Background

[0001] In electronics products, point of load (POL) converters have been widely used to power integrated circuits (ICs). The close proximity of POLs to ICs is important for performance and efficiency. For example, a battery in a smartphone provides a direct current (DC) voltage of about 4 volts (V), while a smartphone central processing unit (CPU) requires direct current supplied at 1 volt. Therefore, a POL converter is necessary to step down the voltage, and it is positioned near the CPU to eliminate long wiring. Long wiring is undesirable as it tends to increase electromagnetic interference issues, contribute undesirable stray inductance and capacitance, and complicate the layout of the circuit board. Uses of POL converters include voltage regulation (VR) in providing power to a processer. [0002] Miniaturization is a continuing demand in electronics, especially in computing devices such as laptops, smartphones, and tablets. It leads to lighter and smaller products with more functionality and larger battery, thus a compact POL with high energy density is highly desired. Components in a POL converter include a power management IC chip, power inductor, and capacitor. Among those, the inductor is typically most bulky and becomes an impediment to miniaturization. In general, two strategies are available to reduce the inductor footprint. One is to increase inductor working frequency (i.e., the switching frequency of semiconductor devices in power management IC chips). The performance of an inductor in a circuit depends on its impedance, which is proportional to the product of working frequency and inductance. For a certain required impedance, the higher the frequency, the lower the needed inductance, thus, a smaller inductor can be used. A second approach to reduce the inductor footprint is to embed an inductor into a printed circuit board (PCB), and thereby reduce the footprint on the surface of the board.

[0003] Minimizing inductor footprint is typically not the only benefit from embedding the inductor and increasing switching frequency. It may also result in reduced capacitor footprint by reducing the need for decoupling capacitance. Moreover, higher switching frequency tends to decrease energy consumption, when, for example, GaN or SiC transistors are used. The energy saving is achieved through better dynamic voltage and frequency scaling, which means the supply voltage will change more dynamically according to the processor workload.

[0004] There are two requirements for increasing the working frequency of an inductor. First is the availability of high frequency semiconductor switching devices at the desired power level. Second, magnetic materials suitable for use as high frequency inductors. In recent years, the emergence of high speed and high power SiC and GaN semiconductor devices satisfies the first condition for increasing working frequency. The second condition on high frequency magnetic material, however, has yet to be met. [0005] Power ferrites are an important category of soft magnetic materials, (e.g. nickel zinc ferrites) and are widely adopted in the MHz frequency range. In integration with electronic devices, however, there are issues with their use, such as sensitivity to stress, relatively low saturation magnetic induction, frangibility, and property deterioration under relatively high bias field or relatively high induction swing. [0006] Amorphous or nanocrystalline ribbons may also be used, but they tend to generate too much loss (i.e., heat) as the frequency is increased into the MHz range. This is due to the impracticability of very thin ribbons (they typically exceed about 18 micrometers in thickness) coupled with their low resistivity (typically < 500 microQ-cm). both of which promote high eddy current loss, although studies (see, e.g., F. Fiorillo et. al., “Magnetic properties of soft ferrites and amorphous ribbons up to radiofrequencies,” J. Magn. Mater., Vol. 322, 2010, pp. 1497-1504; and M. Yagi et. al., “Very low loss ultrathin Co-based amorphous ribbon cores,” J. Appl. Phys., Vol. 64, 1988, pp. 6050-6052) have demonstrated moderate core loss reduction with thinner ribbons. The thinning processes (e.g., melt-spinning in vacuum, chemical etching, and cold rolling) are expensive and difficult to be implemented in mass production. [0007] Another important type of candidate for high frequency application is magnetic metal powders, especially flake shaped powders. Even 0.5 micrometer thin metal flakes tend to generate too much loss at MHz range due to eddy currents and their low ferromagnetic resonance frequency, especially when operated above 5 MHz. (see, e.g. D. Hou et. al., “Very high frequency IVR for small portable electronics with high-current multiphase 3-D integrated magnetics”, IEEE Transactions on Power Electronics, Vol. 32, No. 11, Nov. 2017, pp. 8705-8717)

[0008] Magnetic thin films made by physical vapor deposition (PVD) or electrochemical deposition have been demonstrated with attractive magnetic properties up to GHz frequency range. Due to stress during growth, however, it is very difficult to achieve a thickness of 10s or 100s of micrometers, as required in practice. Another challenge exists in magnetic thin films. During a DC-DC converter operation, there is a DC magnetic bias field acting on the magnetic core, so a slow saturation under bias field in the core material is preferred. NiFe alloy-based magnetic thin films often have fast saturation due to high permeability. It is typically necessary to introduce additional anisotropy into the films to balance permeability and saturation speed. Growing or annealing the films under a magnetic field, or adding other elements into the films can slow down the saturation. If the permeability in the film plane becomes anisotropic, however, the inductor design will become more difficult and complicated.

Summary

[0009] In one aspect, the present disclosure describes a high frequency (e.g., 5 MHz to 150 MHz) power inductor material having first and second opposed major surfaces, including: a polymer binder; and a plurality of multilayered flakes dispersed in the polymeric binder, the multilayered flakes comprising at least two layer pairs, wherein each layer pair comprises a layer of ferromagnetic material and a dielectric electrical isolation layer so that the ferromagnetic layers are electrically isolated from each other by dielectric layers. The multilayered flakes have an average lateral size less than 400 micrometers, less than 300 micrometers, or less than 200 micrometers. The average lateral size refers to a flake lateral size at 50% of a total cumulative distribution by volume. The multilayered flakes are substantially aligned parallel to the first and second major surfaces such that they do not provide an electrically continuous path over a range of, for example, greater than 0.5 millimeters.

