- Email: [email protected]
Wetting behavior and interfacial interactions of molten Cu50Ti alloy with hexagonal BN and TiB2 ceramics
Ceramics International 42 (2016) 9906–9912
Contents lists available at ScienceDirect
Ceramics International journal homepage: www.elsevier.com/locate/ceramint
Wetting behavior and interfacial interactions of molten Cu50Ti alloy with hexagonal BN and TiB2 ceramics Yangwu Mao a,n, Liangxing Peng a, Quanrong Deng a, Dunwei Nie a, Shenggao Wang a, Lixia Xi b a b
Hubei Key Laboratory of Plasma Chemistry and Advanced Materials, Wuhan Institute of Technology, Wuhan 430073, China Institute for Complex Materials, IFW Dresden, Dresden 01171, Germany
art ic l e i nf o
a b s t r a c t
Article history: Received 21 December 2015 Received in revised form 9 March 2016 Accepted 11 March 2016 Available online 12 March 2016
Wetting behavior of molten Cu50Ti alloy on hexagonal BN (h-BN) and TiB2 ceramics has been studied under vacuum using a modified sessile drop method. Final contact angles of 8° and 3° are obtained at 1000 °C on h-BN and TiB2, respectively. Interaction occurs at the interface between the molten alloy and BN, leading to the formation of a reaction layer containing TiB and Ti nitrides. Interfacial interaction of Cu50Ti with TiB2 results in the formation of densely packed TiB layer about 60–100 μm thick and the detachment of TiB2 grains. Spreading wetting of liquid Cu50Ti on h-BN is mainly controlled by the reactions between Ti and BN at the triple line. For Cu50Ti/TiB2 system, spreading is mainly limited by the interfacial reaction in the first stage, and is possibly influenced by both the diffusion of boron atoms and viscous friction of the liquid in the second stage. Finally, brazing of graphite to CuCrZr alloy has been realized using Cu50TiH2 with ceramic additives (including BN and TiB2) as composite fillers. The joints exhibit favorable interfacial bonding between the filler layer and the substrates. The ceramic reinforcements in the filler layer could contribute to the improvement of the shear strength. & 2016 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
Keywords: A. Joining B. Interfaces D. Borides D. Nitrides
1. Introduction Brazing technology, as one of the most effective joining techniques, has been widely used for the fabrication of ceramic/metal joints. For example, carbon/metal (including graphite/Cu and carbon-carbon composite/Cu-clad Mo) joints have been prepared by brazing for the industry applications [1–3]. However, one of the major issues for brazing ceramics to metals is the existence of large residual stresses, which are mainly generated by the mismatch of coefficient of thermal expansion (CTE) and elastic modulus between the substrates [4,5]. The residual stresses of joints may lead to cracks in the ceramic and even the failure of joints upon cooling from the processing temperature. The metal interlayers and composite fillers have been introduced to alleviate the residual stresses of the ceramic/metal joints [6–14]. The composite fillers are generally composed of active metal brazes and ceramic reinforcements. The residual stresses of joints may be relaxed by the introduction of ceramic reinforcements with relatively low CTE. In addition, the ceramic additives in the composite fillers are assumed to be distributed uniformly in the filler layer, which can develop the reinforced n
Corresponding author. E-mail addresses: [email protected], [email protected] (Y. Mao).
http://dx.doi.org/10.1016/j.ceramint.2016.03.090 0272-8842/& 2016 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
structure in the joints. Halbig et al. [11] studied the brazing of SiC using Ag–Cu–Ti braze alloys reinforced with SiC particulates. Based on the theoretical calculations, the CTE of the braze with the incorporation of about 45 vol% SiC could decrease by nearly 45– 60%. As reported in Ref. [13], the synthesized TiB whiskers contributed to the strength improvement of Al2O3/Ti–6Al–4V joints. It should be noted that the microstructure and properties of the joints may be influenced by the wettability and interactions between liquid metal brazes and ceramic reinforcements. Thus, it is crucial to study the wetting behavior between the metal brazes and ceramic reinforcements prior to the design and preparation of the composite fillers for joining. The reinforcements used in the composite fillers generally include carbides, nitrides, borides, and so on. Among them, BN and TiB2 ceramics have been used as reinforcements in Refs. [13,14]. Extensive studies have been performed on the wetting of BN by liquid metals [15–22]. Naidich et al. [15] measured the contact angles in the range of 135–150° at 1000–1500 °C for the metals with a negligible (Ag, Sn, Au) or weak (Ge, Ga, Cu) affinity for both N and B, and contact angles lower than 90 ° for ferrous metals. For Al/BN system, nearly perfect wetting is achieved at 1000 °C and a continuous AlN layer is formed at the interface by the reaction between liquid Al and BN [16]. Nicholas et al. [18] studied the wetting of BN by addition of Ti to Cu and Ag–Cu. Very low contact angles were obtained at both 1150 °C and 950 °C. Furthermore, the
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
wetting of BN by liquid alloys including Si, Ni–Mo (0–40 wt%), GaSb, InSb, and Al–Si has been reported in Refs. [19–22]. Investigations on the wetting of TiB2 by liquid metals, including Cu, Au, Ni, Al, Fe and Ti, have been extensively carried out [23–31]. For Cu/TiB2 system, both wetting and non-wetting contact angles have been reported by different researchers [23]. The molten Cu and Au show a good wetting on non-stoichiometric TiB2 substrates (TiB1.9 and TiB1.95). Some limited boride dissolution and alteration of the substrate composition occur at the TiBx/Cu and TiBx/Au interfaces [24–26]. In the case of Ni/TiB2 system [23], TiB2 is dissolved into the liquid Ni, resulting in a good wetting in either high vacuum or neutral gas environments. Weirauch Jr et al. [27] investigated the wettability of molten aluminum drops on four different types of TiB2 substrates. The effect of the substrate microstructure on the wetting kinetics has been discussed. Mutale et al. [28] introduced the spreading wetting of TiB2 substrates by molten aluminum in the temperature range between 660 °C and 760 °C in different fluxes. Ghetta et al. [29] performed the study on the wetting of sintered TiB2 by pure iron and iron containing dissolved TiB2. The wetting deteriorated with the increase of oxygen content in the TiB2 substrates. Xi et al. [30–31] presented a good wetting of molten Al, Ti and Ti–Al alloys on TiB2. The formation of TiB occurs at the interfaces of Ti/TiB2 and Ti74.3Al25.7/TiB2. In this study, the wetting behavior and interfacial interactions of Cu50Ti alloy with h-BN and TiB2 ceramics have been investigated. Such a study can provide a reference for joining graphite to Cu alloys with the composite fillers composed of Cu50Ti metal braze and ceramic reinforcements (including BN and TiB2).
9907
Fig. 1. Schematic of the modified sessile drop method.
