joining of sic based ceramics and composites with si-16ti ...joining of sic based ceramics and...

440 Int. J. Materials and Product Technology, Vol. 20, Nos. 5/6, 2004

Copyright © 2004 Inderscience Enterprises Ltd.

Joining of SiC based ceramics and composites with Si-16Ti and Si-18Cr eutectic alloys

B. Riccardi* Associazione EURATOM-ENEA, CR Frascati, C.P. 65-00044 Frascati (Rome), Italy E-mail: [email protected] *Corresponding author

C.A. Nannetti ENEA CR Casaccia, 00060 S.Maria di Galeria (Rome), Italy

J. Woltersdorf and E. Pippel Max-Planck-Institut für Mikrostrukturphysik, Weinberg 2, D-06120 Halle/Saale, Germany

T. Petrisor Technical University of Cluj-Napoca, Romania, ENEA Consultant

Abstract: Silicon carbide and SiCf/SiC ceramic matrix composites are attractive materials for energy application because of their chemical stability and mechanical properties at high temperature. For these materials advanced joining techniques are under development and, among them, brazing is one of the most promising.

In this paper a brazing technique based on the use of the Si-16Ti (at.%) and the Si-18Cr (at.%) eutectic alloys (with 1330°C and 1305°C melting points, respectively) is presented and discussed.

The eutectic alloys were prepared by several melting operations, reduced in powders and finally used for the joining experiments. The brazing trials led to joints without discontinuities and defects at the interface and with fine eutectic structures exhibiting a morphology comparable with that of the starting alloys. Microstructure and nanochemistry investigations showed neither interdiffusion nor phase formation at the interface leading to the conclusion that direct chemical bonds are responsible for the adhesion.

Joint specimens showed high shear strength with failure occurring mainly in the base material.

Keywords: brazing; eutectic; silicon carbide; microstructure and nanochemistry; shear strength.

Reference to this paper should be made as follows: Riccardi, B., Nannetti, C.A., Woltersdorf, J., Pippel, E. and Petrisor, T. (2004) ‘Joining of SiC based ceramics and composites with Si-16Ti and Si-18Cr eutectic alloys’, Int. J. Materials and Product Technology, Vol. 20, Nos. 5/6, pp.440–451.

Joining of SiC based ceramics and composites with Si-16Ti and Si-18Cr 441

Biographical notes: Dr. Ing. B. Riccardi graduated in Mechanical Engineering at the University of Naples in 1984 and got the PhD in Materials Engineering at the University of Rome. He’s working at the Nuclear Fusion Department of the Italian Agency for New Technologies, Energy and the Environment (ENEA) in the field of plasma facing components and materials including ceramic matrix composites. More than 70 scientific publications.

Dr. C.A. Nannetti graduated in Industrial Chemistry in 1968 at Pisa University. He’s presently working, as Senior Scientist, at the Material Department of the Italian Agency for New Technologies, Energy and the Environment (ENEA). About 60 scientific publications and patents in the field of technical ceramics as well as ceramic matrix composites.

Prof. Dr. rer. nat. habil. J. Woltersdorf is leading the research group, “Interfaces/New Materials” at the Max-Planck-Institute of Microstructure Physics, Halle, and he is lecturing at the Martin-Luther-University Halle-Wittenberg. PhD Graduation in 1973; Habilitalion in 1988; in 1992, he was appointed professor of surface technology at the Technical University Dresden; Bühler award in 1993; more than 200 scientific publications; editor and coauthor of several technical books and monographs.

Dr. rer. nat E. Pippel is working as a Scientist in the research group “Interfaces/ New Materials’, at the Mu-flancic-Institute of Microstructure Physics, Halle. He graduated in 1982 with a PhD Thesis on pseudomorphous growth in epitaxial bi-crystal systems. Bühler award in 1993; 150 scientific publications; coauthor of several technical books.

Prof. Dr. Tralan Petrisor is the head of the Materials Science Laboratory and lecturing at the Technical University of Cluj-Napoca, Romania, Faculty of Materials Science and Engineering, Department of Physics. PhD Graduation in 1985 with a thesis on “Magnetic properties of titanium and titanium based alloys” In 1999, he was appointed professor at the Technical University of Cluj teaching Thin Films and Applied Superconductivity courses; more than 100 scientific publications.

