4.18.6.1 Joining of CFC to Heat Sink

CFCs, bonded either mechanically or otherwise to a metallic structure, are being used in most of the major fusion devices, including ITER,105,106

Tore Supra,111 Wendelstein 7-X,112 TFTR in United States,113 JT 60U, and JT60SU in Japan.110-113 Silicon-doped carbon is used as the first wall material for the Chinese reactor HT-7.114 See Table 3 for a description of ITER candidate composites (INOX Sepcarb NS-31, Sepcarb NB-31, and Dunlop Concept C1, for example).

As mentioned, CFCs will be used in ITER (see Table 5 for performance specifications), and specifically a 3D CFC for the divertor in the initial phases of the ITER project. The diverter, which is among the most technically challenging ITER com­ponents, is located at the bottom of the plasma cham­ber where the CFCs (and tungsten) are bonded to a copper alloy (CuCrZr). CFCs have been selected

for the lower part of the vertical targets for the initial phase of ITER operation (without tritium), and they must be able to resist ‘steady state’ heat flux up to 10MWm~2 for at least three thousand 400 s pulses and up to 20 MW m~2 in transient events.

As discussed earlier, the use of graphite-based materials (particularly CFCs) in a divertor is believed to be an advantage for the first phase of ITER operations because CFCs have good demonstrated performance in the currently operational plants (e. g., Tore Supra). Their primary competitor, tung­sten, also suffers from macroscopic cracking, melting, and possible melt layer loss, thus making the poten­tial damage to the divertor components more serious than that of CFC. As conditions of additional heating and off-normal events will be very likely during initial operation of ITER, CFC is the present refer­ence design solution for the lower part of the vertical targets for the ITER initial phase, when tritium retention-related issues are not relevant.

In the ITER design, CFC must be joined to the copper alloy CuCrZr-IG (ITER Grade)116-119 in order to transfer the heat loads. Two different designs being considered for this component are the CFC-bonded flat-tile (Figure 33(a)) and monoblock (Figure 33(b)). In order to join the mating surface of the CFC to the copper alloy heat sink, a pure copper interlayer (1-2 mm thick, oxygen free high conduc­tivity copper, 99.95%, CTE: 15.4 at RT, up 20.6 at 700 °C and up to 21.6 at 800 °C) is used to relieve, by plastic deformation, the thermal expansion derived between the CFC and copper alloy heat sink. The CTE of CFC can be found in Table 3, with that of the copper alloy stresses being (16-19) x 10~6K_1 at 700 °C).116 This joining design is shown schematically in Figure 34. Some alternatives to a pure copper interlayer have been proposed within the EU project ExtreMat. For example, an Mo interlayer (1-2 mm),

(b)

Figure 33 (a) Flat tile design. Reproduced from Ferraris, M., et al. In Ceramic Integration and Joining Technologies: From Macro to Nanoscale, 1st ed.; Singh, M., Ohji, T., Asthana, R., Mathur, S., Eds.; Wiley, 2011; Chapter 3,

© 2010 The American Ceramic Society. (b) Monoblock design. Courtesy of J. Linke, FZJ, Germany.

a Cu/W fiber interlayer, and CFC monoblocks with a Cu/W fiber interlayer (Figure 35) have been prepared and tested up to 10MWm~2. Results (unpublished) indicate that this method is less promising compared to the use of a pure Cu thin layer. Ti-doped CFC have also been prepared and tested for flat tiles.117,118

Among the several possible options, the flat tile and monoblock configurations (see Figure 33(a) and 33(b)) with a pure copper interlayer have yielded the most promising results. In particular, the monoblock gives a more robust solution in com­parison with the flat tile for the vertical target and it is now considered as ITER reference geometry.119,120

The monoblock design requires drilled blocks of CFC into which a CuCrZr tube is inserted and joined; also necessary in the monoblock is a pure copper interlayer between the CFC and the copper alloy to relax interface stresses, which have been (modeled and) measured at ±45°, ±90°, ±135°. If 0° is considered to be the flux direction,121,122 the monoblock is preferred to the flat tile design. This design is also much easier to manufacture because of its better heat flux performances and because of its intrinsic ability to attach even in the presence of cracks at the interface of CFC and CuCrZr, caused by its preparation process.

Methods to join CFC to CuCrZr derive from techniques developed to join carbon-based compo­nents to metals. Brazing is a well-known joining process recommended for joining dissimilar

Figure 34 Carbon fiber composite joined to CuCrZr in a (a) flat tile and (b) monoblock configuration (drawings show test components and not full-scale components). Reproduced from Ferraris, M., et al. In Ceramic Integration and Joining Technologies: From Macro to Nanoscale, 1st ed.; Singh, M., Ohji, T., Asthana, R., Mathur, S., Eds.; Wiley, 2011; Chapter 3, © 2010 The American Ceramic Society.

