On high-temperature oxidation and protection of 2:17-type SmCo-based magnets

1 Introduction

Rare earth permanent magnetic materials have been used in electric appliances, microwave tubes, gyroscopes and accelerometers, reaction and momentum wheels to control and stabilize satellites, frictionless bearings and co-axial starter-motors/generators for aero-engines, sensors as well as actuators to perform a basic function – to provide a magnetic flux (Fidler, Schrefl, Hoefinger, & Hajduga, 2004; Goll & Kronmüller, 2000; Gutfleisch, 2009; Gutfleisch et al., 2006; Jiang & An, 2013; Liu, 2006; Ray & Liu, 1992). Among these permanent magnetic materials, the Sm₂(Co,Cu,Fe,Zr)₁₇ alloys are the best high-temperature magnets because of their high Curie temperature, T_c≈850°C (Ray & Liu, 1992).

The 2:17-type Sm(Co_wFe_vCu_xZr_y)_z magnets, w=0.64–0.81, v=0.01–0.28, x=0.05–0.13, y=0.02–0.04, and effective z=7.5–7.9 (Chen, Walmer, Walmer, Liu, & Kuhl, 2000), have a typical cellular precipitation structure, which is composed of a 2:17 rhombohedral (2:17R) cell and 1:5 hexagonal (1:5H) cell boundary (Fidler et al., 2004; Goll & Kronmüller, 2000; Gutfleisch, 2009; Gutfleisch et al., 2011; Jiang & An, 2013; Liu, 2006; Ray & Liu, 1992; Walmer, Chen, & Walmer, 2000; Xiong et al., 2004). Such a cellular precipitation structure is governed by the elastic energy due to lattice misfit between the 2:17R and 1:5H (Gutfleisch et al., 2006), and it develops by thermal homogenization treatment of the sintered alloys at temperatures between 1100°C and 1200°C and subsequent isothermal aging at ∼800°C and slow cooling to 400°C (Gutfleisch et al., 2006; Livingston & Martin, 1977; Rabenberg, Mishra, & Thomas, 1982). Figure 1 shows the cell and boundary precipitates of a Sm(Co_0.68Fe_0.22Cu_0.08Zr_0.02)_7.5 magnet along a direction perpendicular to the c-axis. The rhombic 2:17R cell is surrounded by 1:5H boundary with a width in the nano-size range. The 2:17R cell is a Co- and Fe-rich phase, whereas the 1:5H boundary is a Cu- and Sm-rich phase. The two phases are crossed with a Zr-rich lamellar phase generally with a width of several nanometers, which can be viewed by transmission electron microscopy (TEM) in a direction parallel to the c-axis of the magnet (Gutfleisch et al., 2006; Livingston & Martin, 1977; Rabenberg et al., 1982; Walmer et al., 2000; Wang et al., 2015; Xiong et al., 2004).

Figure 1:

Scanning TEM (STEM) bright-field (BF) image and electron diffraction patterns (upper) as well as corresponding elemental X-ray mappings (lower) of a Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 magnet along a direction perpendicular to the c-axis.