[0010] In one aspect, the present disclosure describes A method of making a high frequency power inductor material, the method including: providing a plurality of multilayered flakes, the multilayered flakes comprising at least two layer pairs, wherein each layer pair comprises a layer of ferromagnetic material and a dielectric electrical isolation layer so that the ferromagnetic layers are electrically isolated from each other by dielectric layers; surface-treating the multilayered flakes with a phosphoric acid solution; and dispersing the multilayered flakes after the surface-treating in a polymeric binder. The multilayered flakes are substantially aligned parallel to the first and second major surfaces such that they do not provide an electrically continuous path over a range of greater than 0.5 millimeters.

[0011] For the purposes of this disclosure, a dielectric is a material wherein the lowest conduction band is at an energy level at least seven times knT higher than the Fermi level, where kn is Boltzmann’s constant (i.e., 1.38 x 10’ ²³ m ² kg/(s ² K)) and where T is the maximum intended use temperature for the power inductor material. The population of the conduction band is determined by the Fermi Function, , and under the stipulated condition no more than 10’ ³ of the electrons in the valence band will be promoted into the conduction band. Further, E is the energy level for the lowest conduction band, and EF is the Fermi level. The quantity (E-EF) is known as the “band gap”. Most dielectrics have band gaps on the order of eV. Charles Kittel, Introduction to Solid State Physics. 6th Ed., New York, John Wiley, 1986, p. 185, shows that semiconductor dielectric materials may have band gaps as low as that of InSb at 0. 17 eV, or about 6.5 times knT at room temperature. For example, SiO as a dielectric has a band gap of about 2 eV (see, for example, Hairen Tan et al., “Wide Bandgap p-type Nanocrystalline Silicon Oxide as Window Layer for High Performance Thin-film Silicon Multi- Junction Solar Cells,” Solar Energy Materials and Solar Cells, Vol. 132, pp. 597-605, January 2015). In addition to SiO, other suitable materials as dielectrics include MgF2, Si, AI2O3, and SiO2.

[0012] Exemplary high frequency power inductor materials described herein are useful, for example, as power inductors in Point of Load (POL) converters, low profile inductors for inductive -capacitive (LC) filters (e.g., for global system mobile communication (GSM) pulse noise suppression in cellular phone speakers), or other applications wherein compact, inductive elements are required on a circuit board. [0013] Advantages of embodiments of high frequency power inductor materials described herein include the capability to achieve a thickness of up to 100s of micrometers, with low core loss density (e.g., less than 10,000 kW/m ³ at 20 MHz and maximum magnetic induction of 10 mT, and less than 20,000 kW/m ³ at 20 MHz and maximum magnetic induction of 15 ml), high saturation magnetic induction (e.g., greater than 0.25 T), high relative permeability (e.g., greater than 20), and soft saturation (e.g., saturation field higher than 1600 A/m) in the MHz range.

[0014] These attributes can enable DC-DC converters working at higher frequencies and can facilitate more efficient circuit board real estate use via a smaller inductor footprint, where inductors may even be embedded as a layer within the circuit board itself. When the inductors are embedded in the board, stray reactance associated with discrete components on the board can be avoided. This reduces the need for decoupling capacitors, thus further decreasing the consumption of board real estate. Another advantage is embedding the inductors into the circuit board reduces the amount of electrical noise and electromagnetic interference (EMI) generated by the POL power converters. Enabling higher working frequency also helps improve battery life through fine dynamic voltage and frequency scaling.

[0015] The high frequency power inductor materials described herein provide superior properties compared to other composite materials such as the multiplayer flake composites described in WO2019/082013. For example, during operation, a power inductor not only has an AC current flowing through it but also has a superimposed DC current. The DC current creates a DC magnetic bias field. At high processor workload, the DC current is high, so is the DC bias field. Composite materials may have an increasing core loss when the DC bias field is increasing, making it unfavorable in high current load applications. The high frequency power inductor materials described herein exhibit a significantly less increased core loss compared to those composite materials. In addition, the multilayered flakes described herein have a relatively smaller lateral size, which makes it more practical to coat using an automated coater. Furthermore, high frequency power inductor materials described herein can have more favorable repeatability in terms of low core loss, which renders further remedy processes (e.g., a second hot press step) unnecessary.

[0016] The high frequency power inductor materials including surface-treated flakes described herein also exhibit a loss tangent at 20 MHz that converges to a narrower band, for example, from 0.055 to 0.064, as compared to samples without the surface treatment. The samples with or without surface treatment may have an initial permeability in substantially the same range.

Brief Description of the Drawings

[0017] FIG. 1 is schematic of an exemplary high frequency power inductor material described herein.

[0018] FIG. 2 is a scanning electron microscope (SEM) image of flakes of Example 4.

[0019] FIG. 3 is plots of exemplary flake lateral size distribution by volume and cumulative distribution by volume versus flake lateral size.

[0020] FIG. 4 is plots of Core loss versus Magnetic DC bias field at 20 MHz for Examples 1-5.

