Journal of Materials Science & Technology 35 (2019) 88–93
Contents lists available at ScienceDirect
Journal of Materials Science & Technology journal homepage: www.jmst.org
Measurement of interfacial residual stress in SiC fiber reinforced Ni-Cr-Al alloy composites by Raman spectroscopy Xixi Niu a,b,1 , Haoqiang Zhang a,c,1 , Zhiliang Pei a , Nanlin Shi a , Chao Sun a , Jun Gong a,∗ a b c
Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China University of Chinese Academy of Sciences, Beijing 100049, China School of Materials Science and Engineering, University of Science Technology of China, Shenyang 110016, China
a r t i c l e
i n f o
Article history: Received 17 April 2018 Accepted 7 July 2018 Available online 15 September 2018 Keywords: Ni-Cr-Al alloy SiC fiber Composite Raman spectra Diffusion barrier coating Residual stress
a b s t r a c t Raman spectroscopy was used to measure Raman spectra of the inner SiC fibers and surface C-rich layers of SiC fibers, composite precursors and SiCf /Ni-Cr-Al composites. The residual stresses of the inner SiC fibers and surface C-rich layers were calculated, and the effect of the (Al + Al2 O3 ) diffusion barrier layer on the interfacial residual stress in the composites was analyzed in combination with the interface microstructure and energy disperse spectroscopy (EDS) elements lining maps. The results show that the existence of (Al + Al2 O3 ) diffusion barrier improves the compatibility of the SiCf /Ni-Cr-Al interface, inhibits the adverse interfacial reaction, and relieves the residual stress inside SiC fibers and at the interface of composite material. Heat treatment can reduce the residual stress at the interface. As the heat treatment time increases, the residual stress at the interface decreases. © 2018 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology.
1. Introduction The new generation of advanced aero-engines requires continuous research and development of new materials, especially composite materials, to reduce the weight of parts and increase the aero-engine thrust-weight ratio [1]. Ni-based superalloys are currently main materials in the preparation of thermal structures used above 700 ◦ C in aero-engines. However, high specific density limits their further application. SiC fiber reinforced Ni-based superalloy composites have higher specific strength, specific rigidity, high temperature resistance and structural stability, which makes it an effective way to reduce the weight of high temperature parts and improve mechanical properties [2]. The fabrication temperatures of SiC fiber reinforced Ni alloy composites are higher than 950 ◦ C. The coefficients of thermal expansion of SiC fibers and Ni alloys are quite different, which will cause great residual stress at the interface during fabrication of the composites. At the same time, SiC fibers and Ni alloys will react seriously and form new phase near the interface, resulting in the change of interface volume and the formation of residual
∗ Corresponding author. E-mail address:
[email protected] (J. Gong). 1 Contributed equally to this work.
stress. Excessive interface residual stress will form stress concentration near the interface, causing the initiation and propagation of interfacial cracks and degrading the strength and decreasing the mechanical properties of the composites [3]. The previous study [4] shows that the addition of an (Al + Al2 O3 ) diffusion barrier coating at the SiCf /Ni interface, while alleviating the thermal expansion coefficient mismatch between the fiber and the matrix alloy and inhibiting excessive interface reaction, can also protect the fiber Crich coating to ensure the bonding between the fiber and the matrix alloy and reduce the interfacial residual stress. There are many methods to determine the residual stress of metal matrix composites. X-ray diffraction [5] is common method of the residual stress measurement at present. However, the X-ray penetration ability is limited, so this method is only suitable to measure and estimate the average residual stress of the surface layer (5–7 m) of composites. Neutron diffraction [6] has strong penetrability, but it is costly and difficult to obtain. Laser Raman method [7,8] can calculate the residual stress in composites by measuring the displacement of the Raman peak under the effect of residual stress. This method has the advantages of high spatial resolution, no contact and damage samples, wide spectrum range and so on, and the residual stress distribution can be obtained, which can better reflect the microscopic residual stress distribution in the interface of the composite material.
https://doi.org/10.1016/j.jmst.2018.09.023 1005-0302/© 2018 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology.
