Engineering soft magnetic properties by doping ions in low-fired M-type hexaferrite with Bi–Co–Ti substitution

Abstract

M-type hexaferrite with appropriate Co, Ti and Bi co-doping has excellent soft magnetic properties in a very high frequency range and can be sintered at low temperatures. This paper systematically investigates the effect of these ion substitutions in different lattice sites for Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ (x = 0.05–0.3 & y = 1.0–1.2). Bi substitution promotes the sintering densification and lowers the sintering temperature by an approximate relation of T_s = 1150 − 1500x + 2500x², where a denser microstructure also works for higher permeability. Co substitution plays a key role in the magnetic properties through magnetocrystalline anisotropy. For all samples, the permeability always reaches a maximum when the Co amount (x + y) is around 1.2 because it has an appropriate planar magnetocrystalline anisotropy at such composition. This work can provide a clear guide for the design of magnetic materials at a very high frequency range.

---

Introduction

For years, electronic technology has developed rapidly towards high frequency and small size, so there is a great demand for various multilayer chip inductive devices and components for several hundred MHz or higher frequencies, i.e. very high frequencies (VHF, 30–300 MHz) and ultrahigh frequencies (UHF, 300–3000 MHz).^1,2 Current multilayer chip inductive components always use spinel ferrites, such as NiCuZn ferrites, whose magnetic properties are excellent below 100 MHz but decay seriously above it, following Snoek’s Law. Instead, hexaferrites with planar magnetocrystalline anisotropy, such as Y-type (Ba₂Me₂Fe₁₂O₂₂, Me = bivalent transition metal ion), Z-type (Ba₃Me₂Fe₂₄O₄₁) and Co–Ti doped M-type (BaFe_12−2x(CoTi)_xO₁₉, x = 1.0–1.4) hexaferrites, have a high cut-off frequency up to GHz and a high permeability at several hundred MHz.^3,4 Among them, Y-type hexagonal ferrites, especially Co₂Y (Ba₂Co₂Fe₁₂O₂₂), have strong planar anisotropy, which endows them with a very high cut-off frequency up to 4 GHz, but the permeability is as low as 2–3.^5–8 Z-type hexagonal ferrites, such as Co₂Z (Ba₃Co₂Fe₂₄O₄₁), have a high permeability of ∼15 below 1 GHz, but the high sintering temperature above 1300 °C and the unstable phase composition restrict their applications.^9–12 Co–Ti doped M-type hexaferrites have even higher permeability than Co₂Z due to high saturation magnetization, so they are desired in many applications.^13–15

M-type hexaferrite has a highly symmetric hexagonal structure stacked by R blocks (BaFe₆O₁₁) and S blocks (Me₂Fe₄O₈) according to a certain order of RSR*S*, where the asterisk (*) stands for 180° rotational symmetry around the hexagonal axis.¹⁶ The perspective drawing of R and S blocks is shown in Fig. 1, where the S block has a cubic spinel structure with the [111] axis vertical and the R block is formed by three oxygen layers with hexagonal packing. In the R and S blocks, O²⁻ and Ba²⁺ ions form a hexagonal closely packed lattice, while Fe³⁺ and bivalent metal ions (Me²⁺) occupy the interstitial positions. The Fe³⁺ ions are allocated in five different crystallographic sites, including three octahedral sites (4f₂, 12k and 2a), one tetrahedral site (4f₁) and one trigonal bipyramidal site (2b), where the 12k, 2a and 2b sites have spin-up configurations and the other two have spin-down configurations.¹⁷ The distribution of the Fe³⁺ ions in the M-type hexaferrite lattice leads to a strong uniaxial anisotropy along the c-axis, which determines the typical hard magnetic features with a high coercive force H_c. Co²⁺ ions with strong planar anisotropy preferentially occupy the 4f₂ and 2b sites.^18–20 Hence, if proper Co substitution is used, the uniaxial magnetocrystalline anisotropy of an M-type hexaferrite can be changed to planar anisotropy and the hard magnetic features are changed to soft magnetic features with high permeability.^13–15

Fig. 1 The perspective drawing of R and S blocks in M-type hexaferrite.