[0021] FIG. 5 is plots of Core loss versus Magnetic DC bias field at 20 MHz for Examples 1-5. [0022] FIG. 6 is plots of initial permeability versus loss tangent at 20 MHz for Examples 6 and 7.

Detailed Description

[0023] Magnetic core material used for the power inductor has become an impediment to the desired increase in switching frequency for POL power supplies. The commercially available magnetic core materials used in smartphone or computer servers may not be suitable for higher switching frequencies. In addition, ferrites are generally fragile and hard to integrate. Metal powder cores either have acceptable permeability with too high core loss (heating) or acceptable core loss with too low permeability.

[0024] The present disclosure provides high frequency power inductor materials containing a polymer binder, and a plurality of multilayered flakes dispersed in the polymeric binder. In some embodiments, the multilayered flakes may have an average lateral size less than 400 micrometers, less than 300 micrometers, or less than 200 micrometers. The average lateral size refers to a flake lateral size at 50% of a total cumulative distribution by volume. In some embodiments, the multilayered flakes can be surface-treated with a phosphoric acid solution. The multilayered flakes after the surface-treating can be dispersed in the binder, where the multilayered flakes are substantially aligned parallel to one another such that they do not provide an electrically continuous path over a range of greater than 0.5 millimeters.

[0025] The high frequency power inductor materials described herein provide superior properties compared to other composite materials such as the multilayer flake composites described in WO2019/082013. For example, during operation, a power inductor not only has an AC current flowing inside but also has a superimposed DC current. The DC current creates a DC magnetic bias field. At high processor workload, the DC current is high, creating a high DC bias field. Composite materials may have an increasing core loss when the DC bias field is increasing, making it unfavorable in high current load applications. The high frequency power inductor materials can exhibit a significantly less increased core loss compared to those composite materials. In addition, the multilayered flakes described herein have a relatively smaller lateral size, which makes it more practical to coat using an automated coater. Furthermore, high frequency power inductor materials described herein can have more favorable repeatability in terms of low core loss, which renders further remedy processes (e.g., a second hot press step) unnecessary.

[0026] Referring to FIG. 1, high frequency power inductor material 100 has first and second opposed major surfaces 101, 102, polymeric binder 104, and plurality of multilayered flakes 106 dispersed in the polymeric binder 104. Multilayered flakes 106 each include at least two layer pairs 110. Each pair 110 include ferromagnetic material layer 111 and adjacent thereto electrically insulating dielectric layer 112 (comprised of electrically insulating material). Multilayered flakes 106 are substantially aligned parallel to first and second major surfaces 101, 102 such that they do not provide an electrically continuous path over a range of greater than 0.5 mm (i.e., multilayered flakes 106 are electrically isolated from each other). For example, the sheet resistance between two vias through the inductor material layer for some embodiments is greater than 10 Q/squarc. while for others it may be greater than 1 kQ/squarc. and yet for some it may be greater than 1 MQ/squarc.

[0027] Exemplary electrically insulating materials may include, on a theoretical basis, at least one of a nitride (e.g., SisN^), fluoride (e.g., MgF2), or oxide (e.g., AI2O3, HfCE, SiO, SiCE, Y2O3, ZnO, B2O3, and ZrCE). Sources of electrically insulating materials include those available from Zhongnuo Advanced Material, Beijing, China; EM Industries, Hawthorn, NY; Materion, Milwaukee, WI; and RD Mathis, Long Beach, CA. Other exemplary electrically insulating materials include high temperature (i.e., with a glass transition temperature, T _g , exceeding 250°C and decomposition temperatures exceeding 350°C) polymeric materials (e.g., polyimides).

[0028] In some embodiments, the ferromagnetic material may include at least one of Co, Fe, or Ni. In some embodiments, the ferromagnetic material may include at least two of Co, Fe, orNi (e.g., soft magnetic alloys of FeCo, NiFe, or FeCoNi). In some embodiments, the ferromagnetic material further may include at least one of Mo, Cr, Cu, V, Si, or Al as additional alloying elements (e.g., soft magnetic alloys of FeSiAl (also commonly referred to as “sendusf ’) or NiFeMo (commonly referred to as “supermalloy”)). In some embodiments, the ferromagnetic material includes crystalline ferromagnetic material (e.g., soft magnetic alloys of FeSiAl, NiFe, NiFeMo, FeCo, or FeCoNi). In some embodiments, the ferromagnetic material may include amorphous ferromagnetic metal (e.g., soft magnetic alloys of FeCoB, or TLTE, where TL is at least one of Fe, Co, or Ni, and TE is at least one of Zr, Ta, Nb, or Hf).

[0029] The use of ferromagnetic metal material layers provides high magnetic saturation induction. Variation in the aspect ratio of the two-dimensional flake can be used to control for higher permeability, or higher ferromagnetic resonance frequency (i.e., less loss coming from resonance). A higher ratio of flake diameter to flake thickness tends to increase permeability. Here, the flake diameter or lateral size refers to the dimension of the flake in a direction substantially perpendicular to the thickness direction. Further, the spaces between flakes form natural magnetically non-susceptible (permeability around 1.0) gaps leading to slow saturation.