spreading processes of the Cu50Ti alloy on the ceramic were monitored by a high-speed CMOS camera (frame rate of 120 fps and resolution of 640 480 pixels). The time-dependent variation in contact angle during isothermal wetting at 1000 °C was tracked. Both the contact angle (the error of 0.5°) and drop diameter were measured from the droplet images by using an axisymmetricdrop-shape analysis (ADSA) program. 2.2. Joining of graphite/CuCrZr using composite fillers
2. Materials and methods 2.1. Wetting of Cu50Ti alloy on h-BN and TiB2 Hexagonal BN (h-BN) ceramics (Φ 20 mm 5 mm) with a purity over 99 wt%, and TiB2 ceramics (20 mm 20 mm 3 mm) with 4–5 wt% sintering aid containing Ni, were used as the wetting substrates. Both the h-BN and TiB2 ceramics were supplied from Key Laboratory of Automobile Materials, Jilin University, China. The surfaces of the substrates were mechanically ground and subsequently polished using diamond suspensions. The Cu50Ti alloy with 50 wt% Ti was prepared from high-purity Cu (99.999 wt%) and Ti (99.995 wt%) plates by arc-melting in a purified Ti-gettered argon atmosphere. The molten alloy was turned in a water-cooled copper crucible and remelted for four times by electromagnetism stirring to ensure a good homogeneity. Then, the alloy was cut into small cubic pieces weighing about 150 mg. The ceramic substrates and the Cu50Ti alloy pieces were ultrasonically cleaned in ethanol prior to wetting tests. The dispensed drop method is unsuitable in our study because of a severe reaction of the Ti-containing melts with the drop dispenser made of alumina. Accordingly, a modified sessile drop method described elsewhere [32] was adopted. The solid bulk alloy, rather than the liquid alloy droplet, was dropped on the substrate surface from the alumina tube. Fig. 1 gives the schematic of the modified sessile drop method. The distance between the end of the tube and the substrate surface was about 6 mm. The ceramic substrate was placed in the vacuum chamber and kept to a horizontal position, while the Cu50Ti alloy specimen was stored in a stainless-steel tube outside the chamber. The chamber was evacuated to a vacuum about 2 10 4 Pa at room temperature, and then heated to the testing temperature of 1000 °C at a rate of 20 °C/min. Then, the solid Cu50Ti alloy was delivered to the surface of the ceramic through an open alumina tube which was connected with the stainless-steel tube. Indeed, the alloy was not melted when it contacted the substrate surface. The melting and
The commercial graphite (10 mm 10 mm 10 mm) with a density of 1.9 g/cm3 and purity of 99.99% was purchased from Changsha Aobo Carbon Co., Ltd., China. The commercial CuCrZr alloy (10 mm 10 mm 10 mm) with a nominal composition of Cu- (0.4–1) Cr- (0.03–0.15) Zr (in wt%) and a density of 8.9 g/cm3 was supported from Shenzhen Heshuo Metal Products Co., Ltd, China. Both the graphite and CuCrZr alloy were used as joining substrates. The substrates were polished with 1.0 μm diamond paste and then ultrasonically cleaned in alcohol prior to brazing experiments. The composite filler was composed of Cu powders, TiH2 powders (replacement of Ti powders, to avoid the oxidation during the mechanical milling) and ceramic additives. The detailed description of Cu powders and TiH2 powders could be found in Ref. [33]. BN powders with the size of about 1 μm and purity of 99.0%, and TiB2 powders with the size of about 3–5 μm and purity of 99.5%, were supported from Qinhuangdao Eno High-Tech Material Development Co., Ltd, China. The powders of Cu, TiH2 and ceramic additives were mixed together by mechanical milling to obtain Cu50TiH2 þ BN (or Cu50TiH2 þTiB2) composite filler. The weight ratio of Cu and TiH2 powders was kept at 1:1, and the content of ceramic additives was 2 wt% in the composite filler. Considering that the melting temperature of the powder filler could be lower than that of bulk alloy, the joining temperature was set at 950 °C (with the holding time of 10 min). The detailed joining procedure and the shear strength measurement of joints were described in Ref. [33]. 2.3. Microstructure characterization The wetting specimens for cross-section view were prepared in an epoxy mount and then polished. The microstructure of the cross-section was characterized by a scanning electron microscope (SEM, Quanta200, Holland) equipped with an energy dispersive X-ray spectrometer (EDS, SDD Inca X-Max50, Holland). The phase identification of interfacial area was determined by an X-ray
9908
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
Fig. 2. Variations in contact angle and normalized drop diameter with time for the wetting of Cu50Ti/h-BN system at 1000 °C.
diffraction spectrometer (XRD, Bruker D8 Advance, Germany) with a scanning speed of 4°/min. The microstructure characterization of the interfacial area for the graphite/CuCrZr joints was also examined by means of SEM.
3. Results and discussions 3.1. Wetting of Cu50Ti/h-BN system The melting temperature of 950 °C can be obtained for Cu50Ti eutectic alloy, according to the Cu–Ti phase diagram [34]. Taking
into account that the brazing temperature is about tens of degree higher than the melting temperature of the filler, the isothermal wetting for liquid Cu50Ti alloy on the ceramics has been investigated at 1000 °C. Fig. 2 shows the variations of contact angle and normalized drop diameter with time for the wetting of Cu50Ti/h-BN system. The measured initial contact angle is 42° at 1000 °C. The contact angle decreases dramatically in the range of 0–10 s, and then decreases slowly in the range of 10–30 s. It ultimately reaches a steady contact angle of 8°. The spreading of the molten Cu50Ti alloy on the h-BN, as reflected by the normalized diameter, is rapid at the beginning, and then becomes slow. The normalized diameter remains unchanged when the wetting reaches equilibrium. Fig. 3 displays the SEM micrograph and EDS element distribution of the cross-section of Cu50Ti/h-BN couple after wetting at 1000 °C. It can be seen from Fig. 3(a) that the reaction layer between the alloy and BN consists of two morphological sub-layers: a continuous layer about 5 μm thick close to BN and a layer containing amounts of needles extending to the alloy drop. The EDS element distribution shown in Fig. 3(b)–(d) indicates that the interfacial area is mainly composed of element Cu and Ti as well as N. The enrichments of element Ti can be observed at the interface, suggesting the formation of reaction layer containing Ti–B and Ti– N compounds. The XRD patterns of the interfacial area, as shown in Fig. 4, confirm the presence of TiB, TiN and Ti2N in the interfacial area. In addition, the cracks are observed in the h-BN substrate. They are mainly caused by the large residual thermal stresses, which are generated due to the big CTE mismatch between the alloy and h-BN during the cooling stage after wetting tests. The reaction evolution between Ti and BN has been investigated by different researchers [18,35–38]. The study
Fig. 3. SEM micrograph (a) and EDS areal distribution (b)–(d) of the cross-section of the Cu50Ti/h-BN couple after wetting at 1000 °C.
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
9909
Fig. 4. The XRD patterns of the interfacial area for the Cu50Ti/BN couple.
performed by Nicholas et al. [18] revealed that Ti nitrides (from TiN to TiN0.43) were formed through the reaction of BN with Ti (BN þTi¼ TiN þB), and then the liberated boron reacted with Ti to form Ti borides (including TiB, Ti3B4 and TiB2) adjacent to the metal drop. However, Faran et al. [35] observed a thin TiB2 layer formed close to BN followed by the much thicker TiB (0 2 0) layer through the reaction between BN and Ti (BN þTi-TimBn þ N). The nitrogen released from the reaction could diffuse through the boride layer and then react with excess Ti to form Ti nitrides. In addition, according to the microstructure analysis obtained by Ding et al. [36], the composite TiB2–TiN reaction layer was developed by the interfacial reaction between the cubic BN and Ti from the braze filler. In our study, a TiB reaction layer was identified close to BN side, but no TiB2 phase was detected in the interfacial area. As argued in Ref. [39], TiB2 is just a transition product of the interfacial reaction because TiB2 is not in equilibrium with Ti, even though the ΔG value of TiB2 formation at 1000 °C is more negative than that of TiB formation. It is, therefore, believable that TiB instead of TiB2 is developed in the reaction layer close to BN area in the Cu50Ti/hBN system. The nitrogen atoms released from the reaction of Ti with BN diffuse through the TiB layer, and then react with Ti to form Ti nitrides, i.e. TiN and Ti2N. According to the Ti–B–N ternary diagram [40] (in Fig. 5), TiB can be in equilibrium with Ti nitrides only when the concentration of nitrogen is less than 40 at% [35]. The diffusion path of the reaction products shown in dashed blue tie line in Fig. 5 is similar to the results in Ref. [35], suggesting that both TiB and Ti nitrides exist in the reaction layer of Cu50Ti/h-BN couple. As illustrated by Eustathopoulos et al. [41], the spreading time of reactive metal/ceramic systems (in the range 10–104 s) could be several orders of magnitude higher than that of non-reactive systems (E 10 2 s). Thus, the spreading rate of the reactive stage for a given reactive system may be controlled by the interfacial reaction itself. In view of the interfacial reactions in Cu50Ti/BN system, the reaction product control model is applied to study the spreading kinetics. The characteristic equation of chemical reaction-limited reactive wetting was proposed by Dezellus et al. [42] as follows:
cos θe − cos θd = (cos θe − cos θ 0 ) exp ( − kt )
Fig. 5. The isothermal section of B–N–Ti ternary diagram at 1000 °C [40] with the corresponding diffusion path.
used as θe and θ0 in the Eq. (1) for the Cu50Ti/BN system at 1000 °C. After taking napierian logarithm, the transformative equation yields:
ln (cos θe − cos θd ) = ln (cos θe − cos θ 0 ) − kt
(2)
The result of ln(cos θe cos θd) versus time for the wetting of Cu50Ti on h-BN is shown in Fig. 6. The experimental data is in good agreement with the linearity prediction for the whole spreading stage, suggesting that the spreading of liquid Cu50Ti alloy on h-BN substrate is mainly controlled by the interfacial reactions. The value of the fitting dynamic constant k is about 0.2/s based on the calculation. 3.2. Wetting of Cu50Ti/TiB2 system The wetting behavior of Cu50Ti/TiB2 system at 1000 °C, as depicted in Fig. 7, is similar to that of Cu50Ti/h-BN system. The contact angle rapidly decreases from the initial 57° to 17° (∼5 s),
(1)
Where θe, θd, and θ0 are the equilibrium, dynamic, and contact angle of the liquid on the unreacted substrate, respectively, and k is a dynamic constant, representing spreading rate. The measured equilibrium and initial contact angles, 8° and 42°, respectively, are
Fig. 6. Napierian logarithm of cos θe cos θd versus time t for the Cu50Ti/h-BN wetting system at 1000 °C.