Coercivity, which represents the ability of the permanent magnets to resist demagnetization with increasing temperature, originates from the domain wall energy difference between the 2:17R cell phase and the 1:5H cell boundary phase. The room temperature coercivity of the 2:17-type magnets value is ∼30 kOe (Gutfleisch, 2009; Gutfleisch et al., 2006; Handstein et al., 2003; Ojima, Tomizawa, Yoneyama, & Hori, 1977; Walmer et al., 2000). However, its irreversible and unacceptable loss occurs when the environmental temperature exceeds 550°C (Walmer et al., 2000). The coercivity loss should be intrinsically related to the degradation of 2:17R cell and 1:5H boundary of the magnets. This has been considered a result of metallurgical changes (Chen et al., 2000), Sm diffusion (Walmer et al., 2000), and evaporation (Chen, Walmer, Kottcamp, & Gong, 2001; Pauw, Lemarchand, & Malandain, 1997). Besides physical reasons (e.g., microstructural defects), chemical variation within the magnets by surface oxidation has been recently proposed to be a decisive (if not an exclusive) factor leading to the collapse of their cellular precipitations structure. Oxidation of the 2:17-type magnets has been investigated by some authors (Chen et al., 2000, 2001; Kardelky et al., 2004; Kardelky, Gebert, Gutfleisch, Hoffmann, & Schultz, 2005; Liu & Walmer, 2005; Liu, Marinescu, Vora, Wu, & Harmer, 2009; Mao et al., 2014; Pauw et al., 1997; Pragnell, Williams, & Evans, 2008; Pragnell, Evans, & Williams, 2009; Wang, Zheng, An, Zhang, & Jiang, 2013; Yang et al., 2012, 2013a,b). The unacceptable coercivity loss may be prevented by adding appropriate surface coatings that can effectively offer oxidation protection to the magnets. Some surface coatings for the 2:17 magnets have been reported by Chen et al. (2006), Chen, Walmer, and Liu (2004), Dong, Peng, Guo, Li, and Wang (2013), Pragnell, Evans, and Williams (2012), Yang et al. (2013a), Zhao et al. (2013a,b, 2014).

In the article, the mechanism for the high-temperature oxidation of the 2:17-type magnets is explained, and the research progress on understanding their structural degradation by oxidation, and correspondingly, their magnetic properties loss in the absence and presence of surface coatings, is reviewed.

2 Oxidation and its effect on magnetic properties

2.1 External oxidation

The 2:17-type SmCo-based magnets are very susceptible to oxidation at temperatures over 550°C, forming a double-layered surface zone as shown in Figure 2. The outer layer (layer 1) is a Sm-free external oxide scale (EOS). It was characterized as an Fe-Co mixed oxide earlier by Chen et al. (2000). Pragnell et al. (2008) indicated that the EOS formed at 600°C consists of CuO, Co₃O₄, and CoFe₂O₄. Most recently, the EOS has been characterized and identified mainly as (Co_xFe_1-x)₃O₄, which is covered with a thin layer of CuO and Cu₂O, as seen in Figure 3 (Wang et al., 2015). Cu₂O is initially formed and later on can be further oxidized into CuO (Wang et al., 2015). The copper oxide layer gradually stops growing with the underlying (Co_xFe_1-x)₃O₄ thickening. Some investigators (Chen et al., 2001; Pauw et al., 1997) attributed the formation of the Sm-free oxide scale to the evaporation of Sm. In fact, the formation of a Sm-containing oxide scale is thermodynamically and kinetically impossible for some reasons, as summarized as follows. First, the oxides in the EOS are metal-deficient type, and their growth is dominated by the outward diffusion of cations of Cu, Co, and Fe. Thus, the EOS has the characteristics of outward growth (Pragnell et al., 2008; Yang et al., 2012). Second, the extreme low diffusivity of Sm in the magnets, according to the Wagner theory (1959), limits them to form a Sm-oxide-rich scale. Third, Sm has a strong affinity for oxygen. It can be locally oxidized into internal Sm₂O₃ particles, as will be detailed later.

Figure 2:

Scanning electron microscopy (SEM) view of the cross section of the Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 magnet for 20-h oxidation in air at 600°C, showing that the magnet formed an EOS (layer 1) and an IOZ (layer 2), in which different areas occur, as indicated by “A”, “B”, and “C”, respectively.

Figure 3:

STEM BF image (upper) and corresponding elemental X-ray mappings (lower) of the EOS on the Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 oxidized for 1 h in air at 600°C.

The electron patterns for CuO and Cu₂O were obtained on a nano-grain of the corresponding oxide in the top area surrounded by the dotted lines using fast Fourier transformation (FFT) and nano-beam electron diffraction (NBD), respectively.