[0030] It is to be understood that the flakes described herein may have various irregular lateral shapes such as shown in FIG. 2, a scanning electron microscope (SEM) image of flakes of Example 4 to be described further below. The flake lateral size can be measured by various methods. In one exemplary optical method, flake equivalent lateral size distributions of various samples can be measured with an optical particle size analyzer (available under the trade designation “Horiba Partica LA950V2”, Horiba Instrument Incorporated, Irvine, CA). Water can be used as a dispersant solvent to disperse the flakes. The measured flake lateral size distribution is volume based. FIG. 3 illustrates measured plots of exemplary flake lateral size distribution and cumulative distribution versus flake size. From the cumulative distribution, V10, V50 and V90 can be extracted, where V10, V50 and V90 represent the respective flake lateral sizes at 10%, 50% and 90% of total cumulative distribution. [0031] In some embodiments, the multilayered flakes may have an average lateral size less than 600 micrometers, less than 500 micrometers, less than 400 micrometers, less than 300 micrometers, less than 200 micrometers, less than 100 micrometers, or less than 50 micrometers. Here, the average lateral size may refer to the flake lateral size at 50% of the total cumulative distribution by volume (e.g., V50 as shown in FIG. 3).

[0032] In some embodiments, the average lateral size of the multilayered flakes may be less than 600 micrometers, and a majority of the multilayered flakes (e.g., 50 wt%, 60 wt%, 70 wt%, 80 wt% or 90 wt% or greater) may have an average lateral size in the range from 10 to 600 micrometers, from 10 to 500 micrometers, from 20 to 500 micrometers, from 20 to 400 micrometers, from 20 to 300 micrometers, or from 20 to 200 micrometers.

[0033] In some embodiments, the average lateral size of the multilayered flakes may have be less than 500 micrometers, and a majority of the multilayered flakes (e.g., 50 wt%, 60 wt%, 70 wt%, 80 wt% or 90 wt% or greater) may have an average lateral size in the range from 10 to 500 micrometers, from 10 to 400 micrometers, from 20 to 400 micrometers, from 20 to 300 micrometers, from 20 to 200 micrometers, or from 20 to 150 micrometers.

[0034] In some embodiments, the average lateral size of the multilayered flakes may be less than 400 micrometers, and a majority of the multilayered flakes (e.g., 50 wt%, 60 wt%, 70 wt%, 80 wt% or 90 wt% or greater) may have an average lateral size in the range from 10 to 400 micrometers, from 10 to 300 micrometers, from 20 to 400 micrometers, from 20 to 300 micrometers, from 20 to 200 micrometers, or from 20 to 150 micrometers.

[0035] In some embodiments, the average lateral size of the multilayered flakes may be less than 300 micrometers, and a majority of the multilayered flakes (e.g., 50 wt%, 60 wt%, 70 wt%, 80 wt% or 90 wt% or greater) may have an average lateral size in the range from 10 to 300 micrometers, from 20 to 300 micrometers, from 20 to 200 micrometers, or from 20 to 150 micrometers.

[0036] In some embodiments, the average lateral size of the multilayered flakes may be less than 200 micrometers, and a majority of the multilayered flakes (e.g., 50 wt%, 60 wt%, 70 wt%, 80 wt% or 90 wt% or greater) may have an average lateral size in the range from 10 to 200 micrometers, from 20 to 200 micrometers, or from 20 to 150 micrometers.

[0037] In some embodiments, the ferromagnetic material layers each may have a thickness up to 1000 (in some embodiments, up to 750, 500, 250, 200, or even up to 150) nm. It is generally desirable for the thickness of a ferromagnetic material layer to be less than /i (in some embodiments, less than ! ) of the skin depth Ds of the layer, wherein the skin depth Ds is calculated from the formula

Ds = 505*sqrt(p/pf), where p is the resistivity (Q-m) of the ferromagnetic layer, p is the relative permeability of the layer itself, and f is frequency (Hz) of the electrical excitation interacting with the inductor. The resulting value of Ds is expressed in meters of thickness.

[0038] In some embodiments, the electrically insulating layers each have a thickness of at least 5 (in some embodiments, up to 10, 15, 20, 25, 30, 35, 40, 50, 75, 100, 125, or even up to 150; in some embodiments, in a range from 5 to 150, 50 to 100, or even 10 to 150) nm. Typically, it is desirable for an electrically insulating layer to be as thin as possible while still ensuring adequate magnetic and electrical isolation of the ferromagnetic metal layers.

[0039] In some embodiments, the multilayered flakes each have a thickness up to 10 (in some embodiments, up to 9, 8, 7, 6, 5, 4, 3, 2, or even up to 1) micrometers.

[0040] In some embodiments, the multilayered flakes are present in an amount of at least 10 (in some embodiments, at least 20, 30, 40, 50, 60, or even 70; in some embodiments, in the range from 30 to 60) percent by volume of the high frequency power inductor material.

[0041] In some embodiments, the polymeric binder is at least one of polyhydric phenols, acrylates, benzoxazines, cyanate ester, polyimide, polyamide, polyester, polyurethanes, or epoxy resins (e.g., epoxy novolac resins).

[0042] In some embodiments, the high frequency power inductor materials described herein have a relative permeability of at least 20 (in some embodiments, at least 30, 40, 50, 75, 100, 150, 200, or even up to 250).

[0043] In some embodiments, the high frequency power inductor materials described herein have a saturation magnetic induction, B _s , of at least 0.2 (in some embodiments, at least 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, or even at least 1) T.