1 Introduction

Silicon carbide and SiC fibre /SiC matrix ceramic composites (SiCf/SiC composites) are attractive materials for functional and structural application because of their stability and strength at elevated temperatures. Furthermore, the good thermal conductivity and the low coefficient of thermal expansion provide a good resistance to thermal shocks, being an important requirement for application at elevated temperatures such as those for energy conversion systems [1,2]. Otherwise, it is difficult to manufacture components of complex geometry made of SiC and SiCf/SiC composites and thus reliable joining processes have to be developed. The formation of a strong joint can be obtained by the achievement of an intimate contact between the two surface to be joined and the formation of an atomically bonded interface. Silicon carbide has a very limited plastic capability and a very low (SiC) or a limited (SiCf/SiC) thoughness, therefore reliable joints can be obtained if the system is able to accommodate thermal expansion mismatch strains that arise during joining, cooling down or in service temperature gradients.

442 B. Riccardi et al.

In the last 20 years several processes have been proposed to join SiC and SiCf/SiC composites but the most promising ones are diffusion bonding, reaction bonding and metallic brazing. Diffusion bonding of silicon carbide [3] has been carried out by using metals or SiC powder but this process requires high pressure and a temperature that is generally too high to join SiCf/SiC composites. Reaction bonding [4] leads to high strength joints obtained starting from silicon carbon mixtures (sometimes with sintering aids) but it usually requires high temperature and often exhibits porosity.

The brazing technique [5,6] uses alloy foils or powders placed between the material to be joined. The process temperature is higher than the liquidus temperature of the alloy so that generally no or very low pressure is necessary. The assembly is generally held in vacuum or in controlled atmosphere during the entire thermal cycle that can thus be tuned. Among all the joining techniques brazing is very attractive because, provided that the brazing filler avoid the formation of brittle compounds and elevated residual strains, it leads to joint of high strength and requires moderate temperatures, no pressure and simple surface preparation; moreover the scaling up from laboratory to industrial production seems not to be a hard issue. Wettability is the fundamental factor for brazing effectiveness. It is well established that for metallic brazing a contact angle θ < 70° is necessary for many joinings [7].

Several brazing alloys have been proposed and they can be divided in two types [8]: noble metal alloys and reactive metal alloys. The first class includes metals such as Ag, Pt, Cu, Ni etc the second includes Ti, Al, Cr etc. Noble metal alloys provide joints with good ductility and strength and generally have a melting point below 1000°C. Reactive metal alloys have higher melting points and can generate undesirable reactions and eventually brittle compounds.

In this paper, a recently developed silicon carbide brazing process is reviewed and discussed. The alloys used are based on eutectic compositions of silicon-titanium and silicon-chromium.

2 Experimental

Pure silicon is currently used to infiltrate SiCf/SiC composites because of the good compatibility and wettability with silicon carbide (the contact angle θ of Si on SiC at 1480°C is only 38°) [9]. Moreover, taking advantage of the Si low thermal expansion coefficient (similar to SiC), previous works have shown that it is possible to join SiC and SiCf/SiC composites by using silicon without additives or active metals [10]. However, the joints performed by using pure silicon exhibited small strength, specially when applied on SiCf/SiC composites and the Si high melting point may degrade fibres or fibre-matrix interphases in the composites.

Joints with improved properties have been obtained by inserting inert additives and reactive metals in the silicon [11]; in this way elevated strength has been reached, but the results are still not well reproducible and some defects affect the morphology of brazing layers.

The key point for the development of the brazing process presented here is the use of Si-Ti and Si-Cr eutectic alloys with the advantage of the favourable features of a Si based alloy with a lower melting point and the presence of active elements able to produce stronger bonds. Another advantage of these alloys respect pure silicon is the presence of a second phase that allow to control the infiltration due to the porosity of SiCf/SiC


composites. The Si-Ti phase diagram [12] shows that there are two eutectics (Si-16Ti at.% and Si-86.5Ti at.%) both with a melting point of 1330°C. The Si-16Ti eutectic is composed of free silicon and Si2Ti, while the Si-86.5Ti is composed of free Ti and Si3Ti5. The second one was discarded because of the reactivity of free Ti with SiC (∆GoTiC < ∆GoSiC at all temperatures). The Si-Cr phase diagram also shows the presence of two eutectics of interest (Si-18Cr at.% and Si-44Cr at.%): the first has a melting point of 1305°C and the second 1390°C. The Si-18Cr eutectic is composed of free silicon and Si2Cr, while the Si-44Cr is composed of Si2Cr and SiCr. The second eutectic was discarded because of the absence of a free silicon phase and the higher melting point.