9910
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
Fig. 7. Variations in contact angle and normalized drop diameter with time for the wetting of Cu50Ti/TiB2 at 1000 °C.
and then gradually towards a value of 3° (∼25 s), which corresponds to the excellent wetting. Further holding at 1000 °C produces negligible changes of the contact angle. The normalized diameter of the Cu50Ti alloy droplet increases rapidly in the beginning and then slowly in the later stage. The diameter of the droplet eventually remains constant. Fig. 8 gives the SEM micrograph, EDS element distribution and XRD patterns of the interfacial area for the Cu50Ti/TiB2 couple after wetting at 1000 °C. In Fig. 8(a), a stack of densely packed whiskers extending up to about 60–100 μm are formed at the interface. The whiskers distributed in the interfacial area are identified as TiB phase according to the EDS analysis and XRD patterns in Fig. 8(d). The ΔG value of the reaction (TiB2 þTi ¼2TiB)
at 1000 °C, calculated using the software HSC Chemistry Version 6 (Outokumpu Ra, Oy, Finland), is negative ( 52.5 kJ/mol), confirming the possibility of TiB whiskers formation in the interfacial area. The formation of TiB by the reaction of Ti with TiB2 is in agreement with the results revealed in Refs. [43,44]. In Ref. [43], TiB whiskers were formed in the interfacial area due to the reaction of TiB2 particles with Ti during brazing Al2O3 to Ti6Al4V alloy with Cu–Ti þTiB2 composite filler. Sobhani et al. [44] demonstrated the formation of TiB whiskers connecting with TiB2 particles for the preparation of TiB2 particles reinforced Cu–Ti composites. In addition, the detachment of the TiB2 grains can be observed at the interface in Cu50Ti/TiB2 couple, which is mainly caused by the penetration of molten Cu50Ti alloy along the grain boundaries of TiB2. Similar microstructure and morphology of the interfacial area have been reported in Ti-25Al/TiB2 couple by Xi et al. [31], who also presented the formation of the stack of densely packed TiB whiskers and the detachment of the TiB2 grains at the interface. The EDS results of microzones E and F marked in Fig. 8(a) give the compositions of Cu36Ti59Ni5 and Cu27Ti72Ni1, respectively, indicating the formation of Ti–Cu intermetallics in the interfacial area besides TiB whiskers. The existence of element Ni in the metallic alloy is attributed to the diffusion of the sintering aid Ni in the TiB2 substrate. The EDS element distribution reveals the distribution of the Cu in the metallic alloy and enrichments of Ti in the interfacial area, which confirms the formation of TiB whiskers and the detachment of TiB2 grains at the interface. In addition, the cracks in the TiB2 substrate are mainly resulted from the residual thermal stresses originated from the big CTE mismatch of the wetting couple. In view of the interfacial reaction (formation of TiB) in Cu50Ti/TiB2 system, the reaction-limited spreading model is
Fig. 8. SEM micrograph (a), EDS element distribution of Ti and Cu (b)–(c) and XRD patterns (d) of the interfacial area of Cu50Ti/TiB2 couple after wetting at 1000 oC.
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
9911
influence the spreading. It is, therefore, possible that the spreading of liquid Cu50Ti alloy on TiB2 is influenced by both the diffusion of boron and viscous friction of the liquid in the second stage (5– 25 s). 3.3. Joining of graphite/CuCrZr with composite fillers
Fig. 9. Napierian logarithm of cos θe cos θd versus time t for the Cu50Ti/TiB2 wetting system at 1000 °C.
applied to interpret the spreading kinetics. Fig. 9 displays the napierian logarithm of cos θe cos θd versus time for Cu50Ti/TiB2 system. It appears that ln(cos θe cos θd) versus time is linear in the first stage (0–5 s) with a slope about 0.5/s. At this stage, the boron atoms can diffuse easily from TiB2 to the liquid alloy drop at the triple line and satisfy the demand of the interfacial reaction. It is indisputable that the spreading wetting of the Cu50Ti/TiB2 system in the first stage is mainly controlled by the interfacial reaction at the triple line. With the progress of the interaction between liquid alloy and TiB2, the spreading rate of liquid on TiB2 is low (less than 0.1 mm/ s), corresponding to a contact angle lower than 20°. The spreading of the liquid at the second stage may be influenced by two aspects. On the one hand, the diffusion of boron atoms outwards from TiB2 to the triple line becomes difficult due to the formation of densely packed TiB whiskers at the interface. As a consequence, the diffusion of boron may influence the spreading of the liquid alloy on TiB2. On the other hand, as argued by Eustathopoulos et al. [23], the inertial force is more predominant than viscous force during the initial spreading in non-reactive systems, since the viscosity of liquid metals and alloys is very low. However, the viscous friction could not be negligible in the late stages of spreading for nearly perfect wetting systems [41]. Taking into account the excellent wetting (very low contact angles) for Cu50Ti/TiB2 system in the second stage (5–25 s), the viscous friction of the liquid may
The composite fillers composed of Cu50TiH2 braze and ceramic additives including BN and TiB2 have been applied for joining graphite to CuCrZr alloy. The shear strength of the joints with Cu50TiH2 þ BN and Cu50TiH2 þTiB2 composite fillers are 17.4 MPa and 16.2 MPa, respectively. These values are much higher than that of joints with the Cu50TiH2 filler (10.8 MPa [33]). Fig. 10 shows the SEM images of the interfacial area of the graphite/CuCrZr joints brazed with the composite fillers, revealing the favorable bonding between the filler layer and the substrates. The ceramic additives could serve as reinforcements in the filler layer. Furthermore, the residual stresses of joints may be relaxed by the introduction of reinforcements with relatively low CTE, which could contribute to the improvement of the shear strength [33].
4. Conclusions The wetting behavior of Cu50Ti alloy on h-BN and TiB2 ceramics has been studied under vacuum by a modified sessile drop method. The molten Cu50Ti alloy shows a good wetting on both hBN and TiB2 ceramics. The interactions occur at the interface between the molten Cu50Ti alloy and BN, leading to the formation of the reaction layer containing TiB and Ti nitrides. For Cu50Ti/TiB2 couple, a densely packed TiB layer about 60–100 μm thick is developed, and the detachment of the TiB2 grains at the interface occurs due to the penetration of molten alloy along the grain boundaries of TiB2. The spreading for the molten Cu50Ti alloy on h-BN is mainly controlled by the reactions between Ti and BN at the triple line. For the wetting of Cu50Ti/TiB2, the spreading is mainly limited by the interfacial reaction between Ti and TiB2 in the first stage, and possibly influenced by both the diffusion of boron atoms and viscous friction of the liquid in the second stage. Joining of graphite to CuCrZr has been realized using the composite filler containing Cu50TiH2 braze and ceramic additives including BN and TiB2. The reinforcements in the filler layer will contribute to the improvement of the shear strength of joints.
Fig. 10. SEM micrographs of the cross-section of graphite/CuCrZr joints brazed at 950 oC with holding time of 10 min using the composite fillers: (a) Cu50TiH2 þ BN and (b) Cu50TiH2 þTiB2.
9912
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
Acknowledgments The authors are deeply grateful to Prof. Ping Shen at Jilin University, China, for the wetting experiments support and discussions. The authors also acknowledge the financial support from the National Natural Science Foundation of China (Grant No. 51304148) and the Scientific Research Project under Hubei Provincial Department of Education (No. D20131504).