Although significantly thickened with an increase of temperature from 500°C to 700°C (Yang et al., 2012), the EOS generally has a similar phase composition and grows following a mechanism that can be described as follows. At the onset of oxidation, oxides of the various component elements of the magnets nucleate. Among the various oxide nuclei, Co₃O₄ grows fastest and quickly forms a layer covering the magnets (Wang et al., 2015). The Co₃O₄ layer is thickened by the dominant outward diffusion of Co cations. Simultaneously, the oxide incorporates Fe and Cu cations. The low solubility of Cu ion in the spinel oxide (which is the reason for the observation of the Cu precipitates in the spinel oxide layer in Figure 3) leads to a slow growth of the outermost Cu₂O. In contrast, Fe ion has an unlimited solubility in the spinel oxide. It incorporates into the spinel oxide with increased concentration with time, due to a gradual decrease of oxygen partial pressure (P_O2) at the interface between the oxide and the metal. Consequently, Co₃O₄ converts into (Co_xFe_1-x)₃O₄.

From Figure 3, voids are visible at the scale/metal interface at the early stage of oxidation. They result from the condensation of the vacancies of the cations of Co, Fe, and Cu injected from the growing EOS. This is the reason why “breakaway” oxidation occurs for the magnet at 600°C, following a mechanical mode of wrinkling-buckling-cracking (Yang et al., 2012). The breakaway oxidation causes the formation of local Fe₂O₃ within the spinel oxide layer. Heavier breakaway oxidation at 700°C causes more Fe₂O₃ to be formed in the oxide scale.

2.2 Internal oxidation

As shown in Figure 2, below the EOS appears a much thicker area (layer 2) with a contrast clearly different from the interior of the magnet. This area, called the reacting layer by Chen et al. (2000), is actually an internal oxidation zone (IOZ) (Pragnell et al., 2008). It formed at and below 600°C and can be divided into three typical areas, as indicated in Figure 2. Area A represents the dispersed micrometer-sized particles of Sm oxide, which are intrinsically formed in the original magnet produced by a conventional powder metallurgy technique (Chen et al., 1998; Feng, Chen, Guo, Yu, & Li, 2010; Liu, Hoffman, & Brown, 1997; Livingston & Martin, 1977; Ma, Liang, & Bounds, 1997; Rabenberg et al., 1982). Area B represents a strip rich in Co and Fe but poor in Sm (Pragnell et al., 2008; Yang et al., 2012). The strip is not clearly seen under SEM in the earlier stage, e.g., the first 1 h at 600°C (Wang et al., 2015), of oxidation, but it appears to be increased in size and number with oxidation time. The strip is mainly a body-centered cubic (bcc) Fe-Co, as seen from its selective area electron diffraction pattern (SAEDP) and elemental distribution presented in Figure 4 (Yang, 2013). It is usually co-precipitated with Cu from the decomposed 2:17R and 1:5H because the solubility of Cu is lower in bcc Co-Fe than in face-centered cubic (fcc) Co-Fe (Bein, Colinet, & Durand-Charre, 2000; Palumbo, Curiotto, & Battezzati, 2006). The latter is the other metallic matrix phase of the IOZ. The major area C has a composition approximate to the original magnet matrix except that the oxygen concentration is higher. It occurs actually as a result of Sm oxidation.

Figure 4:

STEM BF image and corresponding elemental X-ray mappings of a strip in the IOZ indicated by “B” in Figure 2.

The SAEDP is acquired from the “M”-framed area, which shows the strip is bcc structured.

The internal oxidation product of Sm is too fine in size to be observed from Figure 2 under SEM resolution. It was characterized as Sm₂O₃ using X-ray diffraction (Pragnell et al., 2008). Figure 5 shows a high-resolution TEM image of a shallow area of the formed IOZ (Wang et al., 2015). It displays the lattice structure of two Sm₂O₃ particles, which are only ∼10 nm in size. The particles evolve in shape and size with depth, from randomly dispersed, fine spherical particles in the shallow IOZ to rod-like particles in deeper areas along the direction normal to the penetration of oxygen (Wang et al., 2015). Although the length of the rod of the Sm₂O₃ particles also increases with depth, it is still in the nanometer range. The Sm₂O₃ particles evolve in morphology due to a depth-related change of the mechanism of the internal oxidation of Sm from nucleation dominant to growth dominant (Böhm & Kahlweit, 1964). The nucleation-to-growth transition here can be interpreted as follows. Due to the high affinity of Sm for the penetrated O, Sm₂O₃ easily nucleates in the magnet below EOS. The precipitation of numerous Sm₂O₃ particles significantly decreases P_O2 at the oxidation front. This causes difficulty in the nucleation of fresh Sm₂O₃. In this case, the growth of the formed Sm₂O₃ particles along the direction of oxygen penetration becomes a dominant process.