The authors’ previous work reported the properties of a Bi–Co–Ti substituted M-type hexaferrite (Ba_1−xBi_x(Co_1.1+xTi_1.1)Fe_9.8−xO₁₉).^21,22 The Bi–Co–Ti substitution remarkably lowers the high sintering temperature of M-type hexaferrite to below 1000 °C, so it meets the need of low temperature co-fired ceramics (LTCC) technology and can be used to fabricate multilayer chip devices. Moreover, the low heat-treatment temperature controls the valence variation of transition metal ions, which enhances the magnetic properties, such as a higher permeability than that of a Co–Ti substituted M-type hexaferrite. Although good properties were achieved, how to efficiently engineer the magnetic properties has not been discussed in detail. Hence, this paper systematically explored the effect of various substitution ions on the soft magnetic properties of M-type barium hexaferrite. This work can efficiently guide the design of high performance magnetic materials for very high frequency applications.

Experimental

Preparation of hexaferrites

Bi–Co–Ti substituted M-type hexaferrite with a chemical formula of Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ (x = 0.05–0.3 & y = 1.0–1.2) was prepared using a solid-state reaction method. The starting materials, reagent grade Fe₂O₃, BaCO₃, Bi₂O₃, Co₃O₄ and TiO₂, were mixed for 4 hours using a planetary ball-mill. The mixture was calcined at 900 °C for 4 hours, and then ground again for 4 hours. The resultant powders were dry-pressed in a stainless steel die. The pressed pellet and toroidal samples were sintered in the temperature range from 900 °C to 1100 °C for 6 hours in air.

Characterization

The phase composition of the calcined powders was characterized using X-ray diffraction (XRD) using Cu K_α radiation (λ = 0.15418 nm). The density of the sintered samples was measured using the Archimedes’ method. The microstructure of the fractured surface of the sintered samples was observed using scanning electron microscopy (SEM). The complex permeability was measured using an Agilent E4991A RF impedance/materials analyzer from 1 MHz to 1 GHz using the 16454A test fixture.

Results and discussion

Phase formation

Fig. 2 shows the XRD spectra of the Bi–Co–Ti substituted M-type hexaferrites (x = 0.05, 0.25 & y = 1.0, 1.1, 1.2) calcined at 900 °C. After careful indexing with a standard XRD spectrum (PDF#43-002), the samples calcined at 900 °C with Bi–Co–Ti substitution were confirmed to have M-type hexaferrite structures with P63/mmc. Co–Ti substitution does not affect the phase formation temperature of M-type hexaferrite, but a trace amount of Bi substitution (x = 0.05) can lower it obviously. In contrast, the sample calcined at 900 °C without Bi substitution (x = 0 and y = 1.1) has a distinct other phase, which completely disappears after calcination at 1050 °C. The Bi–Co–Ti substituted samples have a much lower phase formation temperature (∼150 °C) than that of the Co–Ti substituted sample without Bi doping.

Fig. 2 XRD spectra of the Bi–Co–Ti substituted M-type hexaferrites (x = 0.05, 0.25 & y = 1.0, 1.1, 1.2) calcined at 900 °C, compared with those of the Co–Ti substituted M-type hexaferrites calcined at 900 °C and 1050 °C, where the red spot indicates the strongest peak of the other phase.