References [1] J. Zhang, T. Wang, C. Liu, Y. He, Effect of brazing temperature on microstructure and mechanical properties of graphite/copper joints, Mater. Sci. Eng. A. 594 (2014) 26–31. [2] Y. Mao, S. Yu, Y. Zhang, B. Guo, Z. Ma, Q. Deng, Microstructure analysis of graphite/Cu joints brazed with (Cu-50TiH2) þ B composite filler, Fusion Eng. Des. 100 (2015) 152–158. [3] M. Singh, R. Asthana, Characterization of brazed joints of carbon-carbon composites to Cu-clad-Mo, Compos. Sci. Technol. 68 (2008) 3010–3019. [4] R. Fedele, A. Ciani, L. Galantucci, V. Casalegno, A. Ventrella, M. Ferraris, Characterization of innovative CFC/Cu joints by full-field measurements and finite elements, Mater. Sci. Eng. A 595 (2014) 306–317. [5] M. Singh, T. Ohji, R. Asthana, S. Mathur, Ceramic Integration and Joining Technologies From Macro to Nanoscale, first ed., Wiley-American Ceramic Society, 2011. [6] M. Salvo, V. Casalegno, S. Rizzo, F. Smeacetto, M. Ferraris, M. Merola, One-step brazing process to join CFC composites to copper and copper alloy, J. Nucl. Mater. 374 (2008) 69–74. [7] Z. Zhong, Z. Zhou, C. Ge, Brazing of doped graphite to Cu using stress relief interlayers, J. Mater. Process. Technol. 209 (2009) 2662–2670. [8] J. Janczak-Rusch, G. Kaptay, L.P.H. Jeurgens, Interfacial design for joining technologies: an historical perspective, J. Mater. Eng. Perform. 23 (2014) 1608–1613. [9] W. Ding, J. Xu, Z. Chen, H. Su, Y. Fu, Brazed joints of CBN grains and AISI 1045 steel with AgCuTi–TiC mixed powder as filler materials, Int. J. Min. Metall. Mater. 18 (2011) 717–724. [10] G. Blugan, J. Kuebler, V. Bissig, J. Janczak-Rusch, Brazing of silicon nitride ceramic composite to steel using SiC-particle-reinforced active brazing alloy, Ceram. Int. 33 (2011) 1033–1039. [11] M.C. Halbig, B.P. Coddington, R. Asthana, M. Singh, Characterization of silicon carbide joints fabricated using SiC particulate-reinforced Ag–Cu–Ti alloys, Ceram. Int. 39 (2013) 4151–4162. [12] G. Lin, J. Huang, H. Zhang, X. Zhao, Joints of Cf/SiC composite to Ti-alloy with in-situ synthesized TiCx improved brazing layers, Mater. Trans. 47 (2006) 1261–1263. [13] M. Yang, T. Lin, P. He, Cu þ TiB2 composite filler for brazing Al2O3 and Ti-6Al-4V alloy, J. Alloy. Compd. 512 (2012) 282–289. [14] Z. Yang, L. Zhang, W. Ren, M. Lei, J. Feng, Interfacial microstructure and strengthening mechanism of BN-doped metal brazed Ti/SiO2-BN joints, J. Eur. Ceram. Soc. 33 (2013) 759–768. [15] Y.V. Naidich, in: D.A. Cadenhead, J.F. Danielli (eds.), Progress in Surface and Membrane Science, vol. 14, Academic Press, New York, p. 353. [16] H. Fujii, H. Nakae, K. Okada, Interfacial reaction wetting in the boron nitride/ molten aluminum system, Acta Metall. Mater. 41 (1993) 2963–2971. [17] P. Shen, H. Fujii, K. Nogi, Effect of temperature and surface roughness on the wettability of boron nitride by molten Al, J. Mater. Sci. 42 (2007) 3564–3568. [18] M.G. Nicholas, D.A. Mortimer, L.M. Jones, R.M. Crispin, Some observations on the wetting and bonding of nitride ceramics, J. Mater. Sci. 25 (1990) 2679–2689.
[19] B. Drevet, N. Eustathopoulos, Wetting of ceramics by molten silicon and silicon alloys: a review, J. Mater. Sci. 47 (2012) 8247–8260. [20] S. Umebayashi, K. Kishi, E. Tani, K. Kobayashi, H. Ito, R. Nakamura, Wetting of some ceramic materials by Ni–Mo alloy, J. Ceram. Soc. Jpn. 93 (1985) 51–57. [21] I. Harter, P. Dusserre, T. Duffar, J.-Ph Nabot, N. Eustathopoulos, Wetting of III-V melts on crucible materials, J. Cryst. Growth 131 (1993) 157–164. [22] S.J. Park, H. Fujii, H. Nakae, Wetting of boron nitride by molten Al–Si alloys, J. Jpn. Inst. Met. 58 (1994) 208–214. [23] N. Eustathopoulos, M.G. Nicholas, B. Drevet, Wettability at High Temperatures, Pergamon, Amsterdam, 1999. [24] M. Aizenshtein, N. Frage, N. Froumin, E. Shapiro-Tsoref, M.P. Dariel, Interface interaction in the (B4C þ TiB2)/Cu system, J. Mater. Sci. 41 (2006) 5185–5189. [25] M. Aizenshtein, N. Froumin, P. Barth, E. Shapiro-Tsoref, M.P. Dariel, N. Frage, How does the composition of quasi-stoichiometric titanium diboride affect its wetting by molten Cu and Au? J. Alloy. Compd. 442 (2007) 375–378. [26] M. Aizenshtein, N. Froumin, N. Frage, The nature of TiB2 wetting by Cu and Au, J. Mater. Eng. Perform. 21 (2012) 655–659. [27] D.A. Weirauch Jr., W.J. Krafick, G. Ackart, P.D. Ownby, The wettability of titanium diboride by molten aluminum drops, J. Mater. Sci. 40 (2005) 2301–2306. [28] C.T. Mutale, W.J. Krafick, D.A. Weirauch Jr., Direct observation of wetting and spreading of molten aluminum on TiB2 in the presence of a molten flux from the aluminum melting point up to 1033 K (760 °C), Metall. Mater. Trans. B 41 (2010) 1368–1374. [29] V. Ghetta, N. Gayraud, N. Eustathopoulos, Wetting of iron on sintered TiB2, Solid State Phenom. 25&26 (1992) 105–114. [30] L. Xi, I. Kaban, R. Nowak, B. Korpala, G. Bruzda, N. Sobczak, N. Mattern, J. Eckert, High-temperature wetting and interfacial interaction between liquid Al and TiB2 ceramic, J. Mater. Sci. 50 (2015) 2682–2690. [31] L. Xi, I. Kaban, R. Nowak, G. Bruzda, N. Sobczak, J. Eckert, Interfacial interactions between liquid Ti–Al alloys and TiB2 ceramic, J. Mater. Sci. 51 (2016) 1779–1787. [32] L.L. Yang, P. Shen, Q.L. Lin, F. Qiu, Q.C. Jiang, Wetting of porous graphite by Cu– Ti alloys at 1373 K, Mater. Chem. Phys. 124 (2010) 499–503. [33] Y. Mao, S. Wang, L. Peng, Q. Deng, P. Zhao, B. Guo, Y. Zhang, Brazing of graphite to Cu with Cu50TiH2 þC composite filler, J. Mater. Sci. 51 (2016) 1671–1679. [34] W.A. Soffa, D.E. Laughlin, High-strength age hardening copper-titanium alloys: redivivus, Prog. Mater. Sci. 49 (2004) 347–366. [35] E. Faran, I. Gotman, E.Y. Gutmanas, Coating of BN via solid state reaction with Ti powder, Mater. Lett. 43 (2000) 192–196. [36] W. Ding, J. Xu, Y. Fu, B. Xiao, H. Su, H. Xu, Interfacial reaction between cubic boron nitride and Ti during active brazing, J. Mater. Eng. Perform. 15 (2006) 365–369. [37] A.K. Chattopadhyay, H.E. Hintermann, On brazing of cubic boron nitride abrasive crystals to steel substrate with alloys containing Cr or Ti, J. Mater. Sci. 28 (1993) 5887–5893. [38] R. Yazdi, S.F. Kashani-Bozorg, Microstructure and wear of in situ Ti/(TiN þTiB) hybrid composite layers produced using liquid phase process, Mater. Chem. Phys. 152 (2015) 147–157. [39] Z. Fan, Z.X. Guo, B. Cantor, The kinetics and mechanism of interfacial reaction in sigma fibre-reinforced Ti MMCs, Compos. Part: A 28 (1997) 131–140. [40] X. Ma, C. Li, W. Zhang, Study on the phase diagram of the Ti–B–N system and the interfacial reaction of the Ti/BN joints, Mater. Sci. Eng. A 392 (2005) 394–402. [41] N. Eustathopoulos, R. Voytovych, The role of reactivity in wetting by liquid metals: a review, J. Mater. Sci. 51 (2016) 425–437. [42] O. Dezellus, F. Hodaj, N. Eustathopoulos, Progress in modeling of chemicalreaction limited wetting, J. Eur. Ceram. Soc. 23 (2003) 2797–2803. [43] P. He, W. Yang, T. Lin, Improving the strength of brazed joints with in situ synthesized TiB whiskers, J. Alloy. Compd. 509 (2011) 289–292. [44] M. Sobhani, H. Arabi, A. Mirhabibi, Microstructure evolution of copper-titanium alloy during in-situ formation of TiB2 particles, Trans. Nonferr. Metal. Soc. 23 (2013) 2994–3001.