Figure 5:

High-resolution TEM (HRTEM) image of an area similar to that indicated by “C” in Figure 2 but formed during shorter-term (1 h) oxidation, showing that the Sm of the magnet was internally oxidized into nano-sized particles of Sm₂O₃ and oxidation decomposed the 2:17R and 1:5H phases into fcc and bcc Co-Fe, in which dislocations (as indicated by “A”) and micro-twins (indicated by “B”) were ubiquitously formed.

From Figure 5, Sm oxidation completely decomposes the 2:17R phase and 1:5H phase into fcc Co-Fe (γ phase) and bcc Fe-Co (α phase). According to the Co-Fe phase diagram, it is possible that the γ phase is originally from the decomposed 1:5H, while the α phase is from the decomposed 2:17R, because the Co content is higher at the 1:5H than at the 2:17R-decomposed location. Accordingly, the internal oxidation of the magnet in the early stage can be written as

(1) Sm 2 (Co, Fe, Cu, Zr) 17 + [O] → Sm 2 O 3 + Fe - Co( α ) + Cu + Zr  (1)

(2) Sm(Co, Fe, Cu, Zr) 5 + [O] → Sm 2 O 3 + Co-Fe( γ ) + Cu + Zr  (2)

where Cu and Zr exist as metallic precipitates. The α-γ transformation may occur in the later stage due to increased Fe consumption with time by external oxidation.

The internal oxidation at higher temperatures, e.g., 700°C (Yang, Peng, Guo, Li, & Wang, 2013b), occurs differently in several aspects as described below. First, FeSmO₃ occurs in a shallow IOZ as presented in Figure 6. Oxide formation by the 2:17R oxidation can be written as

Figure 6:

STEM BF image and corresponding elemental X-ray mappings of a shallower IOZ of the Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 oxidized for 20 h in air at 700°C, showing the formation of submicrometer-sized particles of FeSmO₃ as identified by the electron diffraction pattern and the overlap of the X-ray mappings of Fe, Sm, and O.

(3) Sm 2 (Co, Fe, Cu, Zr) 17 + [O] → FeSmO 3 + Co - Fe ( γ ) + Cu + Zr  (3)

where Co-Fe is in the γ form due to increased consumption of Fe by accelerated external oxidation and formation of FeSmO₃. Second, the Co-based strips do not occur, but the formed oxide particles are larger and visible in the IOZ. Third, a Cu-rich band occurs between the IOZ and the unoxidized part of the magnet. Thus, the internal oxidation at 700°C changes the relative concentrations of Co, Fe, and Cu in the IOZ metallic matrix. This significantly affects both the alloying thermodynamics of these elements in the IOZ and the interdiffusion between the IOZ and the oxidation-free magnet.

The internal oxidation of Sm causes volume expansion in the IOZ. The volume expansion can be roughly calculated to be ∼9.6% if the constituent elements of the magnet are simplified to merely Sm and Co (Wang et al., 2015) and assuming 2:17R occupies 85% volume of the 2:17-type magnets. Such a volume expansion generates compressive strains in the IOZ. Driven by the strains, the metals in the IOZ are extruded, forming a thin Co-based metallic band between the EOS and the IOZ (Wang et al., 2015). They probably diffuse through the dislocation channels in the IOZ as proposed elsewhere (Guruswamy, Park, Hirth, & Rapp, 1986; Kodentsov, Van Dal, Cserháti, Daróczi, & Van Loo, 1999; Zhao, Peng, Wang, & Wang, 2007). From Figure 5, the lattice of the Co-Fe solid solution is also highly distorted, with the ubiquitous formation of dislocations and micro-twins, as indicated by “A” and “B”, respectively. The lattice mismatch is also seen at the marked Sm₂O₃/α-Fe-Co interface in the directions of [431]_Sm2O2//[111]_α-Fe-Co.