Density and microstructure

Fig. 3 shows the density of the samples sintered at different temperatures. The density is sensitive to the Bi amount (x) and the sintering temperature, but not to the Co–Ti substitution (y). The samples with the same Co–Ti amount have a similar density after sintering at the same temperature, but a different Bi amount changes the density obviously. This is because the addition of Bi helps the formation of low-melting compounds to promote the liquid-phase mass transfer during the high temperature sintering process. An appropriate Bi amount (x > 0.1) can lower the sintering temperature to 950 °C to meet the need of LTCC, which is much lower than that of Co–Ti doped M-type hexaferrite without Bi doping (∼1200 °C). For example, the x = 0.15 samples sintered at 950 °C have a high average density of 4.8 g cm⁻³, about 90% of the theoretical density (∼5.3 g cm⁻³). In general, for Bi–Co–Ti substituted M-type hexaferrite, the sintering temperature varies with the Bi amount monotonically but nonlinearly, which follows an approximate relation of T_s (°C) = 1150 − 1500x + 2500x² (0 < x < 0.3), as shown in Fig. 4.

Fig. 3 Density of the Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ samples sintered at different temperatures.

Fig. 4 The plot of the sintering densification temperature for Bi–Co–Ti substituted M-type hexaferrite with the Bi amount and the approximate second-order polynomial fitting.

Fig. 5 shows the microstructure of the sample sintered at 950 °C, with x = 0.05, 0.15, 0.25 & y = 1.0, 1.1, 1.2. Similar to the density results, the microstructure is mainly determined by the Bi amount and is not sensitive to the Co–Ti substitution. The samples sintered at the same temperature and with the same Bi amount have a similar microstructure. With the rise of the Bi amount, the grains grow larger and stack more compactly. After sintering at 950 °C, the samples with x > 0.10 had a dense microstructure and uniform grain size of ∼2 μm, which agrees with the density results. Although the Co–Ti substituted M-type hexaferrites without Bi (x = 0 & y = 1.1) also exhibit high density after sintering at 1200 °C, there are some abnormally large grains among small grains in the microstructure.¹⁵ It implies that the low-fired samples with Bi doping have better microstructures with uniform grain sizes than the high-fired samples without Bi doping. In addition, the grain boundaries are clear and no glass phase is observed. The homogeneous microstructure without a nonmagnetic glass grain boundary can endow the low-fired samples with better magnetic properties.

Fig. 5 SEM photos of the microstructure of the Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ samples sintered at 950 °C. (a) x = 0.05 y = 1.0, (b) x = 0.05 y = 1.1, (c) x = 0.05 y = 1.2, (d) x = 0.15 y = 1.0, (e) x = 0.15 y = 1.1, (f) x = 0.15 y = 1.2, (g) x = 0.25 y = 1.0, (h) x = 0.25 y = 1.1, (i) x = 0.25 y = 1.2.

The low temperature sintering of Bi–Co–Ti substituted M-type hexaferrites facilitates their application in multilayer chip devices and modules, where the criteria of LTCC are a necessary condition.²⁶ The samples with x > 0.1 can be sintered well at 950 °C, which meets the need of LTCC technology. In addition, the low-fired Bi–Co–Ti substituted M-type hexaferrites have a desirable microstructure with homogeneous and compactly stacked grains and without a nonmagnetic glass phase at the boundary, which works for better soft magnetic properties.

Magnetic properties

Fig. 6 shows the frequency dependence of the complex permeability of the samples sintered at 1000 °C. The permeability, either the real or imaginary part, is sensitive to the frequency. For most samples, the real part of the permeability, μ′, shows a steady value up to several hundred MHz and then drops down, which is accompanied by a peak of the imaginary part, μ′′. The plateaus in the frequency spectra are most important for applications, where μ′ has a steady and high value and μ′′ is very low, indicating the efficient operating frequency range of soft magnetic materials. The peak of μ′′ always enhances with the rise of the μ′ value, following the Kramers–Kronig relation, and its position implies the cut-off frequency.

Fig. 6 Frequency dependence of the complex permeability of the Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ (x = 0.05–0.3, y = 1.0–1.2) samples sintered at 1000 °C.