Since obtaining the eutectic structure during the brazing cycle was not successful, the alloys had to be prepared in advance by melting. The procedure used consisted in the plasma melting of Si-Ti and Si-Cr mixtures in which the elements percentage is equal to the eutectic one; in this way a fine eutectic structure was obtained and X-ray diffraction analysis confirmed that no phases different from the eutectic ones were present. Then small ingots were appropriately melted several time by electron beam [13]. Due to the difficulty to produce thin foils, the alloys were reduced in powders by crushing and milling the small ingot and used for joining alone or mixed to an organic colloidal compound being able to form a slurry.

Monolithic SiC and SiCf/SiC composites brazed joints have been manufactured; in the first case Carborundum Hexoloy-SA polycrystalline β-SiC has been used [14], while in the second case the SNECMA CERASEP N31 composite was used (properties are reported in [15]). This composite is manufactured by chemical vapour infiltration of a 3-dimensional woven perform of Nicalon CG SiC fibre and finally SiC coated by chemical vapour deposition. Both materials (monolithic and composite) possess a sufficient chemical stability at high temperature and are suitable for high temperature and vacuum brazing. Conversely, it was verified that composites manufactured by polymer infiltration and pyrolysis do not possess enough stability and were not used for brazing.

Because of its low roughness monolithic SiC did not need any surface preparation, while the composites had to be ground in order to reduce the roughness to a few microns; following this operation some fibres were laid bare. Brazed specimens were manufactured by using 12 × 10 × 3 mm3 plates due to the limited material availability, but the brazing procedures and results obtained can be scaled up to larger areas.

The samples were ultrasonically cleaned in acetone before joining and, after brazing alloy deposition and overlapping, they were inserted in a molybdenum sample holder able to provide a modest load (1N) to keep samples in convenient position. The joining thermal cycle was performed in vacuum (10–6 mbar) or inert atmosphere (Ar + 3% H2) furnaces. The brazing cycles comprised a heating up to the eutectic temperature with a rate of 10°C/min, a holding at melting temperature for 10 min in vacuum and 30 min in Ar, a cooling down to 600°C (at 20°C/min) followed by natural cooling to room temperature.

3 Microstructure

Visual inspection of all the joints manufactured showed absence of macroscopic cracks or detachments. Scanning electron microscopy (SEM) and Energy Dispersive X-ray spectroscopy (EDX) examination have been performed on cross-sections of monolithic SiC and SiCf/SiC composite joints to investigate their microstructure. As a first result


no flaws, unmelted particles and micro-cracks were detected in all joints (Figure 1). For Si-Ti and Si-Cr alloys, the joint thickness of composite samples was in the range of 20–30 µm, but a more pronounced local variation of the thickness could be observed depending on the surface roughness. The joint thickness of monolithic specimens was generally higher due to the absence of infiltration in the impervious SiC. The thickness values ranged from few tenth of microns up to 100 µm depending on the amount of braze deposited between the pieces to be jointed. For both alloys the joints showed a fine size eutectic structure composed of a Si matrix and Ti or Cr silicides phases comparable with that of the starting powder. The Si-18Cr microstructure was generally finer but less elongated than that of Si-16Ti. Due to the relatively short hold time at brazing temperature, the infiltration of the alloys in the composite joints looked sufficiently controlled.

Figure 1 SEM micrography of a joint performed between SiCf/SiC composites

The interface structure and the nature of the bonds between the eutectics and the SiC and SiCf/SiC composites have been also investigated by transmission electron microscopy (TEM) including high resolution or atomic plane imaging (HREM), and electron energy-loss spectroscopy (EELS) for chemical analysis [16]. The brazing alloy and the substrate could be observed side by side on the same image due to a specimen cross sectional preparation.

The EELS technique was able to provide an estimation of the type and the distribution of the chemical elements with a spatial resolution limited by the diameter of the measuring probe (1–2 nm). In particular, the analysis of the near-edge fine structures (ELNES) of the relevant ionisation edges, which are caused by excitations of core-shell electrons into unoccupied states above the Fermi level, allows to characterise the chemical bonding state of individual elements with the same resolution. Characteristic ELNES details are the edge onset as well as the shape, the position and the intensity of individual peaks in a range of some 10 eV above the onset. The ELNES features were interpreted by comparing with related standard spectra. A slight overlapping of the detected elements from both sides of the interfaces cannot be excluded although the interface is atomically sharp. This is why the spatial resolution is limited by the probe diameter. The EELS spectrum of the interface can be simulated by simple adding the spectra of both sides of the interface, thus, the formation of new phases can be excluded.