(a) (b)

11500 mm

Figure 35 Cross-section of a carbon fiber composite monoblock with a Cu/W fiber interlayer. Courtesy of EU-Extremat Project and Pintsuk, G. JFZ, Germany.

materials. If properly done, it results in good mechan­ical strength, high fatigue performance, and minimal thermal resistance at the interface.123-125 In the case of joining of CFC for nuclear applications, some additional restrictions must be considered. A primary consideration is that the joining materials should be low activation materials (LAMs), even if the volume occupied by these materials is negligible in respect of the total volume of the structural materials in the reactor.126 Also important is that the materials be irradiation-stable at the anticipated conditions. Furthermore, the use of pressureless joining
techniques is preferred as the parts to be joined are relatively large. Some joining materials are not allowed, for example, elements with high vapor pres­sure (e. g., zinc or cadmium), or those giving dangerous transmutation reactions. The ITER project mandates thermodynamic and mechanical stability up to at least 800 °C under vacuum for the joint, in order to satisfy requirements of Table 5; the joint must survive the thermal, mechanical, and neutron loads faced by the component, and it is expected to operate in a cyclic mode with an acceptable reliability and lifetime. The joining technology must also be compatible with the overall component manufacture process and in particular with the preservation of the thermomechanical properties of the precipitation — hardened CuCrZr alloy.127

Wettability is a key factor in the joining of CFC to the heat sink. It is not within the scope of this chapter to review this subject, and an overview can be found elsewhere.12 ,12 ,129 However, to restate the most salient point, the Cu interlayer cannot be obtained by directly casting copper on the CFC surface, because Cu does not wet CFC at all, the contact angle of molten copper on carbon substrate being about 1400.128,129 The poor wettability of CFC is related to the non­metallic character of its bonding, whereas the bonding electrons in copper are delocalized.123,124,128

Copper can be directly cast on CFC when the CFC surface is modified to form carbides by a solid state chemical reaction between the composite surface and elements such as Si, Al, Ti, Zr, Cr, Mo,

or W; some metals can form carbides with ‘metal-like’ behavior128 and are usually well wetted by molten metals. Several patents refer to joints between carbon-based materials and copper130-135: for exam­ple, pure Cr and Ti react with C to form carbides. Cr and Ti wet carbon-modified surfaces very well with a contact angle of 35-40° at 1775 °C and of 50-60° at 1740 °C, in Ar, respectively.128

A joining technique based on CFC surface modi­fication is the Active Metal Casting (AMC®) tech­nique originally developed in the eighties by the Austrian company Plansee for nonnuclear purposes. ‘Active’ indicates the activation of the CFC surface to allow it to be wetted by Cu. Physical vapor deposition (PVD) or chemical vapor deposition (CVD) of Ti coating on the CFC surface is followed by a high temperature treatment to form TiC, which improves the wettability ofCFC by molten copper. Active Metal Casting consists of casting a pure Cu layer onto a laser-textured and TiC-modified CFC surface.136,137 The laser texturing enhances Cu infiltration into the CFC, and the TiC-modified CFC surface improves the wetting. The special laser treatment of the CFC surface produces a large number of closed conical holes (diameter ^50-500 pm, depth 100-750 pm), thus increasing the joined area and providing addi­tional crack growth resistance. Due to the open poros­ity of the TiC-modified CFC and laser machining, the cast Cu penetrates into the CFC up to 2 mm. An example of a full-scale component produced by Plansee, Austria, is shown in Figure 36.

AMC® was successfully applied for both flat tile and monoblock geometry. However, AMC® technol­ogy requires laser machining of CFC surfaces, which might not be economically attractive for large-scale production. Laser-induced stresses in the joined area and cracks induced during the joining process have recently been measured and modeled.1 However,

Plansee has recently improved AMC® by using silicon and titanium to modify the CFC surface (TiSi-AMC) (Figure 37(c)).130 Ansaldo Ricerche — Genova, Italy, has proposed a joining technique based on a Cu-Ti-based (Cu ABA) commercial alloy, reinforced by 2D randomly oriented carbon fibers uniformly distributed in the brazing alloy. The joining is carried out at approximately 1000 °C. The Ti reacts with carbon to form a thin TiC layer that promotes wetting.131 Carbon fibers are expected to mitigate thermal expansion mismatch between CFC and the braze (Figure 38) and to react with titanium in the brazing alloy, resulting in beneficial thermal fatigue strength of the joint. This technique was

Plansee

Tungsten (upper part)

Blocks 33

Austenitic steel

Blocks 32

CFC — (lower part)

Blocks 1

Copper/steel
tube joint

Figure 36 Vertical target full-scale prototype manufactured by Plansee (high heat flux units) and Ansaldo Ricerche (support structure and integration). Reproduced from Missirlian, M., et al. J. Nucl. Mater. 2007, 367-370(2), 1330-1336.