2.3 Oxidation effect on magnetic properties

Table 1 presents the variation of the magnetic properties of the Sm(Co_0.68Fe_0.22Cu_0.08Zr_0.02)_7.5 magnet after 20-h oxidation in air at 500°C, 600°C, and 700°C. The values of intrinsic coercivity (_iH_c), remanence (4πM_r), and maximum energy product [(BH)_max] are available from the demagnetization curves of the magnet, which was measured using columnar samples of Ø3×3 mm in a vibrating sample magnetometer with a maximum field of 30 kOe (Wang et al., 2015). Clearly, _iH_c together with 4πM_r and (BH)_max considerably decrease with the temperature increase. At a given temperature, the magnetic properties significantly vary with oxidation time. For example, _iH_c decreases from an original 27 to 26.3 kOe after 1-h oxidation, 21.4 kOe after 10 h, 16.5 kOe after 20 h, and 13 kOe after 50 h at 600°C, as shown in Table 2. After 50-h exposure to air at 600°C, the room temperature _iH_c is reduced to half of the original value.

Table 1

The coercivity, remanence, and maximum energy product of the Sm(Co_0.68Fe_0.22Cu_0.08Zr_0.02)_7.5 magnet without and with oxidation in air for 20 h at 500°C, 600°C, and 700°C.

	iHc (kOe)	4πMr (kG)	BHmax (MGOe)
As-prepared	27.16	10.46	25.12
500°C	25.66	9.04	17.98
600°C	16.56	7.06	9.21
700°C	14.46	5.61	5.62

Table 2

The coercivity, remanence, and maximum energy product of the Sm(Co_0.68Fe_0.22Cu_0.08Zr_0.02)_7.5 magnet after removing its portion oxidized at 600°C for various periods of time.

Oxidation time (h)	iHc (kOe)	4πMr (kG)	BHmax (MGOe)
As-prepared	27.16	10.46	25.12
1	26.37	9.89	21.26
5	25.28	9.70	19.42
10	21.15	7.90	12.46
20	16.56	7.06	9.21
50	12.93	5.75	5.60

The magnetic loss in the oxidized magnet is undoubtedly associated with the oxidation-induced decomposition of the 2:17R and 1:5H phases as addressed above. However, the oxidation-free area (layer 3 in Figure 2) of the oxidized magnet has no significant _iH_c loss. Figure 7 shows almost full overlap of the demagnetization curves of the original magnet and the magnet after removing the oxidized areas during the 20- and 50-h exposure to air at 600°C. From Figure 7 and Table 2, two conclusions can be drawn. First, it is very likely that oxidation is a dominant (if not exclusive) factor leading to the unacceptable magnetic loss of the 2:17-type magnets at high temperatures. Second, the 2:17R cell and 1:5H boundary phases are thermally stable inside the oxidation-free region of the magnet. The two phases are really observed in the magnet area adjacent to the IOZ as shown in Figure 8 (Wang et al., 2015). It is worth noting that the two phases somewhat vary in microchemistry. For example, the 1:5H cell boundary, with respect to the original one, has lower Cu but higher Fe (Wang et al., 2015). The microchemistry variation of the two phases is credibly from the increase of concentrations of Co and Fe in the metallic phases the IOZ, which causes the interdiffusion between IOZ and unoxidized part of the magnet. The interdiffusion should be one reason for the insignificant _iH_c loss in the magnet unaffected directly by oxidation.

Figure 7:

The demagnetization curves of the oxidation-free area of the Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 magnet for various periods of time in air at 600°C.