Contents lists available at ScienceDirect
Ceramics International journal homepage: www.elsevier.com/locate/ceramint
Wetting behavior and interfacial interactions of molten Cu50Ti alloy with hexagonal BN and TiB2 ceramics Yangwu Mao a,n, Liangxing Peng a, Quanrong Deng a, Dunwei Nie a, Shenggao Wang a, Lixia Xi b a b
Hubei Key Laboratory of Plasma Chemistry and Advanced Materials, Wuhan Institute of Technology, Wuhan 430073, China Institute for Complex Materials, IFW Dresden, Dresden 01171, Germany
art ic l e i nf o
a b s t r a c t
Article history: Received 21 December 2015 Received in revised form 9 March 2016 Accepted 11 March 2016 Available online 12 March 2016
Wetting behavior of molten Cu50Ti alloy on hexagonal BN (h-BN) and TiB2 ceramics has been studied under vacuum using a modified sessile drop method. Final contact angles of 8° and 3° are obtained at 1000 °C on h-BN and TiB2, respectively. Interaction occurs at the interface between the molten alloy and BN, leading to the formation of a reaction layer containing TiB and Ti nitrides. Interfacial interaction of Cu50Ti with TiB2 results in the formation of densely packed TiB layer about 60–100 μm thick and the detachment of TiB2 grains. Spreading wetting of liquid Cu50Ti on h-BN is mainly controlled by the reactions between Ti and BN at the triple line. For Cu50Ti/TiB2 system, spreading is mainly limited by the interfacial reaction in the first stage, and is possibly influenced by both the diffusion of boron atoms and viscous friction of the liquid in the second stage. Finally, brazing of graphite to CuCrZr alloy has been realized using Cu50TiH2 with ceramic additives (including BN and TiB2) as composite fillers. The joints exhibit favorable interfacial bonding between the filler layer and the substrates. The ceramic reinforcements in the filler layer could contribute to the improvement of the shear strength. & 2016 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
Keywords: A. Joining B. Interfaces D. Borides D. Nitrides
1. Introduction Brazing technology, as one of the most effective joining techniques, has been widely used for the fabrication of ceramic/metal joints. For example, carbon/metal (including graphite/Cu and carbon-carbon composite/Cu-clad Mo) joints have been prepared by brazing for the industry applications [1–3]. However, one of the major issues for brazing ceramics to metals is the existence of large residual stresses, which are mainly generated by the mismatch of coefficient of thermal expansion (CTE) and elastic modulus between the substrates [4,5]. The residual stresses of joints may lead to cracks in the ceramic and even the failure of joints upon cooling from the processing temperature. The metal interlayers and composite fillers have been introduced to alleviate the residual stresses of the ceramic/metal joints [6–14]. The composite fillers are generally composed of active metal brazes and ceramic reinforcements. The residual stresses of joints may be relaxed by the introduction of ceramic reinforcements with relatively low CTE. In addition, the ceramic additives in the composite fillers are assumed to be distributed uniformly in the filler layer, which can develop the reinforced n
Corresponding author. E-mail addresses: [email protected], [email protected] (Y. Mao).
http://dx.doi.org/10.1016/j.ceramint.2016.03.090 0272-8842/& 2016 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
structure in the joints. Halbig et al. [11] studied the brazing of SiC using Ag–Cu–Ti braze alloys reinforced with SiC particulates. Based on the theoretical calculations, the CTE of the braze with the incorporation of about 45 vol% SiC could decrease by nearly 45– 60%. As reported in Ref. [13], the synthesized TiB whiskers contributed to the strength improvement of Al2O3/Ti–6Al–4V joints. It should be noted that the microstructure and properties of the joints may be influenced by the wettability and interactions between liquid metal brazes and ceramic reinforcements. Thus, it is crucial to study the wetting behavior between the metal brazes and ceramic reinforcements prior to the design and preparation of the composite fillers for joining. The reinforcements used in the composite fillers generally include carbides, nitrides, borides, and so on. Among them, BN and TiB2 ceramics have been used as reinforcements in Refs. [13,14]. Extensive studies have been performed on the wetting of BN by liquid metals [15–22]. Naidich et al. [15] measured the contact angles in the range of 135–150° at 1000–1500 °C for the metals with a negligible (Ag, Sn, Au) or weak (Ge, Ga, Cu) affinity for both N and B, and contact angles lower than 90 ° for ferrous metals. For Al/BN system, nearly perfect wetting is achieved at 1000 °C and a continuous AlN layer is formed at the interface by the reaction between liquid Al and BN [16]. Nicholas et al. [18] studied the wetting of BN by addition of Ti to Cu and Ag–Cu. Very low contact angles were obtained at both 1150 °C and 950 °C. Furthermore, the
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
wetting of BN by liquid alloys including Si, Ni–Mo (0–40 wt%), GaSb, InSb, and Al–Si has been reported in Refs. [19–22]. Investigations on the wetting of TiB2 by liquid metals, including Cu, Au, Ni, Al, Fe and Ti, have been extensively carried out [23–31]. For Cu/TiB2 system, both wetting and non-wetting contact angles have been reported by different researchers [23]. The molten Cu and Au show a good wetting on non-stoichiometric TiB2 substrates (TiB1.9 and TiB1.95). Some limited boride dissolution and alteration of the substrate composition occur at the TiBx/Cu and TiBx/Au interfaces [24–26]. In the case of Ni/TiB2 system [23], TiB2 is dissolved into the liquid Ni, resulting in a good wetting in either high vacuum or neutral gas environments. Weirauch Jr et al. [27] investigated the wettability of molten aluminum drops on four different types of TiB2 substrates. The effect of the substrate microstructure on the wetting kinetics has been discussed. Mutale et al. [28] introduced the spreading wetting of TiB2 substrates by molten aluminum in the temperature range between 660 °C and 760 °C in different fluxes. Ghetta et al. [29] performed the study on the wetting of sintered TiB2 by pure iron and iron containing dissolved TiB2. The wetting deteriorated with the increase of oxygen content in the TiB2 substrates. Xi et al. [30–31] presented a good wetting of molten Al, Ti and Ti–Al alloys on TiB2. The formation of TiB occurs at the interfaces of Ti/TiB2 and Ti74.3Al25.7/TiB2. In this study, the wetting behavior and interfacial interactions of Cu50Ti alloy with h-BN and TiB2 ceramics have been investigated. Such a study can provide a reference for joining graphite to Cu alloys with the composite fillers composed of Cu50Ti metal braze and ceramic reinforcements (including BN and TiB2).
9907
Fig. 1. Schematic of the modified sessile drop method.