[0044] In some embodiments, the high frequency power inductor materials described herein have a magnetic resonance frequency in a range from 50 to 1500 (in some embodiments, 800 to 1400, or even 1000 to 5000) megahertz.

[0045] In some embodiments, the high frequency power inductor materials described herein have a magnetic coercivity, He, not greater than 800 (in some embodiments, not greater than 400) A/m.

[0046] In some embodiments, the flakes have an aspect ratio of up to 100: 1 (in some embodiments, at least 75: 1, 50: 1, 25: 1, or even up to 10: 1; in some embodiments, in a range from 10: 1 to 100: 1).

[0047] Tailoring flake size to control core loss under DC bias

[0048] The present disclosure provides high frequency power inductor materials where the change of core loss under DC bias field can be tuned by selecting different flake sizes. As described further above, during operation, power inductor has a DC current superimposed the AC current. The created DC magnetic bias field may increase core loss through altering permeability and loss characteristic in inductor materials. The present disclosure provides materials and methods to control core loss under DC bias by selecting different flake sizes of the high frequency power inductor materials described herein.

[0049] In some embodiments, the size of the multilayered flakes can be selected such that each have a lateral size less than 500 micrometers, less than 400 micrometers, less than 300 micrometers, or less than 200 micrometers. One or more sieves with certain mesh gauge can be used to select flakes having a lateral size less than a predetermined value.

[0050] In applications, power inductors operate under a DC magnetic bias field superimposed on AC component. The magnitude of the bias field depends on the amount of current handled by the converter. As an example, server a CPU requires higher current, so power inductors in servers experience much higher DC bias field than those in smartphones, the CPU of which uses relatively less current. DC bias fields can be tuned to some extent by adopting different inductor designs. In this disclosure, the larger size multilayer flakes have lower core loss at low DC bias field, but it increases with the bias field. Therefore, it is more suitable for applications with lower currents, such as smartphones. The smaller size multilayer flakes have relatively higher core loss at low DC bias field, and the core loss decreases with the bias field, which makes it fit into applications requiring high currents, such as desktop computers or servers.

[0051] Surface treatment

[0052] In some embodiments, the multilayered flakes can be surface-treated via a phosphoric acid. The flakes and phosphoric acid at a certain concentration can be mixed in a solvent such as, for example, N- propanol, Isopropyl alcohol, ethanol, or acetone. The ratio of acid/flake can be, for example, in the range 1/100, 1.5/100, 2/100, or 0.5/100-5/100 by weight. The mixture can then be heated (e.g., to 50 to 80 °C for 10 to 48 hours) while being mixed by a mixer. After the treatment, the flakes can post rinsed with, for example, n-propanol, and dried, e.g., at 70 °C for 24 hours. The surface-treated flakes are then dispersed in a polymeric binder to form a high frequency power inductor material, where the multilayered flakes are substantially aligned parallel to the first and second major surfaces such that they do not provide an electrically continuous path over a range of greater than 0.5 millimeters.

[0053] The present disclosure found a broad distribution of loss tangent at 20 MHz for samples including flakes without the surface treatment, for example, ranging from around 0.065 to about 0.219, implying a relatively large variation in core loss among repeating runs. For samples including surface- treated flakes, the loss tangent converges to a narrower band, for example, from 0.055 to 0.064. The samples with or without surface treatment may have an initial permeability in substantially the same range.

[0054] Exemplary high frequency power inductor materials described herein are useful, for example, as a power inductor in Point of Eoad (POE) converters, low profile inductors for inductive-capacitive (LC) filters (e.g., for global system mobile communication (GSM) pulse noise suppression in cellular phone speakers), or other applications wherein compact, inductive elements are required on a circuit board.

Listing of Exemplary Embodiments

Exemplary embodiments are listed below. It is to be understood that any one of embodiments 1- 17, 18-20, and 21 can be combined.

[0055] Embodiment 1 is a high frequency power inductor material having first and second opposed major surfaces, comprising: a polymer binder; and a plurality of multilayered flakes dispersed in the polymeric binder, the multilayered flakes comprising at least two layer pairs, wherein each layer pair comprises a layer of ferromagnetic material and a dielectric electrical isolation layer so that the ferromagnetic layers are electrically isolated from each other by dielectric layers, and wherein the multilayered flakes are substantially aligned parallel to the first and second major surfaces, wherein the multilayered flakes have an average lateral size less than 400 micrometers, less than 300 micrometers, or less than 200 micrometers, the average lateral size being a flake lateral size at 50% of a total cumulative distribution by volume.

[0056] Embodiment 2 is the high frequency power inductor material of embodiment 1, wherein the multilayered flakes have an average thickness less than 10 micrometers.

[0057] Embodiment 3 is the high frequency power inductor material of embodiment 1 or 2, wherein the multilayered flakes have an aspect ratio of up to 100: 1.

[0058] Embodiment 4 is the high frequency power inductor material of any preceding embodiment, wherein the ferromagnetic material comprises crystalline ferromagnetic material.

[0059] Embodiment 5 is the high frequency power inductor material of embodiment 4, wherein the ferromagnetic material is a NiFe soft magnetic alloy.

[0060] Embodiment 6 is the high frequency power inductor material of embodiment 4 or 5, wherein the ferromagnetic material is at least one of NiFe, FeCoNi, or FeCo soft magnetic alloy.

