H. Q. Li (UMIST, Manchester, UK) for the thermal conductivity measurements (Table 2) and Lydia Laffont (CEMES/CNRS, Toulouse, France) for the preparation of the commercial carbon fibres for the SThM studies. C. Blanco acknowledges the EU for a postdoctoral Marie Curie Fellowship (HPMF-CT-1999-00233). © 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21–32

31

References Appleyard, S.-P., Zhang, E. & Rand, B. (1998) The use of coke fillers to modify the structure of 1D composites for thermal properties studies. Extended Abstracts Eurocarbon ’98 , Vol. I, pp. 549–550. Strasbourg, France. Appleyard, S.-P., Zhang, E., Rand, B., Li, H.-Q. & Taylor, R. (1999) Control of thermal properties of CC composites by optimisation of structure. Extended Abstracts 24th Biennial Conference on Carbon, Vol. I, pp. 46 – 47. Charleston, USA. Bertram, A., Beasley, K. & De la Torre, W. (1992) An overview of navy composite developments for thermal management. Naval Engineers J. 276 – 285. Bowers, D.-A., Davis, J.-W. & Dinwiddie, R.-B. (1994) Development of 1-D carbon composites for plasma-facing components. J. Nuclear Materials, 212 –215, 1163–1167. Dinwiddie, R.-B., Pylkki, R.-J. & West, P.-E. (1994) Thermal conductivity contrast imaging with a scanning thermal microscope. Thermal Conductivity, 22 (ed. by T. W. Tong), pp. 668– 677. Technomic, Lancaster, PA. Fiege, G.-B.-M., Altes, A., Heiderhoff, R. & Balk, L.-J. (1999) Quantitative thermal conductivity measurements with nanometre resolution . J. Phys. D: Appl. Phys, 32, L13–L17. Fitzer, E. (1987) The future of carbon–carbon composites. Carbon, 25, 163–190. Fitzer, E. & Manocha, L.-M. (1998) Carbon Reinforcements and Carbon/ Carbon Composites. Springer-Verlag, Berlin. Gomès, S., Trannoy, N. & Grossel, P. (1999) DC thermal microscopy: study of the thermal exchange between a probe and a sample. Meas. Sci. Technol. 10, 805 – 811. Gorbunov, V.-V., Fuchigami, N. & Tsukruk, V.-V. (2000a) Microthermal analysis with scanning thermal microscopy. I. Methodology and experimental. Probe Microsc. 2, 53– 63. Gorbunov, V.-V., Fuchigami, N. & Tsukruk, V.-V. (2000b) Microthermal analysis with scanning thermal microscopy. II. Calibration, modelling, and interpretation. Probe Microsc. 2, 65 –75. Hammiche, A., Bozec, L., Conroy, M., Pollock, H.-M., Mills, G., Weaver, J.-M.-R., Price, D.-M., Reading, M., Hourston, D.-J. & Song, M. (2000) Highly localized thermal, mechanical and spectroscopic characterization of polymers using miniaturized thermal probes. J. Vac. Sci. Technol. B. 18, 1322 –1332. Hammiche, A., Pollock, H.-M., Song, M. & Hourston, D.-J. (1996a) Subsurface imaging by scanning thermal microscopy. Meas. Sci. Technol. 7, 142 –150. Hammiche, A., Reading, M., Pollock, H.-M., Song, M. & Hourston, D.-J. (1996c) Localized thermal analysis using a miniaturized resistive probe. Rev. Sci. Instrum. 67, 4268 – 4274. Hammiche. A., Hourston, D.-J., Pollock, H.-M., Reading, M. & Song, M. (1996b) Scanning thermal microscopy: subsurface imaging, thermal mapping of polymer blends, and localized calorimetry. J. Vac. Sci. Technol. B, 14, 1486 –1491. Hishiyama, M., Inagaki, M., Kimmura, S. & Yamada, S. (1974) Graphitization of carbon fiber/glassy carbon composites. Carbon, 12, 249 – 258. Kelly, B.-T. (1981) Physics of Graphite. Applied Science Publishers, London. Majumdar, A. (1999) Scanning thermal microscopy. Annu. Rev. Mater. Sci. 29, 505 –585. Manocha, L.-M. (1994) The effect of heat-treatment temperature on the


32

C. BLANCO ET AL.

properties of polyfurfuryl alcohol based carbon–carbon composites. Carbon, 32, 213 – 223. Moon, I., Androsch, R., Chen, W. & Wunderlich, B. (2000) The principles of micro-thermal analysis and its application to the study of macromolecules. J. Thermal Anal. Calorimetry, 59, 187– 203. Pollock, H.-M. & Hammiche, A. (2001) Micro-thermal analysis: techniques and applications. J. Appl. Phys, 34, R23–R53. Pool, R. (1990) The children of the STM. Science, 247, 634 – 636. Rand, B. (1993) Matrix precursors for carbon– carbon composites. Essentials of Carbon–Carbon Composites (ed. by C. R. Thomas), pp. 67–102. The Royal Society of Chemistry, Cambridge.

Sano, M., Yudasaka, M., Kikuchi, R. & Yoshimura, S. (1997) Anomalous resolutions in scanning thermal microscopy of graphite. Langmuir, 13, 4493 – 4497. Savage, G. (1993) Carbon-carbon Composites. Chapman and Hall, London. Taylor, R., Turner, S.-P., Garner, K. & Jiang, X.-X. (1993) Thermal conductivity of carbon fibres. High Temperatures-High Pressures, 25, 443–350. White, J.-L. & Zimmer, J.-E. (1983) Mesophase alignment within carbon-fiber bundles. Carbon, 21, 323–324. Wickramasinghe, H.-K. (1990) Scanning probe microscopy. Current status and future trends. J. Vac. Sci. Technol. A, 8, 363–368.

© 2002 The Royal Microscopical Society, Journal of Microscopy, 205, 21– 32