X. Niu et al. / Journal of Materials Science & Technology 35 (2019) 88–93
89
3. Result and discussion 3.1. Residual stress inside SiC fiber
Fig. 1. Schematic diagram of test points on cross-section of the SiC fiber.
So far, there are few researches on SiC fiber reinforced Ni alloy composites. A few reports mainly focus on the fabrication of composites and the study of interfacial reactions, and there is no report on the interface residual stress of this kind of composites. In this paper, SiC fiber reinforced Ni-Cr-Al alloy composites with (Al + Al2 O3 ) diffusion barriers were prepared, and the residual stress intensity and distribution of inner SiC and surface C-rich coating in SiC fibers, precursor wires and composites were measured by laser Raman spectroscopy. The interface morphology and element lining maps of the composites were obtained by scanning electron microscopy, and the influence of the (Al + Al2 O3 ) diffusion barrier on the interfacial residual stress of the composites were analyzed. The composite samples were treated by vacuum heat treatment. The influence of heat treatment process on the interfacial residual stress was studied, and the cause for the change of residual stress was analyzed. 2. Experimental The continuous SiC fibers (Institute of Metal Research, Chinese Academy of Sciences (IMR), China) fabricated by chemical vaper deposition(CVD) were used, which have an outer diameter of ∼100 m and the ∼2.5 m thickness of C-rich coating on surface. Firstly, precursor wires were prepared by depositing (Al + Al2 O3 ) diffusion barrier layer and Ni20Cr5Al matrix layer (∼32 m in thickness) on the surface of the SiC fibers using the dual-target mid-frequency magnetron sputtering device. Then the precursor wires were bonded tightly to form precast slabs that were laminated in hexagonal arrangement in a vacuum hot press mold. SiC fiber-reinforced Ni alloy matrix composites were fabricated by vacuum hot press diffusion. The vacuum hot press parameter is 950 ◦ C/60 MPa/1 h. Raman spectroscopy was performed using a LabRam HR800 Raman spectrometer. The excitation light was He-Ne, the laser wavelength was 632.81 nm, the power was 2.5 W, the spatial resolution was 1–2 m, and the scan range of all Raman spectra was 300 to 1800 cm−1 . The test points were selected from the edge of the W core, along the radial direction of the SiC fiber cross-section every 10 m, until the C-rich coating on the fiber surface, as shown in Fig. 1. Peak separation of the obtained Raman spectra was processed by the analytical software Peakfit, and the wavenumber position corresponding to the respective peaks were determined. The interface morphologies of the composite material were observed by scanning electron microscopy (SEM) (FEG XL30), and the distribution information of the elements at the interface was inspected by an energy dispersive spectrometer (EDS).