Figure 8:

TEM BF image and corresponding SAEDPs of the oxidation-free Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 magnet below the IOZ formed for 1-h oxidation in air at 600°C (viewed along a direction vertical to the c-axis), indicative of the maintaining of the 2:17 R and 1:5 H.

The parallel lines below the IOZ are Zr-rich lamellas.

3 Oxidation protection

Since high susceptibility to oxidation is the key reason for the unacceptable loss of the magnetic properties of the 2:17-type magnets, the increase in their oxidation resistance has recently become an important research topic. Liu and Jiang (2011, 2012) proposed that alloying Si can increase the oxidation resistance of a SmCo₇ alloy. The oxidation rate of the SmCo_7-xSi_x at 500°C is significantly decreased, while its coercivity is increased by increasing the Si content from 0.3 to 0.9. Similar results also appear by simultaneous alloying the 1:7 SmCo-based alloy with Si and Zr (Feng et al., 2014). However, they have no evidence showing that Si alloying can guarantee the magnet to grow a SiO₂ scale, which can theoretically fully prevent the external oxidation of Co, Fe, and Cu and highly suppress the internal oxidation of Sm. Works on the alloying of Si and other elements into the 2:17-type magnets are lacking. It is possible that the alloying addition to self-improve the oxidation performance is not a feasible method to noticeably mitigate the magnetic loss of the 2:17-type magnets by oxidation, because, for example, addition of small amounts of Si cannot guarantee the magnet to grow a protective SiO₂ scale, whereas higher amounts will have adverse effect on the structure and composition of the cellular precipitates and, accordingly, on the magnetic properties of the magnets.

Surface coatings should be a preferred choice to increase the resistance of the 2:17-type magnets against oxidation. The surface coatings were designed and developed as a diffusion barrier to oxygen by many investigators (Chen et al., 2004, 2006; Dong et al., 2013; Pragnell et al., 2012; Yang et al., 2013a; Zhao et al., 2013a,b, 2014). Chen et al. (2004, 2006) found that a pure Ni film with a best thickness of 15 μm can offer a good protection against oxidation in air at 500°C to a Sm₂Co₁₇-type magnet, preventing it from a considerable magnetic property loss. This result is well consistent with the work of Wang et al. (2013), who indicated that the deposition of a ∼20-μm-thick Ni coating sharply decreased the mass gain of Sm(Co_balFe_0.1Cu_0.1Zr_0.033)_6.93 magnet exposed for 500 h at 500°C, leading to a loss of the maximum energy product (BH)_max from a normal 40.8% for the uncoated magnet down to only 4.0% in the presence of coating. Pragnell et al. (2012) reported that the oxidation of a Sm(Co_0.74Fe_0.1Cu_0.12Zr_0.04)_8.5 in air at 550°C was significantly suppressed by a 1.5- to 5-μm-thick Pt coating and a paint-like 5- to 10-μm-thick overlay coating containing Ti and Mg oxides, but their processing, composition, and microstructure are not presented in detail.

An ideal coating used for application to high-temperature magnets not only must have exceptional resistance to oxidation, but must also have no adverse effect on the magnetic property. In other words, the coating itself should have the advantages of compact structure, good thermodynamic and chemical stability, and acceptable physical and mechanical compatibility with the magnets. In addition, the coating must be as thin as possible because the loss of magnetic properties of the coating/magnet system is proportional to the volume not only consumed by oxidation but also occupied by the coating. On these bases, two types of thin film with a thickness of below 4 μm were developed by Peng and colleagues: one is oxide ceramics such as Cr₂O₃ and Al₂O₃ (Yang et al., 2013a; Zhao et al., 2013a,b, 2014) and the other is pure Cr film with an ability to thermally grow an oxide scale of Cr₂O₃ at high temperatures (Dong et al., 2013). The Cr₂O₃ and Al₂O₃ films and the Cr film are oxidation resistant even at temperatures over 550°C.