The value of μ′ varies with both compositional variation and the sintering temperature. We extracted the μ′ data at a fixed frequency of 100 MHz within the plateaus from all frequency spectra and plotted them in Fig. 7, showing the composition dependence of the permeability for the samples sintered at different temperatures. For the samples sintered at the same temperature, with increasing x, μ′ always rises first, reaches a maximum and then drops.

Fig. 7 Permeability of the Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ samples (y = 1.0, 1.1 and 1.2) sintered at different temperatures (T_s = 950–1050 °C) as a function of Bi amount.

The permeability is determined by both the chemical composition and the microstructure. It is noted that although the μ′ maximum shifts to a higher x value for the sample with a higher y value, it always occurs at x + y = 1.2–1.25, i.e. the total Co amount. It implies that Co substitution plays a key role in the permeability of Bi–Co–Ti doped M-type hexaferrites. Pure M-type hexaferrite has hard magnetic character with uniaxial anisotropy, which endow it with a high magnetic energy product but with very low permeability at a high frequency. The Co²⁺ ion has strong planar anisotropy and its single-ion anisotropy dominates the magnetocrystalline anisotropy of M-type hexaferrites. When Co substitution is large enough (y > 0.9), the direction of the ease of magnetization changes from the c axis to the basal plane, i.e. from uniaxial magnetocrystalline anisotropy to planar anisotropy, so that the hexaferrite changes from a hard magnetic to a soft magnetic material, which induces a high permeability. If the Co amount rises further, the increasing planar magnetocrystalline anisotropy will reduce the permeability by restricting the rotation of the spin. The permeability of a soft magnet with planar anisotropy can be expressed as

where M_s is the saturation magnetization, and H_θ and H_φ stand for the anisotropy fields vertical and parallel to the basal plane. Hence, the permeability of Bi–Co–Ti doped M-type hexaferrite reaches a maximum when it has a proper planar magnetocrystalline anisotropy.

Besides chemical composition, the microstructure also affects the permeability because pores and grain boundaries block spin rotation serving as an equivalent magnetocrystalline anisotropic field. If sintered at the same temperature, the sample with a lower Bi amount (x value) has a relatively loose microstructure, i.e. more pores and grain boundaries, which works against the permeability. The effect of pores and grain boundaries on the permeability can be evaluated using a magnetic circuit model²³

where μ and μ_B stand for the effective permeability of the real sample and of ideal bulk ferrite without any defects, respectively, D stands for the average size of ferrite particles, and δ stands for the average gap between them, i.e. the thickness of the grain boundaries. The δ/D value is calculated approximately from the measured density based on the following formula,²⁴where ρ and ρ_t stand for the measured and theoretical density. The sample with a loose microstructure has a small grain size and a large gap, i.e. a large δ/D value, so the effective permeability will be lower. Fig. 8 shows the measured permeability as a function of the δ/D value, where the curve shows the permeability calculated from eqn (2), using μ_B = 80. In general, the measured μ′ is in good agreement with the calculated curve, especially in the high δ/D part. In the low δ/D part, all samples have a dense microstructure so the materials’ intrinsic characters become the dominating factor. It should be noted that the sample with a Co amount of x + y = 1.2–1.25 always has an effective permeability higher than the calculated curve, while the sample with a higher or lower Co amount has an effective permeability much lower than the calculated curve, which agrees with the previous discussion.

Fig. 8 Plot of the measured μ′ as a function of the d/D value. The curve is calculated from eqn (2), using μ_B = 80.

The modification of Co substitution on anisotropy is also reflected in the cut-off frequency f_r, which is another key parameter for soft magnetic materials and can be expressed as

Fig. 9 plots the composition dependence of f_r for the samples (y = 1.0 and 1.1) sintered at 1000 °C.²⁵ For the samples sintered at 1000 °C, the f_r has a minimum at the composition of x + y = 1.10–1.15, and then increases with the rise of the Co amount.

Fig. 9 Composition dependence of f_r for the Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ samples (y = 1.0 and 1.1) sintered at 1000 °C.