successfully tested on CFC NB31-Cu flat-tile and monoblock joints. Several other solutions have been investigated to modify CFC surface, for example, by TiN or TiC, within the EU project ExtreMat.118

A method has been proposed based on the CFC surface modification by reaction of Cr, Mo, and W. Both Cr and Mo have been extensively used as active elements in brazing alloys for copper active brazing and in patents referring to nonnu­clear applications.130-135,138,139 As example W, Mo, and Cr powders were deposited on CFC (CFC NB31, Snecma Propulsion Solide, France) by the slurry technique: details related to the process can be found elsewhere.132-134 Cr-carbide-modified CFC appears to have yielded the best results (Figure 39). A 15-pm-thick carbide (Cr23C6, Cr7C3) layer has been identified by XRD on CFC; the CTE of the carbides lies between that of CFC and copper (reported above) (CTE of Cr7C3 is 10 x 10-6K-1).135

A commercial brazing alloy Gemco® (87.75 wt% Cu, 12 wt% Ge and 0.25 wt% Ni; Wesgo Metals) has been used to braze Cr-modified CFC to Cu and

(c)

Figure 37 (a and b) Laser structuring of carbon fiber composite (CFC) in Active Metal Casting (AMC®) process and cross-section of the AMC® CFC-Cu joint. Courtesy of Chevet, G. Ph. D. Thesis 2010, University of Bordeaux, France.

to CuCrZr in a single step process,140 which is an advantage in comparison with other joining technol­ogies that require two steps: first joining CFC to Cu, and then CFC-Cu to CuCrZr. Flat-tile (a) and monoblock (b) mock-ups have been obtained by this technique and tested (Figure 40).

ENEA, Italy, has manufactured several actively cooled mock-ups of flat tile and monoblock type, by using different technologies; a new process (pat­ented) for the production of monoblocks is based on prebrazed casting and hot radial pressing (PBC+HRP) (Figure 41). The CFC surface modification is obtained by a titanium-copper-nickel commercial brazing alloy, which is followed by a Cu casting, then a radial diffusion bonding between the cooling tube and the CFC by pressurizing only the internal tube and keeping the joining zone in vacuum at the

120,141,142

required bonding temperature.

Complete manufacture and testing of this vertical target medium-scale mock-up (Figure 41) can be

(a)

31 mm

38 mm

(b)

Figure 39 Carbon fiber composite-Cu active brazing with Cu-Ti alloy and dispersed carbon fibers in the braze (b). Reproduced from Schedler, B.; Huber, T.; Eidenberger, E.; Scheu, C.; Pippan, R.; Clemens, H. Fusion Eng. Des.

2007, 82(15-24), 1786-1792. Reproduced from Ferraris, M., et al. In Ceramic Integration and Joining Technologies: From Macro to Nanoscale, 1st ed.; Singh, M., Ohji, T., Asthana, R., Mathur, S., Eds.; Wiley, 2011; Chapter 3,

© 2010 The American Ceramic Society.

considered as a success for both PBC and HRP pro­cesses, which can be an alternative to current techniques.

Several joining techniques are based on active braz­ing, which do not require any manner of CFC surface

Figure 40 Optical micrograph of the cross-section of a Cr-carbide modified carbon fiber composite (CFC)-Cu joint, (b) Cr-carbide modified CFC-Cu-CuCrZr mock-up. Reproduced from Ferraris, M., et al. In Ceramic Integration and Joining Technologies: From Macro to Nanoscale, 1st ed.; Singh, M., Ohji, T., Asthana, R., Mathur, S., Eds.; Wiley, 2011; Chapter 3, © 2010 The American Ceramic Society.

modification. In this case, the active brazing alloys for CFC include elements such as Ti, Zr, Cr, and Si, which allow CFC wettability by the molten brazing alloy. A drawback of active brazing can be that active elements may form brittle intermetallics or com­pounds of low melting point. In one study, a TiCuNi brazing alloy produced by Wesgo Metals in the form of sheets (70Ti-15Cu-15Ni) was used to join CFC and silicon-doped CFC to pure copper, but the presence of Ni and Ti brittle intermetallics at the joint interface had a detrimental effect on thermal fatigue resistance tests of the joined component.

In Japan, a brazing technology was developed for ITER143 to join CFC to Cu by a NiCrP brazing alloy, followed by the joining of the CFC-Cu to the CuCrZr by low temperature HIP (500 °C). Recently,

a new brazing process was developed, based on the NiCuMn alloy, after metallization of the CFC surface. Several other brazing alloys have been devel­oped for CFC-Cu joints: Ag-based (63Ag-35Cu-2Ti, 59Ag-27Cu-13In-1Ti), or Cu-based brazing alloys (Cu-3Si-2Al-2Ti; Cu-Mn; Cu—Ti). Ag was discarded in view of nuclear transmutation-related issues.144