spreading processes of the Cu50Ti alloy on the ceramic were monitored by a high-speed CMOS camera (frame rate of 120 fps and resolution of 640 480 pixels). The time-dependent variation in contact angle during isothermal wetting at 1000 °C was tracked. Both the contact angle (the error of 0.5°) and drop diameter were measured from the droplet images by using an axisymmetricdrop-shape analysis (ADSA) program. 2.2. Joining of graphite/CuCrZr using composite fillers
2. Materials and methods 2.1. Wetting of Cu50Ti alloy on h-BN and TiB2 Hexagonal BN (h-BN) ceramics (Φ 20 mm 5 mm) with a purity over 99 wt%, and TiB2 ceramics (20 mm 20 mm 3 mm) with 4–5 wt% sintering aid containing Ni, were used as the wetting substrates. Both the h-BN and TiB2 ceramics were supplied from Key Laboratory of Automobile Materials, Jilin University, China. The surfaces of the substrates were mechanically ground and subsequently polished using diamond suspensions. The Cu50Ti alloy with 50 wt% Ti was prepared from high-purity Cu (99.999 wt%) and Ti (99.995 wt%) plates by arc-melting in a purified Ti-gettered argon atmosphere. The molten alloy was turned in a water-cooled copper crucible and remelted for four times by electromagnetism stirring to ensure a good homogeneity. Then, the alloy was cut into small cubic pieces weighing about 150 mg. The ceramic substrates and the Cu50Ti alloy pieces were ultrasonically cleaned in ethanol prior to wetting tests. The dispensed drop method is unsuitable in our study because of a severe reaction of the Ti-containing melts with the drop dispenser made of alumina. Accordingly, a modified sessile drop method described elsewhere [32] was adopted. The solid bulk alloy, rather than the liquid alloy droplet, was dropped on the substrate surface from the alumina tube. Fig. 1 gives the schematic of the modified sessile drop method. The distance between the end of the tube and the substrate surface was about 6 mm. The ceramic substrate was placed in the vacuum chamber and kept to a horizontal position, while the Cu50Ti alloy specimen was stored in a stainless-steel tube outside the chamber. The chamber was evacuated to a vacuum about 2 10 4 Pa at room temperature, and then heated to the testing temperature of 1000 °C at a rate of 20 °C/min. Then, the solid Cu50Ti alloy was delivered to the surface of the ceramic through an open alumina tube which was connected with the stainless-steel tube. Indeed, the alloy was not melted when it contacted the substrate surface. The melting and
The commercial graphite (10 mm 10 mm 10 mm) with a density of 1.9 g/cm3 and purity of 99.99% was purchased from Changsha Aobo Carbon Co., Ltd., China. The commercial CuCrZr alloy (10 mm 10 mm 10 mm) with a nominal composition of Cu- (0.4–1) Cr- (0.03–0.15) Zr (in wt%) and a density of 8.9 g/cm3 was supported from Shenzhen Heshuo Metal Products Co., Ltd, China. Both the graphite and CuCrZr alloy were used as joining substrates. The substrates were polished with 1.0 μm diamond paste and then ultrasonically cleaned in alcohol prior to brazing experiments. The composite filler was composed of Cu powders, TiH2 powders (replacement of Ti powders, to avoid the oxidation during the mechanical milling) and ceramic additives. The detailed description of Cu powders and TiH2 powders could be found in Ref. [33]. BN powders with the size of about 1 μm and purity of 99.0%, and TiB2 powders with the size of about 3–5 μm and purity of 99.5%, were supported from Qinhuangdao Eno High-Tech Material Development Co., Ltd, China. The powders of Cu, TiH2 and ceramic additives were mixed together by mechanical milling to obtain Cu50TiH2 þ BN (or Cu50TiH2 þTiB2) composite filler. The weight ratio of Cu and TiH2 powders was kept at 1:1, and the content of ceramic additives was 2 wt% in the composite filler. Considering that the melting temperature of the powder filler could be lower than that of bulk alloy, the joining temperature was set at 950 °C (with the holding time of 10 min). The detailed joining procedure and the shear strength measurement of joints were described in Ref. [33]. 2.3. Microstructure characterization The wetting specimens for cross-section view were prepared in an epoxy mount and then polished. The microstructure of the cross-section was characterized by a scanning electron microscope (SEM, Quanta200, Holland) equipped with an energy dispersive X-ray spectrometer (EDS, SDD Inca X-Max50, Holland). The phase identification of interfacial area was determined by an X-ray
9908
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
Fig. 2. Variations in contact angle and normalized drop diameter with time for the wetting of Cu50Ti/h-BN system at 1000 °C.
diffraction spectrometer (XRD, Bruker D8 Advance, Germany) with a scanning speed of 4°/min. The microstructure characterization of the interfacial area for the graphite/CuCrZr joints was also examined by means of SEM.
3. Results and discussions 3.1. Wetting of Cu50Ti/h-BN system The melting temperature of 950 °C can be obtained for Cu50Ti eutectic alloy, according to the Cu–Ti phase diagram [34]. Taking
into account that the brazing temperature is about tens of degree higher than the melting temperature of the filler, the isothermal wetting for liquid Cu50Ti alloy on the ceramics has been investigated at 1000 °C. Fig. 2 shows the variations of contact angle and normalized drop diameter with time for the wetting of Cu50Ti/h-BN system. The measured initial contact angle is 42° at 1000 °C. The contact angle decreases dramatically in the range of 0–10 s, and then decreases slowly in the range of 10–30 s. It ultimately reaches a steady contact angle of 8°. The spreading of the molten Cu50Ti alloy on the h-BN, as reflected by the normalized diameter, is rapid at the beginning, and then becomes slow. The normalized diameter remains unchanged when the wetting reaches equilibrium. Fig. 3 displays the SEM micrograph and EDS element distribution of the cross-section of Cu50Ti/h-BN couple after wetting at 1000 °C. It can be seen from Fig. 3(a) that the reaction layer between the alloy and BN consists of two morphological sub-layers: a continuous layer about 5 μm thick close to BN and a layer containing amounts of needles extending to the alloy drop. The EDS element distribution shown in Fig. 3(b)–(d) indicates that the interfacial area is mainly composed of element Cu and Ti as well as N. The enrichments of element Ti can be observed at the interface, suggesting the formation of reaction layer containing Ti–B and Ti– N compounds. The XRD patterns of the interfacial area, as shown in Fig. 4, confirm the presence of TiB, TiN and Ti2N in the interfacial area. In addition, the cracks are observed in the h-BN substrate. They are mainly caused by the large residual thermal stresses, which are generated due to the big CTE mismatch between the alloy and h-BN during the cooling stage after wetting tests. The reaction evolution between Ti and BN has been investigated by different researchers [18,35–38]. The study
Fig. 3. SEM micrograph (a) and EDS areal distribution (b)–(d) of the cross-section of the Cu50Ti/h-BN couple after wetting at 1000 °C.
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
9909
Fig. 4. The XRD patterns of the interfacial area for the Cu50Ti/BN couple.
performed by Nicholas et al. [18] revealed that Ti nitrides (from TiN to TiN0.43) were formed through the reaction of BN with Ti (BN þTi¼ TiN þB), and then the liberated boron reacted with Ti to form Ti borides (including TiB, Ti3B4 and TiB2) adjacent to the metal drop. However, Faran et al. [35] observed a thin TiB2 layer formed close to BN followed by the much thicker TiB (0 2 0) layer through the reaction between BN and Ti (BN þTi-TimBn þ N). The nitrogen released from the reaction could diffuse through the boride layer and then react with excess Ti to form Ti nitrides. In addition, according to the microstructure analysis obtained by Ding et al. [36], the composite TiB2–TiN reaction layer was developed by the interfacial reaction between the cubic BN and Ti from the braze filler. In our study, a TiB reaction layer was identified close to BN side, but no TiB2 phase was detected in the interfacial area. As argued in Ref. [39], TiB2 is just a transition product of the interfacial reaction because TiB2 is not in equilibrium with Ti, even though the ΔG value of TiB2 formation at 1000 °C is more negative than that of TiB formation. It is, therefore, believable that TiB instead of TiB2 is developed in the reaction layer close to BN area in the Cu50Ti/hBN system. The nitrogen atoms released from the reaction of Ti with BN diffuse through the TiB layer, and then react with Ti to form Ti nitrides, i.e. TiN and Ti2N. According to the Ti–B–N ternary diagram [40] (in Fig. 5), TiB can be in equilibrium with Ti nitrides only when the concentration of nitrogen is less than 40 at% [35]. The diffusion path of the reaction products shown in dashed blue tie line in Fig. 5 is similar to the results in Ref. [35], suggesting that both TiB and Ti nitrides exist in the reaction layer of Cu50Ti/h-BN couple. As illustrated by Eustathopoulos et al. [41], the spreading time of reactive metal/ceramic systems (in the range 10–104 s) could be several orders of magnitude higher than that of non-reactive systems (E 10 2 s). Thus, the spreading rate of the reactive stage for a given reactive system may be controlled by the interfacial reaction itself. In view of the interfacial reactions in Cu50Ti/BN system, the reaction product control model is applied to study the spreading kinetics. The characteristic equation of chemical reaction-limited reactive wetting was proposed by Dezellus et al. [42] as follows:
cos θe − cos θd = (cos θe − cos θ 0 ) exp ( − kt )
Fig. 5. The isothermal section of B–N–Ti ternary diagram at 1000 °C [40] with the corresponding diffusion path.
used as θe and θ0 in the Eq. (1) for the Cu50Ti/BN system at 1000 °C. After taking napierian logarithm, the transformative equation yields:
ln (cos θe − cos θd ) = ln (cos θe − cos θ 0 ) − kt
(2)
The result of ln(cos θe cos θd) versus time for the wetting of Cu50Ti on h-BN is shown in Fig. 6. The experimental data is in good agreement with the linearity prediction for the whole spreading stage, suggesting that the spreading of liquid Cu50Ti alloy on h-BN substrate is mainly controlled by the interfacial reactions. The value of the fitting dynamic constant k is about 0.2/s based on the calculation. 3.2. Wetting of Cu50Ti/TiB2 system The wetting behavior of Cu50Ti/TiB2 system at 1000 °C, as depicted in Fig. 7, is similar to that of Cu50Ti/h-BN system. The contact angle rapidly decreases from the initial 57° to 17° (∼5 s),
(1)
Where θe, θd, and θ0 are the equilibrium, dynamic, and contact angle of the liquid on the unreacted substrate, respectively, and k is a dynamic constant, representing spreading rate. The measured equilibrium and initial contact angles, 8° and 42°, respectively, are
Fig. 6. Napierian logarithm of cos θe cos θd versus time t for the Cu50Ti/h-BN wetting system at 1000 °C.
