Fabrication and Characterization of Ferroﬂuidic-Based Wire-Wound and Wire-Bonded Type Inductor for Continuous RF Tunable Inductor

Featured Application: The growing market of Internet of Things (IOT) wireless multi-band systems creates the demand for high-quality and highly linear variable RF components. These components, including discrete and microelectronic mechanical systems (MEMS) inductors, have a wide range of applications in RF (radio frequency) transceivers, such as voltage-controlled oscillator (VCO), low-noise ampliﬁer (LNA), direct current—direct current (DC–DC) converters, matching networks, multi-band ﬁlters, multi-band RF circuits, RF power ampliﬁers in radio transmitters, high-isolated switches, and reconﬁgurable antennas. Abstract: This paper reports on the novel implementation of a tunable solenoid inductor using the ﬂuid-based inductance varying technique. The concept utilized material permeability variation that directly modiﬁes self-induced magnetic ﬂow density inside the coil, which in turn creates a variation of the inductance value. The core is formed by a channel which allows the circulation of a liquid through it. The liquid proposed for this technique has ferromagnetic behavior, called ferroﬂuid, with a magnetic permeability higher than unity. To evaluate the proposed technique, two di ﬀ erent types of solenoid inductor were designed, simulated and measured. The two structures are wire-wound and wire bond solenoid inductors. The structures are simulated in a 3D EM analysis tool followed by fabrication, test and measurement. The wire-bonded-based inductor showed a quality factor of 12.7 at 310 MHz, with 81% tuning ratio, by using ferroﬂuid EMG 901. The wire-wound-based inductor showed that the maximum tuning ratio is 90.6% with quality factor 31.3 at 300 MHz for ferro ﬂuidic EMG 901. The maximum measured tuning ranges were equal to 83.5% and 56.2% for the wire-wound type and the wire-bonded one, respectively. The measurement results for the proposed technique showed a very high tuning range, as well as high quality factor and continuous tunability.

shows an image of the implemented wire-wound inductor. Insulated copper wire was used as the coil because of its high electrical conductivity to improve the quality factor and its flexibility to easily formed into coil. In addition, micropipe and polydimethylsiloxane (PDMS) were used to realize the inner channel. The designed wire-wound inductors have 10 turns and 1-mm inner diameter with 0.5 mm wire thickness, as shown in Table 1. The solenoid core is realized by a 1-mm micropipe. Figure 1a shows the micropipe used to create microchannel mold, while Figure 1b,c show dimension of the device and the completed wire-wound inductor with microchannel PDMS, respectively. Table 1. Dimension of fabricated wire-wound inductor structure.

Type Wire Thickness Number of Turns Inner Diameter
Wire-wound solenoid 0.5 mm

Wire Bond Inductor
The implemented wire bond inductor is as shown in Figure 2. Coil structure was formed by two components; copper traces on the PCB and bonding wire as demonstrate in Figure 2a. The PCB used in this work is ROGERS4003, with 0.8-mm dielectric thickness, relative permittivity of 3.38, loss tangent of 0.0027, metal conductor thickness of 18µ m and metal conductivity of 5.8 × 10 7 S/m. The metal width at the base is 250 µ m with a space of 250 µ m. The bond wire is gold with 25 µm diameter.
As for catalyst to change the substrate property, thus changing the inductance value, PDMS based microchannel was attached onto the PCB, as shown in Figure 2b. The specification of wire bond

Wire Bond Inductor
The implemented wire bond inductor is as shown in Figure 2. Coil structure was formed by two components; copper traces on the PCB and bonding wire as demonstrate in Figure 2a. The PCB used in this work is ROGERS4003, with 0.8-mm dielectric thickness, relative permittivity of 3.38, loss tangent of 0.0027, metal conductor thickness of 18µm and metal conductivity of 5.8 × 10 7 S/m. The metal width at the base is 250 µm with a space of 250 µm. The bond wire is gold with 25 µm diameter.

Characterization of Tunable Inductor
The fabricated wound and bonded structure inductor contains two ports to connect to the network analyzer, with sub-miniature version A (SMA) connectors. The network analyzer was calibrated up to the end of its ports' connectors (up to the ports of the device under test), so that the insertion and mismatch loss due to the ports and their corresponding connectors are eliminated. The calibration kit used for this work is SOLT, which is short, open, load and through. The wire-wound solenoid inductor was fabricated using Rogers 4003 laminate.
In addition, to remove the connectors and lines effect of the PCB until the heads of the inductor, a SOLT calibration kit was designed and fabricated using FR-4 PCB. The FR-4 laminate is the identical material used for fabricated spiral inductor. The calibration kit fabricated for this work is shown in Figure 3.
The inductor was measured in an empty state and the state when the ferrofluid was injected to the channel. The measured S-parameter data were used to determine the inductance and quality factor. The measurement setup, including the network analyzer and injection tools used for measuring inductance and quality factors, is shown in Figure 3. The wire-wound solenoid inductor was first measured when its channel was empty. Then, the ferrofluids, such as EMG901, were injected into the channel gradually until it was completely filled. This causes the inductance to increase As for catalyst to change the substrate property, thus changing the inductance value, PDMS based microchannel was attached onto the PCB, as shown in Figure 2b. The specification of wire bond solenoid inductor for metal thickness, number of turns, space between turns and bond diameter were 250 µm, 10, 250 µm and 25 µm, respectively.