Figures 2 and 3 show TEM micrographs of SiCf/SiC samples joined with Si-16Ti and Si-18Cr alloys indicating the separation of the eutectic alloy into Si and Si2Ti for the first alloy and Si and Si2Cr for the second. The two alloys are composed of a Si matrix with an intermetallic phase but the contact to the substrate is mostly performed by Si even if some TiSi2 and Si2Cr particles are linked to SiC. A consistent hooking together can also be seen between the brazing and the rough SiCf/SiC matrix surface. The TEM image of Figure 3(b) shows that some strain contrast contours have formed along the interface. This mechanical strain results from the thermal expansion mismatch between the SiC bodies and the brazing alloy and hints at strong bonding in the interface.

Figure 2 TEM image of the interface area between SiCf/SiC and Si-16Ti brazing

Figure 3 Microstructure of the brazing area between SiC and 82Si-18Cr wt% – (a) HV-TEM image of the complete brazing layer – low magnification and (b) interface area between SiC and the brazing material)


From the atomic plane images (Figures 4 (a), (b), (c)) it is found that no phases have formed between SiC and Si and between SiC and Si2Ti or Si2Cr, i.e., the contact is atomically sharp.

Figure 4 High resolution TEM images of the interface area between SiC and brazing eutectic phases: (a) interface Si-SiC; (b) interface Si2Ti -SiC and (c) interface Si2Cr -SiC


The chemical analysis of a SiC/Si , SiC/Si2Ti and SiC/Si2Cr interface (Figures 5–7) are shown in a nanometer scale as performed by a set of EEL spectra (a) and across the individual interfaces (b). In (c) the chemical bond specific ELNES of the Si-L23 edge of selected spectra are magnified with the background subtracted: it can be seen that the interface spectrum is almost the sum of those of the adjacent materials recorded far away from the interface. From these findings it must be concluded that in all cases no other phases have formed, i.e., the interface is sharp and no Ti or Cr diffusion into SiC occurs (within a resolution limit of 1–2 nm).

All the analyses performed confirm that the interfaces between the substrate bodies and the brazing alloy are nearly atomically sharp, therefore there is no detectable interdiffusion or formation of new phases. Thus, the high strength of the joints is related to the direct chemical Si-Si and Si-C bonds in case of a Si/SiC interface with additional Si-Ti and Ti-C or Si-Cr and Cr-C bonds depending on the alloy used.

Figure 5 Nanochemistry of the interface between Si and SiC, (a) series of EEL spectra (separation 2.5 nm), (b) TEM image and (c) Si-L23 ELNES of selected spectra


Figure 6 Nanochemistry of the interface between Si2Ti and SiC, (a) series of EEL spectra (separation 2.5 nm), (b) TEM image and (c) Si-L23 ELNES of selected spectra

Figure 7 Nanochemistry of the interface between Si2Cr and SiC, (a) series of EEl spectra (separation 2.5 nm), (b) TEM image, (c) Si-L23 ELNES of selected spectra


4 Mechanical testing

Among all the mechanical features of the joints, the shear strength is one of the leading properties to assess the reliability of any joint technique. Several methods to test the shear strength of joints have been proposed [17]. Generally, each of them is based on a different specimen geometry (single or double lap joint, asimmetric bending bar, Iosipescu specimen etc.), but a pure shear strain field is extremely difficult to be achieved.

In this work, the shear tests were performed following a procedure which consists in a modification of the ASTM D905-89 test procedure (Figure 8) [18]. This method is able to provide a rather good estimation of the shear strength and is an effective way for a comparative evaluation. Due to some limitations of the used apparatus, testing has been performed mainly at room temperature and only some Si-16Ti joints were tested at 600°C. Clearly, this value is far away from the working temperature of SiC brazed components but can evidence some degradation if this occur. The crosshead speed was 0.6 mm/min.

Figure 8 Scheme of the shear test arrangement

The results of shear tests performed in this way were remarkable. The samples joined by using the Si-16Ti alloy on composite substrates exhibited a shear strength (measured at least on 5 specimens for each experimental point) of 71 ± 10 MPa at RT and up to 70 MPa at 600°C; the strength seems to be scarcely influenced by the testing temperature. The samples joined by using the Si-18Cr alloy applied on monolithic SiC substrates cracked at 150 MPa at RT, while the composite samples exhibited a shear strength of 80 ± 10 MPa at RT. Moreover, all the tests carried out on the same type of brazed joints gave sufficiently reproducible results with a limited scattering.

The shear load-displacement curves for composite samples showed, with the exception of an initial adjusting phase, a trend practically linear up to failure which occurred sometimes with a very limited toughness. For these joints, the shear strength was slightly affected by the residual roughness and open porosity.