Figure 9 shows the cross sections of the Cr₂O₃-coated Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 with oxidation in air at 600°C for 1 and 20 h. The Cr₂O₃ film was only ∼1.2-μm-thick and deposited by a reactive arc ion plating (Zhao et al., 2013a). Its deposition process has no significant effect on the chemical composition and precipitation structure of the magnet. Thin film deposition leads to the formation of a very shallow oxidation-affected zone (OAZ) on the magnet, which is not clearly seen for 1 h [only ∼0.33 μm thick based on TEM investigation (Zhao, 2015) but ∼1.5 μm thick for 20 h under SEM resolution]. The OAZ is double-layered, consisting of a thin outer Sm-free Co-rich metallic band and a thicker inner IOZ where Sm has been internally oxidized (Zhao et al., 2013a,b). Figure 10 shows the distribution of elements in the cross-sectioned OAZ image at high magnification for the Cr₂O₃-coated magnet oxidized in air for 1 h. The Co-rich outer band and the inner IOZ with dispersion of Sm-rich precipitates are clearly seen. The Sm-rich precipitates are actually nano-sized Sm₂O₃ particles formed by the internal oxidation of Sm (Zhao, 2015). The latter decomposed the 2:17R and 1:5H into the α- and γ-Co-Fe by reactions (1) and (2). The Sm oxidation extruded the unoxidized elements, forming the outer Co-rich band in the OAZ. A similar Co-rich thin layer occurs between EOS and IOZ of the uncoated magnet, as described and explained above. The Co-rich band, in comparison to the IOZ, is much thinner, i.e., it was ∼0.25 μm thick, whereas the IOZ was ∼1.25 μm for the coated magnet oxidized for 20 h at 600°C (Zhao et al., 2013b). The OAZ increased in thickness from ∼1.5 μm in the condition to ∼3.2 μm for the 20-h oxidation at 700°C (Zhao et al., 2013a). It is nearly two orders of magnitudes thinner than the OAZ in the uncoated magnet with oxidation in the same condition (Yang et al., 2012).

Figure 9:

SEM view of the cross section of the Cr₂O₃-coated Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 magnet for (A) 1 h and (B) 20 h in air at 600°C, with the insets correspondingly showing the magnified image.

Figure 10:

STEM BF image and corresponding elemental X-ray mappings of the cross section of the Cr₂O₃-coated Sm(Co_balFe_0.22 Cu_0.08Zr_0.02)_7.5 oxidized in air for 1 h at 600°C, showing that the OAZ of the magnet is divided into an outer layer rich in Co but free of Sm and an inner layer where Sm has been precipitated.

The original cellular precipitation structure of the magnet, which has collapsed in the OAZ, remains below it.

Compared to the thin Cr₂O₃ film, the Al₂O₃ film of a similar thickness and was deposited from an α-Al₂O₃ target material by magnetron sputtering offered slightly better protection to the Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 in the same exposure condition, leading to the formation of a thinner double-layered OAZ (Zhao et al., 2014). The two types of the oxide films experienced no cracking or spalling during isothermal oxidation at 600°C and 700°C and the subsequent cooling. The thin thickness should be one reason for the good mechanical compatibility of the two films with the magnet. In addition, the interface bonding between the films and the magnet should be increased because the displacement reaction of Cr₂O₃ and Al₂O₃ occurs with the Sm of the magnet. The Cr and Al reduced from the reaction diffuse into the IOZ (Yang et al., 2013a; Zhao et al., 2013a,b, 2014). The occurrence of the displacement reaction can also be confirmed from Figure 10, where Sm was obviously doped in the Cr₂O₃ film while Cr was diffused into the OAZ in the magnet. From Figure 10, the original cellular precipitation structure of the magnet remains in the oxidation-free area below the OAZ magnet. Similar results can also be seen from the oxidation-free area of the magnet coated with the Al₂O₃ thin film (Zhao et al., 2014).