Characterization of Tunable Inductor
The fabricated wound and bonded structure inductor contains two ports to connect to the network analyzer, with sub-miniature version A (SMA) connectors. The network analyzer was calibrated up to the end of its ports' connectors (up to the ports of the device under test), so that the insertion and mismatch loss due to the ports and their corresponding connectors are eliminated. The calibration kit used for this work is SOLT, which is short, open, load and through. The wire-wound solenoid inductor was fabricated using Rogers 4003 laminate.
In addition, to remove the connectors and lines effect of the PCB until the heads of the inductor, a SOLT calibration kit was designed and fabricated using FR-4 PCB. The FR-4 laminate is the identical material used for fabricated spiral inductor. The calibration kit fabricated for this work is shown in Figure 3. continuously. The maximum inductance is achieved when the channel is fully injected. The same procedure was done for another three ferrofluids with different permeability and magnetic saturation. This method was implemented for both structures.

Ferrofluidic Properties
Ferrofluid is a type of material that can act like a liquid and a magnetic solid. It is made of nanoscale ferromagnetic particles and suspended in a liquid (such as oil). When no magnet is nearby, the material acts like a liquid, but when a magnet is nearby, the particles become temporarily magnetized and the ferrofluid acts like a solid. When the magnet is removed, it returns to its liquid state. The particles, which have an average size of about 100 Å (10 nm), are coated with a stabilizing dispersing agent (surfactant) which prevents particle agglomeration, even when a strong magnetic field gradient is applied to the ferrofluid. In this work, the ferrofluid was supplied from Ferrotec Co. Ltd. (www.ferrotec.com). As shown in Table 2, in this work, four ferrofluids, referred to as EMG901, EMG905, EMG909 and EMG911, were applied to, respectively, initial magnetic permeabilities of 5.4, 3.1, 1.9 and 1.3; magnetic saturation equal to 66 mT, 44 mT, 22 mT and 11 mT, and magnetic particle concentration 11.8%, 7.8%, 3.9% and 2%. The inductor was measured in an empty state and the state when the ferrofluid was injected to the channel. The measured S-parameter data were used to determine the inductance and quality factor. The measurement setup, including the network analyzer and injection tools used for measuring inductance and quality factors, is shown in Figure 3. The wire-wound solenoid inductor was first measured when its channel was empty. Then, the ferrofluids, such as EMG901, were injected into the channel gradually until it was completely filled. This causes the inductance to increase continuously. The maximum inductance is achieved when the channel is fully injected. The same procedure was done for another three ferrofluids with different permeability and magnetic saturation. This method was implemented for both structures.

Ferrofluidic Properties
Ferrofluid is a type of material that can act like a liquid and a magnetic solid. It is made of nanoscale ferromagnetic particles and suspended in a liquid (such as oil). When no magnet is nearby, the material acts like a liquid, but when a magnet is nearby, the particles become temporarily magnetized and the ferrofluid acts like a solid. When the magnet is removed, it returns to its liquid state. The particles, which have an average size of about 100 Å (10 nm), are coated with a stabilizing dispersing agent (surfactant) which prevents particle agglomeration, even when a strong magnetic field gradient is applied to the ferrofluid. In this work, the ferrofluid was supplied from Ferrotec Co. Ltd. (www.ferrotec.com). As shown in Table 2, in this work, four ferrofluids, referred to as EMG901, Appl. Sci. 2020, 10, 3776 6 of 15 EMG905, EMG909 and EMG911, were applied to, respectively, initial magnetic permeabilities of 5.4, 3.1, 1.9 and 1.3; magnetic saturation equal to 66 mT, 44 mT, 22 mT and 11 mT, and magnetic particle concentration 11.8%, 7.8%, 3.9% and 2%.

Results and Discussion
The wire bond solenoid inductor is modeled and simulated in High Frequency Structure Simulator (HFSS) as shown in Figure 4a. As can be seen, the metal base is implemented on a ROGERS4003 dielectric. It is recommended to use a relative permittivity of 3.55 for simulation. The wire bonds are modeled, as well, to form the solenoid inductor. Two excitation ports act as the terminals with a common reference ground. A three-dimensional solenoid inductor for both inductor types was designed in the HFSS EM simulation tool. Figure 4b shows the bond wire (bridge) connections on the base part. The PCB used in this work is ROGERS4003, with 0.8-mm dielectric thickness, a relative permittivity of 3.38, a loss tangent of 0.0027, a metal conductor thickness of 18 µm and a metal conductivity of 5.8 × 10 7 S/m. The metal width at the base is 10 mL with a space of 10mL. The bond wire is gold (Au) with 25 µm diameter. Ferrofluid was injected into the channel, then it remains static in channel during measurement. The characterization of the inductance was performed based on the percentage of liquid in the channel. As explained in Section 3.2, ferrofluid was modeled as solid magnetic bar when the particles were magnetized.