Observation of the fracture surfaces revealed that failure was always cohesive in all tested specimens. In the composite samples, the failure always occurred in the composites (Figure 9), leading to the conclusion that the limiting parameter of the performance was the shear strength of the composite itself. In monolithic samples, the failure mainly occurred in the substrate; sometimes, even though starting at the joint interface, it also propagated in the bulk SiC.


Figure 9 Composite specimen failure (light micrograph)

5 Conclusions

The proposed joining technique, which employs the eutectic Si-16Ti and Si-18Cr alloys, appears suitable for joining of SiC and SiCf/SiC composites. Joints with a shear strength comparable with that of base material were obtained.

Following SEM analysis all the joints investigated did not show any defects in the brazing layer which maintained a fine eutectic structure. Systematic investigations of the microstructure and of the nanochemistry (HREM, EELS, esp. ELNES) of both Si-16Ti and Si-18Cr joints led to the conclusion that direct chemical bonds are responsible for the adhesion with no interdiffusion or forming of additional phases. Shear tests of the joints of SiCf/SiC composites showed remarkable values of the bonding strength (about 70–80 MPa), while joints of monolithic SiC (Si-18Cr) exhibited excellent strength (up to 140 MPa). or the joining of large pieces.

Acknowledgements

The authors are grateful to Mr. Marcello Sacchetti (ENEA Frascati) who has successfully manufactured all the joints presented and discussed in the paper.

References 1 Naslain, R. (1999) ‘Materials Design and Processing of High Temperature Ceramic Matrix

Composites: State of the Art and Future Trends’, Adv. Comp. Mater., Vol. 8, No. 1, pp.3–16. 2 Mühlratzer, A. and Leuchs, M. (2001) ‘Application of non oxide CMCs’, in Krukel, W.,

Naslain, R. and Schneider, H. (Eds.): Proceedings of the 4th International Conference on High Temperature Ceramic Matric Composites, Münich (Germany) 1–3 October.

3 Akselsen, O.M. (1992) ‘Review: Diffusion bonding of ceramics’, J. Mat. Sci., Vol. 27, pp.569–579.


4 Singh, M. (1997) ‘A reaction forming method for joining silicon carbide-based ceramics’, Scripta Mater., Vol. 37, No. 8, pp.1151–1154.

5 Iseki, T., Arakawa K. and Suzuki, H. (1980) J. Mat. Sci. Lett., Vol. 15, p.1049. 6 Nakamura, M. and Shigematsu, I. (1996) J. Mat. Sci, Vol. 31, pp.6099–6104. 7 Kapoor, R.R. and Eagar, T.W. (1989) Ceram.Eng.Sci.Proc., Vol. 101, p.602. 8 Schwartz, M.M. (1990) ‘Ceramic joining’, ASM International, p.103 9 Li, J.G. and Husner, H. (1991) J. Mat. Sci. Letts, Vol. 10, pp.1275–1276. 10 Ferraris, M., Badini, C., Montorsi, M., Appendino, P. and Scholz, H.W. (1994) J. Nucl. Mat.,

pp.212–215, 1613–1616. 11 Gasse, A., Chaumat, V., Rado, C. and Eustathopoulos, N. (1996) J. Mater. Sc. Letts, Vol. 15,

pp.1630–1632. 12 ASM Handbook (1992) ‘Alloy phase diagrams’, ASM International, Vol. 3, pp.160–367 13 ENEA Patent 478 (2001) RM2001A000101. 14 Hexoloy, S.A. (1998) ‘Technical specification’ Carborundum Italia s.r.l. 15 La Barbera, A., Riccardi, B., Nannetti, C.A., Donato, A. and Moreschi, L.F. (2001) J. Nucl.

Mat., Vol. 294, pp.223–231. 16 Schneider, R., Lichtenberger, O. and Woltersdorf, J. (1996) ‘Phase identification in composite

materials by EELS fine structure analysis’, J. Microsc., Vol. 183, pp.39–51. 17 Cordeau, S. and Taffarel, C. (1997) ‘Characterisation de la resistance au cisaillement de

junctions’, Technical Note CEA-CEREM Grenoble (France) DEM n.38/97. 18 Riccardi, B., Donato, A., Colombo, P. and Scarinci G. (1998) ‘Development of homogeneous

joining techniques for SiC/SiCf composites’, in Beaumont, R., Libeure, P., de Gentile, B. and Tonon, G. (Eds.): Fusion Technology – 20th SOFT Marseille.

joining of sic based ceramics and composites with si-16ti ...joining of sic based ceramics and...

Documents