Study [9] shows that -SiC is a face-centered cubic structure, and the Si and C atoms in the crystal cell form covalent bonds to form a non-centrosymmetric diamond-like structure. In the Raman spectrum, there are two distinct first-order phonon scattering peaks of the SiC crystal. They are the transverse optical phonon mode (TO) peak at 790 cm−1 and the longitudinal optical phonon mode (LO) peak at 973 cm−1 . The stress on SiC fiber will cause the shift of Raman peaks, and the displacement value of the peak shift has an approximately linear relationship with the stress. According to the displacement of the TO peak of the SiC crystal, the stress value of the SiC crystal can be determined [10]. When compressive stress is present on SiC crystal, the TO peak shift toward to higher wave number, and when the tensile stress is present, the TO peak shifts to the lower wave number. The displacement value of TO peak of the SiC crystal and the axial stress of the crystal satisfy the following linear relationship [11]: ωTO (cm−1 ) = −3.27(GPa)
(1)
where ωTO =ωTO -ω0TO , ω0TO is the TO peak value in the unstressed state, ω0TO = 790 cm−1 . The Raman spectra is separated into peaks to obtain TO peak. According to the TO peak value, the displacement value ω of the TO peak can be calculated, and the stress values at different positions in the SiC fiber were obtained by Formula (1). Fig. 2 shows Raman spectra of SiC fiber, precursor wire and composite obtained from different position along the radial of the SiC fibers. It can be seen that the Raman spectra at different positions inside the SiC fiber have obvious changes, with increasing distance from the W core. According to the displacement value ω of the TO peak in Fig. 2 and Formula (1), the value and distribution of residual stress along the radial of the SiC fiber are calculated, as shown in Fig. 3. From Fig. 3, it can be seen that the SiC fibers are under residual compressive stress, and the internal stress of the fibers is parabolic. Along the radial of the fiber, the SiC fibers are composed of the first deposition layer with large SiC grains, good crystallinity, and few stacking faults, the second deposition layer with small and uniform grains size and a little bit of crystalline silicon, and C-rich deposition layer [12]. At the first test point (10 m from the W core) position, the TO peak is strong and symmetrical, and the stress is greater. The reason is that this point is located in the first deposition layer where the large grains are pressed against each other and the stress in the radial direction of the fiber is less dispersed. At the next tested position (20 m), the TO peak is weaker and broadened. This point is located in the transitional deposition layer, where the SiC grain size becomes smaller and the faults such as stacking faults increase. The radial stress of the fiber here is more dispersed, so the stress decreases. At the third tested position (30 m), the TO peak intensity continues to weaken and the waveform is broadened. This point is located in the second deposition layer, where the SiC grain size is fine and homogeneous and the layer disorder increases [13], so the stress is smaller here. The TO peak at the fourth tested point (40 m) is similar to that at the third tested position. However, the stress increase, because it is affected by compressive stress produced by C-rich coating. As shown in Fig. 3, after the process that the SiC fibers are fabricated to the precursor wires, the compressive stress inside the fibers is reduced, and the stress at the portion near the C-rich coating becomes tensile stress. This is due to the deposition of (Al + Al2 O3 ) diffusion barrier layer and Ni-Cr-Al alloy layer on the surface of the SiC fibers. The higher thermal expansion coefficient of the deposited layers causes tensile stress on the fibers, which weakens the internal compressive stress of the SiC fiber. SiC fibers are
90
X. Niu et al. / Journal of Materials Science & Technology 35 (2019) 88–93
Fig. 3. Value and distribution of residual stress along the radial of the fiber in SiC fiber, precursor wire and composite.
Fig. 4. Morphologies of SiCf /Ni-Cr-Al Composites: (a)without (Al + Al2 O3 ) coating, (b) with (Al + Al2 O3 ) coating.
Fig. 2. Raman spectra of SiC obtained from different positions along the radial of the fiber: (a) SiC fiber, (b) precursor wire and (c) SiCf /Ni-Cr-Al composite.
brittle material, and the excessive compressive stress may cause the fiber to buckle. Therefore, the residual stress inside the fiber should not be too great before fibers prepared into the composite. The alleviating trend of the internal compressive stress of the SiC fibers in the precursor wires is beneficial to the improvement of the interface physical compatibility of the composites.
From Fig. 3, it can also be seen that, after the precursor wires are fabricated to composite by hot pressing diffusion, the stress difference decreases at different positions inside the SiC fibers, which means the compressive stress is further relieved. Fig. 4 shows the cross morphology of the composites with and without (Al + Al2 O3 ) diffusion barrier layer. In the process of composites fabrication without (Al + Al2 O3 ) layer, due to volume expansion caused by production of the interface product at high temperature and great mismatch of the thermal expansion coefficient between Ni-Cr-Al alloy matrix and SiC fibers, large compressive stress will generate inside the SiC fiber, which may cause reaction
X. Niu et al. / Journal of Materials Science & Technology 35 (2019) 88–93
91
Table 1 Value of G peak position and residual stress for C-rich coating.
