The next generation of radar electronics will use simultaneous transmit and receive (STAR) technology, which incorporates magnetic ceramics in multiport devices that allow the simultaneous sending and receiving of electromagnetic (EM) signals. The utility of these devices can be seen in radar systems that enable continuous tracking of multiple targets.

A key component to radar electronics is the circulator. The basis of operation for a circulator is that an EM signal introduced to one port is wholly coupled to an adjacent port while isolating it from the other port(s). Circulators with three or more ports have nonreciprocal properties that are highly valued in many microwave systems. Magnetic ceramics, such as ferrites, that are biased in a DC magnetic field exhibit anisotropic wave propagation characteristics. This anisotropy in the propagation of EM waves gives rise to a non-reciprocal rotation of the plane of polarization of the EM wave, a non-reciprocal phase shift and a non-reciprocal displacement of the microwave field pattern.¹ Ferrites-magnetic oxides that have both desirable dielectric and magnetic properties and, most importantly, are insulators-have been an integral part of nearly all circulator device designs that operate at µ- and mm-wave frequencies. Although spinel ferrites and garnets have been used in many applications from 1-10 GHz, the hexaferrites have proven to be versatile for applications from 1-100 GHz. One advantage of the hexaferrite materials is that they might not require a permanent magnet when used for microwave applications.

Traditionally, circulator designs have included ferrites as pressed compacts with biasing magnets placed above and/or below to provide the necessary biasing magnetic field for operation. These designs result in three-dimensional constructs that are often bulky and costly to fabricate. Special circuitry and packaging features of the permanent magnet need to be considered in all microwave devices. The next generation of magnetic microwave devices will be planar, self-biased and low-loss, and will operate well beyond the performance metrics of today's devices. Self-biasing is an important property that eliminates the need for the permanent magnet and reduces the size and cost of microwave devices. The elimination of this magnet is an essential step in making microwave devices planar and compatible with other technologies, such as semiconductor integrated devices.

To achieve these goals, ferrites must possess high saturation magnetization (4πM_s), high remanent magnetization (M_r), tunable (or variable) magnetic anisotropy fields (H_A), low microwave losses (i.e., low FMR linewidths, ΔH_FMR) and, for many applications, have the easy axis of magnetization perpendicular to the film plane (i.e., perpendicular magnetic anisotropy, PMA). In physical terms, the films should be thick (>400 µm), dense (low levels of porosity that are responsible for added microwave loss) and pure-phase. For many applications the microstructure should possess a strong crystallographic orientation, although true epitaxy is not required.

Barium Hexaferrite

Enlarge this picture Figure 2. a) A SEM image of the surface morphology of a screen printed film after binder burnout (low-temperature heat treatment) and b) after sintering, and c) the cross section of the same film illustrating elongated grains oriented with their long axis perpendicular to the film plane.

Ba (M-type) hexaferrite (BaM) has the magnetoplumbite structure and a stoichiometry of BaFe₁₂O₁₉.² The magnetoplumbite structure has 32 atoms per formula unit and 64 atoms in a single unit cell (see Figure 1). One property of this compound that is of particular value in microwave device design is the strong uniaxial anisotropy field with the magnetic easy direction being along the c-axis (H_A~17,000 Oe).^3,4 The high magnetic anisotropy field can be adjusted by using an appropriate substitute for the Fe ion, which allows for the tuning of resonance frequencies from 1-100 GHz.⁵ This degree of freedom makes BaM a choice material for many monolithic microwave integrated circuit (MMIC) applications.

The structure consists of seven ferric ions distributed onto five inequivalent interstitial sites (one trigonal bipyramidal, two octahedral and two tetrahedral). The magnetism arises from the super-exchange interaction between ferric cations mediated by the oxygen anion. This is a negative exchange interaction (i.e., antiferromagnetism), with an amplitude from each Fe-O-Fe pairing dependent on the angle made between the ferric cations and oxygen (i.e., a larger exchange occurs with larger angles). The saturation magnetization is reported to be 4000 G (as 4πM_s) with a Neel temperature of 450°C and an FMR linewidth for a single crystal ranging between 10-20 Oe.^4,6 This compound is stable in air.