It is noted that the composition for the f_r minimum is a bit lower than that for the μ′ maximum. This is because the maximum of μ′ is determined chiefly by H_φ while the minimum of f_r is determined by the product effect of H_φ and H_θ which stand for the energy for the spin rotating in or out of the basal plane. The planar magnetocrystalline anisotropy of hexaferrite prevents spins from rotating out of the basal plane, which is mainly reflected in the large value of H_θ. H_φ is similar to the anisotropic field of cubic spinel ferrites, while H_θ is about two orders of magnitude larger. That is the reason why the f_r value of hexaferrites is about one order of magnitude higher than that of spinel ferrites. The substitution of Co²⁺ on Fe³⁺ changes the uniaxial anisotropy to planar anisotropy and is notable around the composition of x + y = 1.10–1.15 so that a minimum f_r occurs. More Co substitution further increases the planar anisotropic field, both H_θ and H_φ, so that f_r rises remarkably. In addition, the y = 1.1 samples always have higher f_r than the y = 1.0 samples, although they can be of the same Co amount because of more nonmagnetic Ti ions.

Snoek’s product, i.e. (μ′ − 1)f_r, is the most important limitation for the application of ferrites at high frequencies, which reflects the comprehensive performance of the permeability and the cut-off frequency. The Snoek’s product of hexaferrite with planar anisotropy has a different expression as follows:

A hexaferrite with a strong planar anisotropy can have a large Snoek’s product, which makes it suitable to be used in VHF or higher frequency. Fig. 10 shows the Snoek’s product of Bi–Co–Ti substituted M-type hexaferrite (y = 1.0 and 1.1) sintered at different temperatures,²⁵ calculated using the data from Fig. 6, 7 and 9. In general, Snoek’s product rises with increasing x due to the variation of planar anisotropy. Co²⁺ substitution enhances the planar anisotropy so Snoek’s product rises with the increment of H_θ/H_φ according to eqn (5). The y = 1.0 samples always have a low Snoek’s product of about 5–8 GHz, which is similar to those of spinel ferrites. In contrast, the y = 1.1 samples have a much higher Snoek’s product of 12–18 GHz. For example, the x = 0.2 & y = 1.1 sample sintered 1050 °C has a high Snoek’s product of 17.69 GHz (μ′ = 30 & f_r = 0.61 GHz). The large Snoek’s product greatly promotes applications above hundreds of MHz.

Fig. 10 Snoek’s product of the Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ samples (y = 1.0 and 1.1) sintered at different temperatures.

Conclusions

This paper systematically investigated the effects of Bi, Co and Ti ion substitutions in different lattice sites of M-type hexaferrites with compositions of Ba_1−xBi_x(Co_x+yTi_y)Fe_12−x−2yO₁₉ (x = 0.05–0.3 & y = 1.0–1.2). Bi substitution promotes sintering densification, so that the samples can be sintered well at a low sintering temperature to fit LTCC technology. Co substitution plays a key role in the magnetic properties through magnetocrystalline anisotropy. The permeability of Bi–Co–Ti substituted M-type hexaferrite always reaches a maximum when the Co amount (x + y) is around 1.2 due to proper planar magnetocrystalline anisotropy. The y = 1.1 samples have both high permeability and a high cut-off frequency, which induces a very high Snoek’s product up to ∼18, so they are excellent soft magnetic materials suitable for applications above hundreds of MHz. This work can provide a clear guide for magnetic material design in the VHF and UHF range, and has brilliant prospects in the rapid development of microelectronic and communication technologies towards higher frequencies.

Acknowledgements

This work was supported by grants from the National Science Foundation of China (51172020 and 51372018), the National Program for Support of Top-Notch Young Professionals, the Program for New Century Excellent Talents in University (NCET-12-0780), the Beijing Higher Education Young Elite Teacher Project (YETP0414), and the Program for Changjiang Scholars and Innovative Research Team in University (IRT1207).