G wavenumber (cm−1 ) (GPa)
Fig. 5. Raman spectra of C-rich coating on the surface of SiC: (a) SiC fiber, (b) precursor wire and (c) SiCf /Ni-Cr-Al composite.
and stress patterns inside the fibers (Fig. 4(a)). In our work, since the (Al + Al2 O3 ) diffusion barrier layer added to the composite interface, the fibers remain intact and no obvious reaction layer at the interface (Fig. 4(b)). The matrix has little effect on the internal stress of the SiC fibers, and the stress inside the fibers tends to be uniform under the effect of thermal residual stress. 3.2. Residual stress at the interface of SiCf /Ni-Cr-Al composites Fig. 5 shows the Raman spectra of the C-rich coating on the surface of SiC fiber in fibers, precursor wires, and composites. There are two distinct absorption peaks in the spectra of graphite-like structure C coating in the wave number range of 1000–2000 cm−1 . They are the D peak (1330–1360 cm−1 ) and the G peak (1585–1600 cm-l ), which represent different chemical structures. The peak characterizes the E2g vibrational mode of the C atom in the graphite structure is G peak, which is the vibration generated in the opposite direction in the adjacent plane of the carbon network plane. This peak appears to indicate that the C coating has a graphite-like lamellar structure. The D peak characterizes the A1g vibrational mode in the graphite plane, which resulting from the edges disorder or defects of the graphite crystallites and long-range order destruction of the graphite platelets. The intensity of D peak reflects disorder degree of C-rich coating. The bonds of A1g vibrational mode reflected by D peak mostly exist between the surface and the graphite plane, so the D peak is insensitive to the axial residual stress of the fibers [14,15]. The graphite plane in the C-rich coating on the surface of the SiC fibers is highly oriented along the fiber direction, so the wavenumber change of G peak can character the axial residual stress of the fibers. A lot of researches [16,17] show that the displacement of G peak has a linear relationship with the strain in the graphite-like structure, and the relationship between the stress and the G peak displacement is given: =
E ( − 0 ) = −0.1038 J
(2)
where 0 is the G peak position in the stressfree state (0 = 1584.5 cm−1 ), E is the Young’s modulus of the C coating (E = −110 GPa), J is the constant of −10.6 cm−1 /%. According to the displacement value of the G peak and formula (2), the residual stress values at C-rich coating of SiC fiber, precursor wires and composites are calculated, as shown in Table 1. The residual stress at C-rich coating on the surface of the SiC fibers is −1.90 GPa. During the deposition of the C-rich coating on the
SiC
Precursor wire
Composite
1602.8 −1.90
1598.0 −1.40
1596.3 −1.22
SiC surface, due to the growth of the amorphous carbon and the graphite crystal, the internal stress is caused by the volume expansion. At the same time, the thermal residual stress is caused due to the mismatch of thermal expansion coefficient between the surface C-rich coating and the SiC. Residual stress is composed of the internal stress and the thermal residual stress. The residual stress at the C-rich coating of the precursor wires reduces to −1.4 GPa. This is due to the difference of thermal expansion coefficient between the C-rich coating and the (Al + Al2 O3 ) diffusion barrier layer on SiC surface can cause tensile stress, which weakens the compressive stress at C-rich coating. The residual stress at the C-rich coating of the composites reduces to −1.22 GPa, which was decreased by 0.18 GPa. Interface residual stress of composites consists of the residual stress at Crich coating in the precursor wires before hot pressing, the thermal residual stress caused by the thermal expansion coefficients mismatch of the C-rich coating and the matrix layer, and internal stress caused by interface reaction products. The thermal expansion coefficient of (Al + Al2 O3 ) diffusion barrier is between the