9910
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
Fig. 7. Variations in contact angle and normalized drop diameter with time for the wetting of Cu50Ti/TiB2 at 1000 °C.
and then gradually towards a value of 3° (∼25 s), which corresponds to the excellent wetting. Further holding at 1000 °C produces negligible changes of the contact angle. The normalized diameter of the Cu50Ti alloy droplet increases rapidly in the beginning and then slowly in the later stage. The diameter of the droplet eventually remains constant. Fig. 8 gives the SEM micrograph, EDS element distribution and XRD patterns of the interfacial area for the Cu50Ti/TiB2 couple after wetting at 1000 °C. In Fig. 8(a), a stack of densely packed whiskers extending up to about 60–100 μm are formed at the interface. The whiskers distributed in the interfacial area are identified as TiB phase according to the EDS analysis and XRD patterns in Fig. 8(d). The ΔG value of the reaction (TiB2 þTi ¼2TiB)
at 1000 °C, calculated using the software HSC Chemistry Version 6 (Outokumpu Ra, Oy, Finland), is negative ( 52.5 kJ/mol), confirming the possibility of TiB whiskers formation in the interfacial area. The formation of TiB by the reaction of Ti with TiB2 is in agreement with the results revealed in Refs. [43,44]. In Ref. [43], TiB whiskers were formed in the interfacial area due to the reaction of TiB2 particles with Ti during brazing Al2O3 to Ti6Al4V alloy with Cu–Ti þTiB2 composite filler. Sobhani et al. [44] demonstrated the formation of TiB whiskers connecting with TiB2 particles for the preparation of TiB2 particles reinforced Cu–Ti composites. In addition, the detachment of the TiB2 grains can be observed at the interface in Cu50Ti/TiB2 couple, which is mainly caused by the penetration of molten Cu50Ti alloy along the grain boundaries of TiB2. Similar microstructure and morphology of the interfacial area have been reported in Ti-25Al/TiB2 couple by Xi et al. [31], who also presented the formation of the stack of densely packed TiB whiskers and the detachment of the TiB2 grains at the interface. The EDS results of microzones E and F marked in Fig. 8(a) give the compositions of Cu36Ti59Ni5 and Cu27Ti72Ni1, respectively, indicating the formation of Ti–Cu intermetallics in the interfacial area besides TiB whiskers. The existence of element Ni in the metallic alloy is attributed to the diffusion of the sintering aid Ni in the TiB2 substrate. The EDS element distribution reveals the distribution of the Cu in the metallic alloy and enrichments of Ti in the interfacial area, which confirms the formation of TiB whiskers and the detachment of TiB2 grains at the interface. In addition, the cracks in the TiB2 substrate are mainly resulted from the residual thermal stresses originated from the big CTE mismatch of the wetting couple. In view of the interfacial reaction (formation of TiB) in Cu50Ti/TiB2 system, the reaction-limited spreading model is
Fig. 8. SEM micrograph (a), EDS element distribution of Ti and Cu (b)–(c) and XRD patterns (d) of the interfacial area of Cu50Ti/TiB2 couple after wetting at 1000 oC.
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
9911
influence the spreading. It is, therefore, possible that the spreading of liquid Cu50Ti alloy on TiB2 is influenced by both the diffusion of boron and viscous friction of the liquid in the second stage (5– 25 s). 3.3. Joining of graphite/CuCrZr with composite fillers
Fig. 9. Napierian logarithm of cos θe cos θd versus time t for the Cu50Ti/TiB2 wetting system at 1000 °C.
applied to interpret the spreading kinetics. Fig. 9 displays the napierian logarithm of cos θe cos θd versus time for Cu50Ti/TiB2 system. It appears that ln(cos θe cos θd) versus time is linear in the first stage (0–5 s) with a slope about 0.5/s. At this stage, the boron atoms can diffuse easily from TiB2 to the liquid alloy drop at the triple line and satisfy the demand of the interfacial reaction. It is indisputable that the spreading wetting of the Cu50Ti/TiB2 system in the first stage is mainly controlled by the interfacial reaction at the triple line. With the progress of the interaction between liquid alloy and TiB2, the spreading rate of liquid on TiB2 is low (less than 0.1 mm/ s), corresponding to a contact angle lower than 20°. The spreading of the liquid at the second stage may be influenced by two aspects. On the one hand, the diffusion of boron atoms outwards from TiB2 to the triple line becomes difficult due to the formation of densely packed TiB whiskers at the interface. As a consequence, the diffusion of boron may influence the spreading of the liquid alloy on TiB2. On the other hand, as argued by Eustathopoulos et al. [23], the inertial force is more predominant than viscous force during the initial spreading in non-reactive systems, since the viscosity of liquid metals and alloys is very low. However, the viscous friction could not be negligible in the late stages of spreading for nearly perfect wetting systems [41]. Taking into account the excellent wetting (very low contact angles) for Cu50Ti/TiB2 system in the second stage (5–25 s), the viscous friction of the liquid may
The composite fillers composed of Cu50TiH2 braze and ceramic additives including BN and TiB2 have been applied for joining graphite to CuCrZr alloy. The shear strength of the joints with Cu50TiH2 þ BN and Cu50TiH2 þTiB2 composite fillers are 17.4 MPa and 16.2 MPa, respectively. These values are much higher than that of joints with the Cu50TiH2 filler (10.8 MPa [33]). Fig. 10 shows the SEM images of the interfacial area of the graphite/CuCrZr joints brazed with the composite fillers, revealing the favorable bonding between the filler layer and the substrates. The ceramic additives could serve as reinforcements in the filler layer. Furthermore, the residual stresses of joints may be relaxed by the introduction of reinforcements with relatively low CTE, which could contribute to the improvement of the shear strength [33].
4. Conclusions The wetting behavior of Cu50Ti alloy on h-BN and TiB2 ceramics has been studied under vacuum by a modified sessile drop method. The molten Cu50Ti alloy shows a good wetting on both hBN and TiB2 ceramics. The interactions occur at the interface between the molten Cu50Ti alloy and BN, leading to the formation of the reaction layer containing TiB and Ti nitrides. For Cu50Ti/TiB2 couple, a densely packed TiB layer about 60–100 μm thick is developed, and the detachment of the TiB2 grains at the interface occurs due to the penetration of molten alloy along the grain boundaries of TiB2. The spreading for the molten Cu50Ti alloy on h-BN is mainly controlled by the reactions between Ti and BN at the triple line. For the wetting of Cu50Ti/TiB2, the spreading is mainly limited by the interfacial reaction between Ti and TiB2 in the first stage, and possibly influenced by both the diffusion of boron atoms and viscous friction of the liquid in the second stage. Joining of graphite to CuCrZr has been realized using the composite filler containing Cu50TiH2 braze and ceramic additives including BN and TiB2. The reinforcements in the filler layer will contribute to the improvement of the shear strength of joints.
Fig. 10. SEM micrographs of the cross-section of graphite/CuCrZr joints brazed at 950 oC with holding time of 10 min using the composite fillers: (a) Cu50TiH2 þ BN and (b) Cu50TiH2 þTiB2.
9912
Y. Mao et al. / Ceramics International 42 (2016) 9906–9912
Acknowledgments The authors are deeply grateful to Prof. Ping Shen at Jilin University, China, for the wetting experiments support and discussions. The authors also acknowledge the financial support from the National Natural Science Foundation of China (Grant No. 51304148) and the Scientific Research Project under Hubei Provincial Department of Education (No. D20131504).