[0061] Embodiment 7 is the high frequency power inductor material of any preceding embodiment, wherein the ferromagnetic material layers each have a thickness up to 1000 nanometers.

[0062] Embodiment 8 is the high frequency power inductor material of any preceding embodiment, wherein the electrically insolating layers have an average thickness of at least 5 nanometers.

[0063] Embodiment 9 is the high frequency power inductor material of any preceding embodiment, wherein the multilayered flakes are present in an amount of at least 10 percent by volume of the high frequency power inductor material. [0064] Embodiment 10 is the high frequency power inductor material of any preceding embodiment, wherein the polymeric binder is at least one of polyhydric phenols, acrylates, benzoxazines, cyanate ester, polyimide, polyamide, polyester, polyurethanes, or epoxy resins.

[0065] Embodiment 11 is the high frequency power inductor material of any preceding embodiment having a relative permeability of at least 20.

[0066] Embodiment 12 is the high frequency power inductor material of any preceding embodiment having a saturation magnetic induction, Bs, of at least 0.2 Tesla.

[0067] Embodiment 13 is the high frequency power inductor material of any preceding embodiment having a magnetic resonance frequency in a range from 500 to 1500 megahertz.

[0068] Embodiment 14 is the high frequency power inductor material of any preceding embodiment having a magnetic coercivity, He, not greater than 10 Oersted or 800 Ampere/meter.

[0069] Embodiment 15 is the high frequency power inductor material of any preceding embodiment having a skin depth, wherein the ferromagnetic layer thickness is less than the skin depth at an electrical excitation of 20 MHz.

[0070] Embodiment 16 is the high frequency power inductor material of any preceding embodiment having a core loss density no greater than 10,000 kW/m ³ at 20 MHz with a maximum magnetic induction of 10 mT under a magnetic DC bias field from 0 to 2500 Ampere/meter (A/m).

[0071] Embodiment 17 is the high frequency power inductor material of any preceding embodiment having a core loss density no greater than 20,000 kW/m ³ at 20 MHz with a maximum magnetic induction of 15 mT under a magnetic DC bias field from 0 to 2500 A/m.

[0072] Embodiment 18 is a method of making a high frequency power inductor material, the method comprising: providing a plurality of multilayered flakes, the multilayered flakes comprising at least two layer pairs, wherein each layer pair comprises a layer of ferromagnetic material and a dielectric electrical isolation layer so that the ferromagnetic layers are electrically isolated from each other by dielectric layers; surface-treating the multilayered flakes with a phosphoric acid solution; and dispersing the multilayered flakes after the surface-treating in a polymeric binder, wherein the multilayered flakes are substantially aligned parallel to the first and second major surfaces.

[0073] Embodiment 19 is the method of embodiment 18, wherein surface-treating the multilayered flakes further comprises mixing the multilayered flakes and the phosphoric acid solution, optionally, heating the mixture up to 90 °C. [0074] Embodiment 20 is the method of embodiment 18 or 19, wherein the high frequency power inductor material has a distribution range of loss tangent TanO no greater than 0.12, or no greater than 0.07 at 20 MHz.

[0075] Embodiment 21 is a high frequency power inductor material having first and second opposed major surfaces, comprising: a polymer binder; and a plurality of multilayered flakes dispersed in the polymeric binder, the multilayered flakes comprising at least two layer pairs, wherein each layer pair comprises a layer of ferromagnetic material and a dielectric electrical isolation layer so that the ferromagnetic layers are electrically isolated from each other by dielectric layers, and wherein the multilayered flakes are substantially aligned parallel to the first and second major surfaces, wherein the multilayered flakes have an average lateral size less than 400 micrometers, less than 300 micrometers, or less than 200 micrometers, the average lateral size being a flake lateral size at 50% of a total cumulative distribution by volume, and wherein the multilayered flakes are surface-treated with a phosphoric acid solution. Examples

[0076] Unless otherwise noted or readily apparent from the context, all parts, percentages, ratios, etc. in the Examples and the rest of the specification are by weight. Table 1 provides abbreviations and a source for all materials used in the Examples below:

[0077] Table 1

Core Loss Measurement Test Method

[0078] The core loss was measured at 20 MHz as described in D. Hou et. al., “New high-frequency core loss measurement method with partial cancellation concept”, pp. 2987-2994, IEEE Transactions on Power Electronics, Vol. 32, No. 4, (2017), the disclosure of which is incorporated herein by reference.

Permeability Spectrum Measurement Test Method

[0079] Permeability spectrum from 1 MHz to 100 MHz is measured using an impedance analyzer

(available under the trade designation “KEYSIGHT E4990A” Keysight Technologies Inc., Santa Rosa, CA) and a terminal adapter (available under the trade designation “42942A” from Keysight Technologies Inc.).

Evaluation of Core loss via Loss Tangent TanO and initial permeability

[0080] Under no DC bias field, the magnetic core loss is related to initial permeability of the magnetic core through the following Equation (1): where Pv is magnetic core loss per volume, f is AC field frequency, go is the vacuum permeability, and B is maximum magnetic induction during the AC magnetic field swing, p’ and TanO are real part of permeability and loss tangent respectively. Loss tangent TanO is defined as a ratio between imaginary ( p ) and real (p’) part of the initial permeability. (J. K. Watson, Application of Magnetism. Gainesville, FL: Storter Printing Company, 1985, pp. 209-213) Initial permeability is defined as a complex permeability measured under low AC magnetic field (HAC«80 A/m). When TanO is small, Equation (1) can be approximated as:

[0081] Core loss density Pv becomes inversely proportional to p’/TanO. So, high p’ and low TanO are necessary to achieve low core loss and the two parameters provide a fast evaluation of the corresponding core loss density Pv. A broad distribution of loss tangent may imply unfavorable large variation in core loss among repeating runs.