fiber and the matrix, which relieves the thermal residual tensile stress caused by the matrix on the C-rich coating. On the other hand, the diffusion barrier prevents the interdiffusion of atoms at the interface and inhibits the formation of brittle products which are not conducive to the residual stress release. Fig. 6 shows the interface morphology and EDS elements lining map of the SiCf /Ni-Cr-Al composites. At the interface, Si and C atoms diffuse in the same direction, and Ni, Cr, and Al atoms diffuse in opposite direction. But obviously, Si, Ni, Cr, and Al atoms cannot continue to diffuse cross the (Al + Al2 O3 ) diffusion barrier. There is no obvious brittle interface product at the C-rich coating. As interstitial diffusion atoms, the diffusion rate of C atoms is faster than other atoms, and the diffusion path is farther. C atoms enter the matrix and react with Cr to form reactants. Since the number of C atoms entering the matrix under the effect of barrier layer are limited, the reactants are nanometer level, and the particulate carbides are dispersed in the matrix near the barrier layer, so they have low impact on the C-rich coating. Therefore, the vacuum hot pressing process of the composites fabrication has a little effect on the interface residual stress at C-rich coating. Fig. 7 shows Raman spectra of the C-rich coating on the surface of SiC fiber in composites after vacuum heat treatment at 850 ◦ C for 50, 100, 150, and 200 h. Since the integral intensity ratio value R(ID /IG ) of D peak and G peak is direct proportional to the graphitization degree of the C coating and inverse proportional to the crystal integrity [15]. With the increases of heat treatment time, R value decreases, which indicates that C atoms diffuse gradually, the disorder degree of the C-rich coating increases, and the defects in C-rich coating increase. The shoulder line on the right side of the D peak is called the D¨line. Some scholars [18] have found similar features in the Raman spectroscopy study of amorphous carbon. The reason for the occurrence of this peak is not yet clear, while some scholars [19] believe that it is related to the sp3 -like tetrahedral cross-linking structure of C atoms forming and existence of a few impurity atoms in amorphous carbon. In our study, the reason ¨ for Dpeak appears may be that under high temperature heat treatment for a long time, the diffusion of C atoms cause a large number of defects, stacking faults and impurity atoms in C-rich coating. The G peak wavenumber was obtained by peak separation of the Raman spectra in Fig. 7. According to Formula (2), the displacement value and residual stress for C-rich coating in composites
92
X. Niu et al. / Journal of Materials Science & Technology 35 (2019) 88–93
Fig. 7. Raman spectra of C-rich coating on the surface of SiC fiber in composites. Table 2 Value of G peak position and residual stress for C-rich coating in composites under different heat treatment conditions. 850 ◦ C/50 h 850 ◦ C/100 h 850 ◦ C /150 h 850 ◦ C /200 h −1
G wavenumber (cm (GPa)
Fig. 6. Interface morphology and EDS elements lining map of SiCf /Ni-Cr-Al composites.
) 1601.8 −1.26
1698.6 −1.18
1594.4 −1.02
1592.3 −0.81
under different heat treatment conditions can be calculated, as shown in Table 2. With the increase of the vacuum heat treatment time, the residual stress of the C-rich coating at the interface of the composite material gradually decreases. In this study, the main reason for the decrease of residual stress at the interface was analyzed from the interface behavior of the composites. The interface morphology and EDS elements lining map of SiCf /Ni-Cr-Al composites under different heat treatment conditions are shown in Fig. 8. As the heat treatment time increases, the C-rich coating gradually
Fig. 8. Interface morphologies and EDS elements lining maps of SiCf /Ni-Cr-Al composites under different heat treatment conditions. (a) 850 ◦ C /50 h, (b) 850 ◦ C/100 h, (c) 850 ◦ C/150 h (d) 850 ◦ C/200 h.