BaM bulk polycrystalline compacts, produced through typical ceramic processing techniques, possess relatively low to moderate hysteresis loop squareness and FMR linewidths of 2000-3000 Oe.⁴ However, Dionne and Fitzgerald have shown that in indium (In)-doped BaM the use of a magnetic field during compaction provided hysteresis loop squareness values of >0.90.⁷ With such high loop squareness values, these compacts show promise as self-biased materials and therefore can operate without permanent magnets in circulator applications. (Note: With the substitution of In for Fe, the magnetic anisotropy field reduces dramatically, lowering the zero-field resonance frequency and redefining the device operational frequency and bandwidth.) Research activity from 1990 to the present has focused on the development of BaM as thin and thick films. Processing of BaM films have been performed using pulsed laser deposition (PLD),⁸ liquid phase epitaxy (LPE)⁹ and, more recently, screen printing.¹⁰

Recently, Y. Chen et al. have demonstrated that screen printing (SP) is capable of processing thick, self-biased, low-loss BaM films.¹⁰ In this technique, pure-phase Ba-hexaferrite powders are mixed with a binder to form a suitable paste for screen printing. The printing is done through a template or mask, and the still wet film is baked at 250°C to burnout the binder. During this low-temperature heat treatment, a DC magnetic field is applied perpendicular to the plane of the film. A second heat treatment at ~1200°C is required for recrystallization and sintering. Figure 2 shows SEM images of the screen printed films at different stages of processing. Prior to sintering (Figure 2a), roughly micron size particles are loosely arranged in a high-porosity film (>50%). After sintering (Figure 2b), the film has a closely packed polycrystalline structure in which the grains range in size from small(~1 µm) to large (8-10 µm) with porosity levels of approximately 13-15%. In Figure 2c, the structure (in a cross section at a 45-degree orientation) of the film is revealed to show elongated grains with the long axis perpendicular to the film plane. Some pores are clearly visible, suggesting that further refinement is possible. X-ray diffraction analysis of this sample displayed (0,0,2n) reflections with enhanced intensity consistent with the c-axis texture perpendicular to the film plane. Collectively, these data confirm that screen printing was successful in the processing of dense, thick, pure-phase films with a strong c-axis orientation out of the film plane.

Magnetic Properties

Enlarge this picture Figure 3. Typical hysteresis loops for screen printed films illustrating high loop squareness for the easy axis loop perpendicular to the film plane. The saturation magnetization, as 4πMs, is 4000 G.

Figure 3 shows hysteresis loops (magnetization as 4πM_s vs. applied magnetic field) that display the characteristic uniaxial perpendicular anisotropy necessary for circulator operation. Compared to PLD and LPE films (see Table 1), these films have a higher coercivity, as well as high hysteresis loop squareness (~0.95). (Hysteresis loop squareness is defined as the remanent magnetization divided by the saturation magnetization, M_r/M_s.) A high hysteresis loop squareness indicates that the film has a magnetization even when the applied magnetic field is removed. This result demonstrates for the first time that thick BaM films can be made self-biased so that the remanent magnetization is about 95% of the saturation magnetization.

Enlarge this picture

Another important property is the microwave loss, which is largely proportional to ferromagnetic resonance (FMR) linewidth (DH). FMR measurements of these films reveal DH values as low as 210 Oe at f=55.6 GHz. Although these values are comparatively large compared to the high-quality films processed by PLD and LPE, i.e., ΔH<30Oe¹⁰, they are markedly improved compared with polycrystalline compacts whose linewidth is between 2000-3000 Oe.¹¹ For most microwave applications at high frequencies, ΔH≤500 Oe is tolerable. For example, for ΔH~800 Oe an insertion loss of <0.5 dB was estimated in a circulator design at 35 GHz utilizing BaM films. When these results are viewed collectively, the development of planar self-biased circulators looks encouraging.

Outlook

Screen printing is often used in integrated circuit fabrication for depositing interconnects, dielectric components, metal ground planes, etc. However, comparatively little work has been done in applying this processing technique to the deposition of magnetic materials. The study described in this article demonstrates that screen printing is capable of producing thick films of good microwave quality (FMR linewidths of ~210 Oe). The loop squareness values are in excess of 0.95, making these the first demonstrated thick films with self-biased properties. However, much work is still needed to improve film density and further reduce microwave loss.

One remaining challenge is to overcome the "self-bias paradox," which calls for the growth of films with high crystal quality (which translates to low microwave loss) while sustaining a high remanence magnetization. Growing a film of high crystal quality promotes a low coercivity, which, along with the demagnetizing energy of a thin film, shears the easy axes loop and results in the loss of self-biased properties. This paradox thus reduces the need to increase coercivity and maintain low microwave loss. Many mechanisms that result in enhanced coercivity-such as grain boundaries, inhomogeneities, local anisotropy fields, pores, etc.-increase microwave loss to unacceptable levels. Screen printing has shown that this is indeed possible. The fact that screen printing is suitable for large substrates at a high throughput and low cost makes this technique attractive for industrial processing.

Authors' Note

This research was supported by grants from the Office of Naval Research (N00014-05-10349) and the Defense Advanced Research Program Agency (HR0011-05-1-0011).