Flake equivalent lateral size measurement test method

[0082] The multilayer flakes have irregular shapes, such as shown in FIG. 2, a scanning electron microscope (SEM) image of flakes of Example 4 to be described further below. Flake equivalent lateral size distributions for various samples were measured with an optical particle size analyzer (available under the trade designation “Horiba Partica LA950V2”, Horiba Instrument Incorporated, Irvine, CA). Sample Refractive index was set to 3.320 - O.OOOi and water was used as a dispersant solvent.

Circulation was set to 6 and ultrasonication was applied to improve flake dispersion in water. The measured size distribution was volume based. From cumulative distribution, V10, V50 and V90 were extracted, where V10, V50 and V90 represent the respective flake lateral sizes at 10%, 50% and 90% of total cumulative distribution.

Example 1 (EX-1)

[0083] Permeable multi-layered NiFe/insulator particulate material consisting of multiple sub-skindepth magnetic layers alternated with dielectric spacer layers (FFDM) particles (permeable multi-layered NiFe/insulator particulate material of multiple sub-skin-depth magnetic layers alternated with dielectric spacer layers available from 3M Company, St. Paul, MN, under the trade designation “3M FLUX FIELD DIRECTIONAL MATERIALS PARTICLE EM05EC”) were in the form of flakes with total flake thickness of about 6 micrometers, and a lateral size less than 500 micrometers. A first sieving was performed with a sieve shaker (available under the trade designation “vibratory sieve shaker AS 200” from Retsch, Haan, Germany) and a sieve with mesh gauge of 125 micrometers. Subsequently, larger flakes left in the sieve were broken down using a blender for 2 minutes. Then, the flakes were sieved for a second time using a sieve with mesh gauge of 32 micrometers. Flakes passed the sieve were collected for the following process. Flake equivalent lateral size measurement test method was used on the collected flakes and the results are displayed in Table 2.

[0084] Surface treatment was subsequently performed. 20 grams of flakes with the selected size were pre rinsed with heptane and acetone. The flakes and 0.47 gram phosphoric acid with concentration of 85% were added into 100 gram n-propanol. The mixture was heated to 70 °C for 24 hours while being mixed by a mixer at 1800 rpm. After treatment, the flakes were post rinsed with n-propanol, and dried in an oven set at 70 °C for 24 hours.

[0085] Materials of 11.8 gram hydroxypropyl cellulose (HPC) solution (2.5% in PGME) with molecular weight of 370,000, 0.2 gram polymeric dispersant Solsperse 41000 (available from Lubrizol, Wickliffe, OH) and 0.5 gram fluorosurfactant FC-4430 (available from 3M, Maplewood, MN) were mixed together using a mixer (available under the trade designation “ARE-310” from Thinky, Laguna Hills, CA) at 2000 rpm for 2 minutes. Next, 13 grams of sieved multilayer flakes were added and mixed again at 2000 rpm for 2 minutes. After that, a premixed epoxy solution of 2.5 grams were added under the same mixing conditions. The epoxy solution is made by dissolving a 80/20 mixture of EPON 1001F and EPON 154 (both available from Hexion Inc., Columbus, OH) in propylene glycol methyl ether (PGME) at 50% by weight. Finally, 1 gram PGME was mixed in and 0.2 gram diaminodiphenyl sulfone (DDS) was added as a curing agent with the same mixing conditions.

[0086] Right after mixing, the slurry was coated onto a polyethylene terephthalate (PET) substrate using HIRANO coater (available under the trade designation “TM-MC multi coater” from). Bar gap was set to 300 micrometers during coating with a line speed of 0.7 meter per minute and a tension of 60 Newton. Two heating zones with temperatures of 70 °C and 110 °C dried the coated film. The coated film then was released from the PET substrate and placed in a vacuum chamber for 24 hours with a roughing pump to remove all solvent residue.

[0087] After the dried film was cut and stacked, a heated press was used to press and cure the stack into a final composite sheet at 190 °C for 3 hours with a pressure of 0.83 ton per square inch. A pair of 0.09 inch thick 304 stainless steel plates with mirror-like surface sandwiched the stack during heated pressing. A dry film release agent (available under the trade designation “MR311” from Sprayon, Cleveland, OH) was sprayed on the plate surfaces before pressing. Rings with inner diameter of 8 mm and outer diameter of 12 mm were cut out for testing.