Because of high melting point and exceptional ability to thermally grow a protective layer of Cr₂O₃, pure Cr is an alternative coating material for application to the magnet. Such a Cr thin film can be deposited using a double-glow sputtering method, and it behaves similar to the Cr₂O₃ film in inhibiting magnet oxidation (Dong et al., 2013). The Cr-film coated Sm(Co_balFe_0.22Cu_0.08Zr_0.02)_7.5 forms a ∼2.9-μm-thick OAZ for 20 air exposures at 700°C. One difference is that the Co-rich band is not seen between the film and the formed IOZ because the compressive strains generated by Sm internal oxidation can be released by the interdiffusion between Cr and the component elements (mainly Co) of the magnet. The interdiffusion causes the Cr of the film to penetrate into the deeper area of the magnet; it is, however, helpful to increase the interface bonding between the film and the magnet.

Figure 11 shows the comparison of the demagnetization curves of the uncoated and Cr₂O₃-coated magnets (with the dimensions of Ø3×3 mm) without oxidation and with 20-h oxidation in air at 600°C. Unlike the dramatic drop of _iH_c for the uncoated magnet, almost no _iH_c loss is seen for the coated magnet after oxidation. The overlay of the curves of the coated magnet with oxidation and the uncoated magnet without oxidation suggests that the oxidized magnet also has no significant loss in remanence and maximum energy product. Hence, the degradation of the magnetic properties, although unacceptable for the uncoated magnet, is negligible for the coated magnet at high temperature. This is consistent with the observation that the portion of the volume occupied by the thin Cr₂O₃ film and the OAZ is insignificant (only ∼0.03%) (Zhao, 2015) in the columnar sample, in which the oxidation-free area maintains the typical cellular precipitation structure of the magnet.

Figure 11:

The demagnetization curves of the uncoated and Cr₂O₃-coated Sm(Co_balFe_0.22Cu_0.08Zr_0.02)₇ without and with 20-h oxidation in air at 600°C.

4 Summary

Oxidation of the 2:17-type SmCo-based magnets in air at temperatures over 550°C forms a Sm-free EOS and an IOZ where Sm is internally oxidized into Sm₂O₃ particles. The EOS is mainly composed of (Co_xFe_1-x)₃O₄ spinel, in which the Fe content is increased with time. The outward growth of the spinel oxide rather than Sm sublimation is the reason for non-detection of Sm in the external scale. The thickness of the spinel oxide is significantly increased with temperature. At a given temperature, EOS grows dramatically slower than IOZ. The Sm internal oxidation decomposes the two main magnetic matrix phases of 2:17R and 1:5H into the α- and γ-Co-Fe phases with the precipitation of metallic Cu and Zr. The sizes of the Sm₂O₃ particles formed are temperature-dependent, which are generally in the nano-range at and below 600°C and in the submicrometer range at 700°C. Sm oxidation generates compressive strain on the IOZ. Driven by the strain, metals in the IOZ are extruded to form a thin Co-rich band below the EOS.

Because the 2:17R and 1:5H cellular precipitation structure remains, the unoxidized part of the magnet exhibits only a small loss of coercivity. It suggests that oxidation is the dominant reason for the unacceptable magnetic loss of magnets. One reason for the small _iH_c loss in the oxidation-free area is plausibly correlated with the formation of an interdiffusion zone below IOZ, where the 2:17R and 1:5H phases suffer a variation in microchemistry.

Oxygen diffusion barrier materials can be coated onto the 2:17-type magnets to increase their resistance against oxidation at high temperatures. Oxide ceramics of chromia and alumina or metals with the ability to thermally grow a protective scale of an oxide such as chromia are ideal oxygen diffusion barrier materials. Despite being coated with these materials with a thickness of several micrometers, the magnets are not significantly affected by oxidation, and in the case, _iH_c loss is prevented. The oxide ceramic films and magnets have a large difference in thermal expansion coefficient and mechanical properties. Accordingly, the cracking and spalling resistance of these films should be fully investigated in the future. Metallic films are generally more adhesive to the magnets than oxide films, through diffusion bonding with the magnets. However, the interdiffusion between the films and the magnets inevitably more or less changes the chemistry of the 2:17R and 1:5H phases below OAZ. Understanding the interdiffusion effect on the degree of the chemistry variation of the cellular precipitates is also important for the practical application of the metallic films.