Notes and references

1. Y. Bai, W. J. Zhang, L. J. Qiao and J. Zhou, J. Adv. Ceram., 2012, 1, 100.
2. S. Kracunovská and J. Töpfer, J. Electroceram., 2009, 22, 227.
3. J. Smit and H. B. J. Wijn, Ferrites, Cleaver-Hume Press, London, 1959.
4. S. Kracunovská and J. Töpfer, Key Eng. Mater., 2010, 434, 361.
5. Y. Bai, J. Zhou, Z. Gui and L. Li, Mater. Lett., 2002, 57, 807.
6. Y. Bai, J. Zhou, Z. Gui, Z. Yue and L. Li, J. Magn. Magn. Mater., 2003, 264, 44.
7. Y. Bai, J. Zhou, Z. Gui and L. Li, J. Magn. Magn. Mater., 2002, 246, 140.
8. Y. Bai, J. Zhou, Z. Gui, L. Li and L. Qiao, J. Alloys Compd., 2008, 450, 412.
9. H. G. Zhang, L. T. Li, J. Zhou, Z. X. Yue, Z. W. Ma and Z. L. Gui, J. Eur. Ceram. Soc., 2001, 21, 149.
10. Y. Bai, F. Xu, L. Qiao and J. Zhou, Mater. Res. Bull., 2009, 44, 898.
11. S. Kracunovská and J. Töpfer, J. Mater. Sci.: Mater. Electron., 2011, 22, 473.
12. H. I. Hsiang, L. T. Mei, C. S. Hsi, W. C. Wu, L. B. Cheng and F. S. Yen, J. Magn. Magn. Mater., 2011, 323, 1011.
13. C. S. Wang, L. T. Li, J. Zhou, X. W. Qi, Z. X. Yue and X. H. Wang, J. Magn. Magn. Mater., 2003, 257, 100.
14. S. Bierlich and J. Töpfer, IEEE Trans. Magn., 2012, 48, 1556.
15. W. Zhang, Y. Bai, X. Han, L. Wang, X. Lu and L. Qiao, J. Alloys Compd., 2013, 546, 234.
16. E. P. Wohlfarth, Ferromagnetic materials: a Handbook on the properties of magnetically ordered substances, North-Holland Publishing Company, Amsterdam, New York, Oxford, 1982.
17. X. Batlle, X. Obradors, J. Rodriguez-Carvajal, M. Pernet, M. V. Cabanas and M. Vallet, J. Appl. Phys., 1991, 70, 1614.
18. Z. Mimsa, S. Lego, R. Gerber and E. Pollert, J. Magn. Magn. Mater., 1995, 140–144, 2103.
19. J. M. Williman, J. Adetunji and M. Gregori, J. Magn. Magn. Mater., 2000, 220, 124.
20. Y. P. Wang, L. C. Li, H. Liu, H. Z. Qiu and F. Xu, Mater. Lett., 2008, 62, 2060.
21. W. Zhang, Y. Bai, X. Han, L. Wang, X. Lu, L. Qiao, J. Cao and D. Guo, Mater. Res. Bull., 2013, 48, 3850.
22. W. Zhang, Y. Bai, X. Han, L. Wang, X. Lu, L. Qiao, J. Cao and D. Guo, J. Alloys Compd., 2013, 556, 20.
23. T. Tsutaka, M. Ueshima, T. Tokunaga, T. Nakamura and K. Hatakeyama, J. Appl. Phys., 1995, 78, 3983.
24. T. Nakamura, J. Appl. Phys., 2000, 88, 348.
25. The cut-off frequencies of most y = 1.2 samples are out of the upper limit (1 GHz) of impedance analyzer, so that the data are not included in Fig. 9 and 10.
26. L. B. Kong, Z. W. Li, G. Q. Lin and Y. B. Gan, Acta Mater., 2007, 55, 6561.

---