X. Niu et al. / Journal of Materials Science & Technology 35 (2019) 88–93
depletes, which make the tensile stress relieved. Under the action of (Al + Al2 O3 ) diffusion barrier layer, the increase ratio of the of reaction layer thickness is very slow, which mainly controlled by diffusion of C atoms. The tensile stress of reaction layer to the Crich coating changes very little and gradually decreases, and it has a tendency to gradually weaken the compressive stress of the C-rich coating. 4. Conclusions (1) During the fabrication of SiCf /Ni-Cr-Al alloy composites, due to the presence of (Al + Al2 O3 ) diffusion barriers, the residual tensile stress in SiC fibers is relieved, and the non-uniform internal residual stress of the fibers tends to be consistent. (2) During the fabrication of SiCf /Ni-Cr-Al alloy composites, the residual compressive stress at the interface of the C-rich coating was gradually relieved from 1.9 GPa to 1.4 GPa; (Al + Al2 O3 ) diffusion barrier improved the physical compatibility of the SiC fiber and the Ni-Cr-Al alloy matrix, inhibited excessive chemical reactions at the SiCf /Ni-Cr-Al interface and avoided high residual stresses at the interface of the composites. (3) Vacuum heat treatment can reduce the residual stress at the interface of composite materials. As the heat treatment time increases, the residual stress of the C-rich coating gradually decreases. Acknowledgment This work was supported by the National Natural Science Foundation of China (No. 51371170).
93
References [1] Z.Q. Wang, Aeron. Manuf. Technol. 22 (2015) 34–38 (in Chinese). [2] S.T. Mileiko, V.M. Kiiko, A.A. Kolchin, A.V. Serebryakov, V.P. Korzhov, M.Y. Starostin, N.S. Sarkissyan, Compos. Sci. Technol. 62 (2002) 167–179. [3] Y.M. Wang, G.X. Zhang, X. Zhang, Q. Yang, L.N. Yang, R. Yang, Acta Metall. Sin. 52 (2016) 1153–1170 (in Chinese). [4] L. Zhang, N.L. Shi, J. Gong, Z.L. Pei, L.J. Gao, C. Sun, Acta Metall. Sin. 47 (2011) 497–501 (in Chinese). [5] T. Adachi, T. Sekino, T. Nakayama, T. Kusunose, K. Niihara, Mater. Lett. 57 (2003) 3057–3062. [6] P. Rangaswamy, M.B. Prime, M. Daymond, M.A.M. Bourke, B. Clausen, H. Choo, N. Jayaraman, Mater. Sci. Eng. A 259 (1999) 209–219. [7] Y. Ward, R.J. Young, R.A. Shatwell, Compos. Part A 33 (2002) 1409–1416. [8] C.A. Taylor, M.F. Wayne, W.K.S. Chiu, Thin Solid Films 429 (2003) 190–200. [9] D. Ghosh, G. Subhash, N. Orlovskaya, Acta Mater. 56 (2008) 5345–5354. [10] G. Pezzotti, H. Suenobu, T. Nishida, O. Sbaizero, J. Am. Ceram. Soc. 82 (1999) 1257–1262. [11] W.L. Zhu, J.L. Zhu, S. Nishino, G. Pezzotti, Appl. Surf. Sci. 252 (2006) 2346–2354. [12] B.H. Jin, N.L. Shi, J. Mater. Sci. Technol. 24 (2008) 261–264. [13] B. Liu, Y.Q. Yang, X. Luo, B. Huang, Spectrosc. Spect. Anal. 31 (2011) 2956–2959 (in Chinese). [14] H. Hiura, T.W. Ebbesen, K. Tanigaki, H. Takahashi, Chem. Phys. Lett. 202 (1993) 509–512. [15] Y. Ward, R.J. Young, R.A. Shatwell, J. Mater. Sci. 36 (2001) 55–66. [16] P. Colomban, G. Gouadec, J. Mathez, J. Tschiember, P. Peres, Compos. Part A 37 (2006) 646–651. [17] R.J. Young, S.J. Eichhorn, Polymer 1 (2007) 2–18. [18] S.K. Sze, N. Siddique, J.J. Sloan, R. Escribano, Atmos. Environ. 35 (2001) 561–568. [19] W.W. Cao, B. Zhu, M. Jing, C.G. Wang, Spectrosc. Spect. Anal. 28 (2008) 2885–2889 (in Chinese).