References [1] J. Zhang, T. Wang, C. Liu, Y. He, Effect of brazing temperature on microstructure and mechanical properties of graphite/copper joints, Mater. Sci. Eng. A. 594 (2014) 26–31. [2] Y. Mao, S. Yu, Y. Zhang, B. Guo, Z. Ma, Q. Deng, Microstructure analysis of graphite/Cu joints brazed with (Cu-50TiH2) þ B composite filler, Fusion Eng. Des. 100 (2015) 152–158. [3] M. Singh, R. Asthana, Characterization of brazed joints of carbon-carbon composites to Cu-clad-Mo, Compos. Sci. Technol. 68 (2008) 3010–3019. [4] R. Fedele, A. Ciani, L. Galantucci, V. Casalegno, A. Ventrella, M. Ferraris, Characterization of innovative CFC/Cu joints by full-field measurements and finite elements, Mater. Sci. Eng. A 595 (2014) 306–317. [5] M. Singh, T. Ohji, R. Asthana, S. Mathur, Ceramic Integration and Joining Technologies From Macro to Nanoscale, first ed., Wiley-American Ceramic Society, 2011. [6] M. Salvo, V. Casalegno, S. Rizzo, F. Smeacetto, M. Ferraris, M. Merola, One-step brazing process to join CFC composites to copper and copper alloy, J. Nucl. Mater. 374 (2008) 69–74. [7] Z. Zhong, Z. Zhou, C. Ge, Brazing of doped graphite to Cu using stress relief interlayers, J. Mater. Process. Technol. 209 (2009) 2662–2670. [8] J. Janczak-Rusch, G. Kaptay, L.P.H. Jeurgens, Interfacial design for joining technologies: an historical perspective, J. Mater. Eng. Perform. 23 (2014) 1608–1613. [9] W. Ding, J. Xu, Z. Chen, H. Su, Y. Fu, Brazed joints of CBN grains and AISI 1045 steel with AgCuTi–TiC mixed powder as filler materials, Int. J. Min. Metall. Mater. 18 (2011) 717–724. [10] G. Blugan, J. Kuebler, V. Bissig, J. Janczak-Rusch, Brazing of silicon nitride ceramic composite to steel using SiC-particle-reinforced active brazing alloy, Ceram. Int. 33 (2011) 1033–1039. [11] M.C. Halbig, B.P. Coddington, R. Asthana, M. Singh, Characterization of silicon carbide joints fabricated using SiC particulate-reinforced Ag–Cu–Ti alloys, Ceram. Int. 39 (2013) 4151–4162. [12] G. Lin, J. Huang, H. Zhang, X. Zhao, Joints of Cf/SiC composite to Ti-alloy with in-situ synthesized TiCx improved brazing layers, Mater. Trans. 47 (2006) 1261–1263. [13] M. Yang, T. Lin, P. He, Cu þ TiB2 composite filler for brazing Al2O3 and Ti-6Al-4V alloy, J. Alloy. Compd. 512 (2012) 282–289. [14] Z. Yang, L. Zhang, W. Ren, M. Lei, J. Feng, Interfacial microstructure and strengthening mechanism of BN-doped metal brazed Ti/SiO2-BN joints, J. Eur. Ceram. Soc. 33 (2013) 759–768. [15] Y.V. Naidich, in: D.A. Cadenhead, J.F. Danielli (eds.), Progress in Surface and Membrane Science, vol. 14, Academic Press, New York, p. 353. [16] H. Fujii, H. Nakae, K. Okada, Interfacial reaction wetting in the boron nitride/ molten aluminum system, Acta Metall. Mater. 41 (1993) 2963–2971. [17] P. Shen, H. Fujii, K. Nogi, Effect of temperature and surface roughness on the wettability of boron nitride by molten Al, J. Mater. Sci. 42 (2007) 3564–3568. [18] M.G. Nicholas, D.A. Mortimer, L.M. Jones, R.M. Crispin, Some observations on the wetting and bonding of nitride ceramics, J. Mater. Sci. 25 (1990) 2679–2689.
[19] B. Drevet, N. Eustathopoulos, Wetting of ceramics by molten silicon and silicon alloys: a review, J. Mater. Sci. 47 (2012) 8247–8260. [20] S. Umebayashi, K. Kishi, E. Tani, K. Kobayashi, H. Ito, R. Nakamura, Wetting of some ceramic materials by Ni–Mo alloy, J. Ceram. Soc. Jpn. 93 (1985) 51–57. [21] I. Harter, P. Dusserre, T. Duffar, J.-Ph Nabot, N. Eustathopoulos, Wetting of III-V melts on crucible materials, J. Cryst. Growth 131 (1993) 157–164. [22] S.J. Park, H. Fujii, H. Nakae, Wetting of boron nitride by molten Al–Si alloys, J. Jpn. Inst. Met. 58 (1994) 208–214. [23] N. Eustathopoulos, M.G. Nicholas, B. Drevet, Wettability at High Temperatures, Pergamon, Amsterdam, 1999. [24] M. Aizenshtein, N. Frage, N. Froumin, E. Shapiro-Tsoref, M.P. Dariel, Interface interaction in the (B4C þ TiB2)/Cu system, J. Mater. Sci. 41 (2006) 5185–5189. [25] M. Aizenshtein, N. Froumin, P. Barth, E. Shapiro-Tsoref, M.P. Dariel, N. Frage, How does the composition of quasi-stoichiometric titanium diboride affect its wetting by molten Cu and Au? J. Alloy. Compd. 442 (2007) 375–378. [26] M. Aizenshtein, N. Froumin, N. Frage, The nature of TiB2 wetting by Cu and Au, J. Mater. Eng. Perform. 21 (2012) 655–659. [27] D.A. Weirauch Jr., W.J. Krafick, G. Ackart, P.D. Ownby, The wettability of titanium diboride by molten aluminum drops, J. Mater. Sci. 40 (2005) 2301–2306. [28] C.T. Mutale, W.J. Krafick, D.A. Weirauch Jr., Direct observation of wetting and spreading of molten aluminum on TiB2 in the presence of a molten flux from the aluminum melting point up to 1033 K (760 °C), Metall. Mater. Trans. B 41 (2010) 1368–1374. [29] V. Ghetta, N. Gayraud, N. Eustathopoulos, Wetting of iron on sintered TiB2, Solid State Phenom. 25&26 (1992) 105–114. [30] L. Xi, I. Kaban, R. Nowak, B. Korpala, G. Bruzda, N. Sobczak, N. Mattern, J. Eckert, High-temperature wetting and interfacial interaction between liquid Al and TiB2 ceramic, J. Mater. Sci. 50 (2015) 2682–2690. [31] L. Xi, I. Kaban, R. Nowak, G. Bruzda, N. Sobczak, J. Eckert, Interfacial interactions between liquid Ti–Al alloys and TiB2 ceramic, J. Mater. Sci. 51 (2016) 1779–1787. [32] L.L. Yang, P. Shen, Q.L. Lin, F. Qiu, Q.C. Jiang, Wetting of porous graphite by Cu– Ti alloys at 1373 K, Mater. Chem. Phys. 124 (2010) 499–503. [33] Y. Mao, S. Wang, L. Peng, Q. Deng, P. Zhao, B. Guo, Y. Zhang, Brazing of graphite to Cu with Cu50TiH2 þC composite filler, J. Mater. Sci. 51 (2016) 1671–1679. [34] W.A. Soffa, D.E. Laughlin, High-strength age hardening copper-titanium alloys: redivivus, Prog. Mater. Sci. 49 (2004) 347–366. [35] E. Faran, I. Gotman, E.Y. Gutmanas, Coating of BN via solid state reaction with Ti powder, Mater. Lett. 43 (2000) 192–196. [36] W. Ding, J. Xu, Y. Fu, B. Xiao, H. Su, H. Xu, Interfacial reaction between cubic boron nitride and Ti during active brazing, J. Mater. Eng. Perform. 15 (2006) 365–369. [37] A.K. Chattopadhyay, H.E. Hintermann, On brazing of cubic boron nitride abrasive crystals to steel substrate with alloys containing Cr or Ti, J. Mater. Sci. 28 (1993) 5887–5893. [38] R. Yazdi, S.F. Kashani-Bozorg, Microstructure and wear of in situ Ti/(TiN þTiB) hybrid composite layers produced using liquid phase process, Mater. Chem. Phys. 152 (2015) 147–157. [39] Z. Fan, Z.X. Guo, B. Cantor, The kinetics and mechanism of interfacial reaction in sigma fibre-reinforced Ti MMCs, Compos. Part: A 28 (1997) 131–140. [40] X. Ma, C. Li, W. Zhang, Study on the phase diagram of the Ti–B–N system and the interfacial reaction of the Ti/BN joints, Mater. Sci. Eng. A 392 (2005) 394–402. [41] N. Eustathopoulos, R. Voytovych, The role of reactivity in wetting by liquid metals: a review, J. Mater. Sci. 51 (2016) 425–437. [42] O. Dezellus, F. Hodaj, N. Eustathopoulos, Progress in modeling of chemicalreaction limited wetting, J. Eur. Ceram. Soc. 23 (2003) 2797–2803. [43] P. He, W. Yang, T. Lin, Improving the strength of brazed joints with in situ synthesized TiB whiskers, J. Alloy. Compd. 509 (2011) 289–292. [44] M. Sobhani, H. Arabi, A. Mirhabibi, Microstructure evolution of copper-titanium alloy during in-situ formation of TiB2 particles, Trans. Nonferr. Metal. Soc. 23 (2013) 2994–3001.