[0088] The Core Loss Measurement Test Method was used to measure the core loss of the composite EX-1. Test was performed at frequency f = 20 MHz. Magnetic induction swing B was set at 10 mT and 15 mT respectively. In both cases, core loss is highest at close to zero DC bias field. As DC bias field increases to around 1500 A/m, core loss decreases monotonically. Further increase in DC bias field makes little change in core loss. Comparing to other examples, the decreasing trend of core loss makes EX-1 composed of small flakes a beter candidate for application with high DC bias field. Example 2

[0089] EX-2 was made with the same raw materials and processes described in EX-1, except the second sieving, where a sieve with mesh gauge of 53 micrometers was used. Flake equivalent lateral size measurement test method was used on the collected flakes and the results are displayed in Table 2. The Core Loss Measurement Test Method was used to measure the core loss of the composite EX-1. Test conditions were the same as described in EX-1. At B=10 mT, as DC bias field increases from zero, core loss starts with 5900 kW/m ³ and decreases slightly to a botom of around 5700 kW/m ³ between 330 A/m and 750 A/m. Then it slowly grows to 6800 kW/m ³ at a bias field of 2600 A/m. At B=15 mT, core loss starts from 14000 kW/m ³ and decreases slightly to a botom of 12200 kW/m ³ between 750 A/m and 1100 A/m. Then it slowly grows to 14100 kW/m ³ at 2600 A/m. Overall, EX-2 has a rather small change of core loss over DC bias field.

[0090] EX-3 was made with the same raw materials and processes described in EX-1, except that in the second sieving process, a sieve with mesh gauge of 90 micrometers was used. Flake equivalent lateral size measurement test method was used on the collected flakes and the results are displayed in Table 2. Test conditions were the same as described in EX-1. At B=10 mT, as DC bias field increases from zero, core loss decreases from 4000 kW/m ³ to 3700 kW/m ³ at 270 A/m. Then the trend is switched. Core loss gradually increase to 6700 kW/m ³ at 2430 A/m. At B=15 mT, core loss first decreases slightly to a botom of 8830 kW/m ³ then increase gradually to 14100 kW/m ³ at 2480 A/m. In both cases, at around 1170 A/m, core loss of EX-3 surpasses that of EX-1, making it favorable for application with lower DC bias field.

[0091] EX-4 was made with the same raw materials and processes described in EX-1, except that in the second sieving process, a sieve with mesh gauge of 125 micrometers was used. Flake equivalent lateral size measurement test method was used on the collected flakes and the results are displayed in Table 2. Test conditions were the same as described in EX-1. In both cases of B=10 mT and 15 mT, core loss at close to zero DC field is the lowest among those of all five examples. Core loss gradually increases and exceeds that of EX-1 at DC bias field of 880 A/m and 960 A/m with B=10 mT and 15 mT respectively. The low core loss at low DC field makes EX-4 more suitable for application with DC bias field less than around 1000 A/m.

[0092] FFDM particles (described in EX-1) were sieved with mesh gauge of 500 micrometers. Flakes passed the sieve were collected. Flake equivalent lateral size measurement test method was used on the collected flakes and the results are displayed in Table 2. Four grams of the selected particles were mixed with 2.5 grams of polyimide resin (PIR) (available under the trade designation “UN1866 CPI” from NeXolve Corporation, Huntsville, AL) and 1 milliliter of diethylene glycol dimethyl ether (available from Alfa Aesar, Lancashire, United Kingdom) in a mixing jar. After mixing with a mixer, the slurry was coated onto a polyethylene terephthalate (PET) substrate using a film applicator (available under the trade designation “MICROM II FILM APPLICATOR” from Gardco, Pompano Beach, FL). The coated film was dried at 90°C for 1 hour. The composite sheet was then peeled off the substrate backing.

[0093] Subsequently, the composite sheet was cut and stacked for heated pressing. A heated press was used to densify the composite at 5 tons on a 4-inch (10-cm) diameter ram at 275°C for 60 minutes.

[0094] The Core Loss Measurement Test Method was used to measure the core loss of the composites EX-5. Test conditions were the same as described in EX-1. In both cases of B=10 mT and 15 mT, core loss increases dramatically starting from zero DC field. It quickly surpasses core loss of all previous samples at around DC field of 450 A/m, making EX-5 less favorable at high DC bias field.

[0095] FIGS. 4 and 5 are plots of Core Loss versus Magnetic DC bias field for Examples 1-5 (EX-1, EX-2, EX-3, EX-4, and EX-5) at B=10 mT and 15 mT, respectively.

[0096] Table 2

Example 6 (EX-6)

[0097] EX-6 is a set of 13 samples repeatedly made by the same procedure as EX-4.

Example 7 (EX-7)

[0098] EX-7 was made by the same procedure as EX-4 except no Surface treatment before the Slurry preparation was performed. The process of making EX-7 was repeated to make a set of 35 samples.

[0099] FIG. 6 shows Initial permeability and loss tangent for samples with surface treatment (EX-6) and samples without surface treatment (EX-7) at 20 MHz. The slope of each data point with respect to origin of the coordinates (0, 0) is p’/TanO. Data points close to top left comer are preferred, which means high initial permeability combined with low loss tangent. A broad distribution of loss tangent was observed for samples of EX-7, ranging from around 0.065 to about 0.219, implying a relatively large variation in core loss among repeating runs. The initial permeability for samples of EX-7 ranges from 105 to 158. For samples of EX-6, the loss tangent converges to a narrower band from 0.055 to 0.064. The initial permeability ranges from 112 and 160 without damage, similar to the samples of EX-7.

[00100] Foreseeable modifications and alterations of this disclosure will be apparent to those skilled in the art without departing from the scope and spirit of this disclosure. This disclosure should not be restricted to the embodiments that are set forth in this application for illustrative purposes.