Results and Discussion
The wire bond solenoid inductor is modeled and simulated in High Frequency Structure Simulator (HFSS) as shown in Figure 4a. As can be seen, the metal base is implemented on a ROGERS4003 dielectric. It is recommended to use a relative permittivity of 3.55 for simulation. The wire bonds are modeled, as well, to form the solenoid inductor. Two excitation ports act as the terminals with a common reference ground. A three-dimensional solenoid inductor for both inductor types was designed in the HFSS EM simulation tool. Figure 4b shows the bond wire (bridge) connections on the base part. The PCB used in this work is ROGERS4003, with 0.8-mm dielectric thickness, a relative permittivity of 3.38, a loss tangent of 0.0027, a metal conductor thickness of 18 µ m and a metal conductivity of 5.8  10 7 S/m. The metal width at the base is 10 mL with a space of 10mL. The bond wire is gold (Au) with 25 µ m diameter. Ferrofluid was injected into the channel, then it remains static in channel during measurement. The characterization of the inductance was performed based on the percentage of liquid in the channel. As explained in Section 3.2, ferrofluid was modeled as solid magnetic bar when the particles were magnetized.
Using the simulated data, the discrepancy between measurement results and the ideal solution can be obtained. Figure 4a,b show the simulated 3D view of the wire-wound inductor and wire bond inductor, respectively.
(a)   Figure 5a illustrates the simulation results of inductance variation, corresponding to the full injection of the different ferrofluids. As can be seen in this figure, the minimum inductance is obtained at vacuum. When the channel is empty, the core is filled by air, where the permeability is approximately equal to vacuum. As tunable fluidic based inductors, ferrofluids will be used for injection into the core. According to the permeability of the injected liquid, the inductance is varied. In this work, four ferrofluids, referred to as EMG901, EMG905, EMG909 and EMG911, were applied to, respectively, initial magnetic permeabilities of 5.4, 3.1, 1.9 and 1.3, magnetic saturation equal to 66 mT, 44 mT, 22 mT and 11 mT, and magnetic particle concentrations 11.8%, 7.8%, 3.9% and 2%.
The maximum inductance is achieved for EMG901. EMG901 has the highest permeability among the applied ferrofluids. The simulated inductance for vacuum is 54.4 nH at 100 MHz, and increases to 65.4, 86.1, 105.6 and 187.3 nH for EMG911, EMG909, EMG905 and EMG901, respectively. Figure  5b shows the simulated corresponding quality factor. For all fluids, the quality factor obtained is over 300 at 100 MHz. The quality factor slightly decreases when the ferrofluids are injected. However, the quality factor reduction is more severe for EMG901. Figure 5c shows the simulated tuning ratio. The maximum tuning ratio obtained is larger than 243% for EMG901 at 100 MHz. For other ferrofluids, the tuning ratio obtained is from 20.22% to 243%, for a frequency range of a few MHz to 900 MHz. Using the simulated data, the discrepancy between measurement results and the ideal solution can be obtained. Figure 4a,b show the simulated 3D view of the wire-wound inductor and wire bond inductor, respectively. Figure 5a illustrates the simulation results of inductance variation, corresponding to the full injection of the different ferrofluids. As can be seen in this figure, the minimum inductance is obtained at vacuum. When the channel is empty, the core is filled by air, where the permeability is approximately equal to vacuum. As tunable fluidic based inductors, ferrofluids will be used for injection into the core. According to the permeability of the injected liquid, the inductance is varied. In this work, four ferrofluids, referred to as EMG901, EMG905, EMG909 and EMG911, were applied to, respectively, initial magnetic permeabilities of 5.4, 3.1, 1.9 and 1.3, magnetic saturation equal to 66 mT, 44 mT, 22 mT and 11 mT, and magnetic particle concentrations 11.8%, 7.8%, 3.9% and 2%.

Wire-Wound Solenoid Inductor
The maximum inductance is achieved for EMG901. EMG901 has the highest permeability among the applied ferrofluids. The simulated inductance for vacuum is 54.4 nH at 100 MHz, and increases to 65.4, 86.1, 105.6 and 187.3 nH for EMG911, EMG909, EMG905 and EMG901, respectively. Figure 5b shows the simulated corresponding quality factor. For all fluids, the quality factor obtained is over 300 at 100 MHz. The quality factor slightly decreases when the ferrofluids are injected. However, the quality factor reduction is more severe for EMG901. Figure 5c shows the simulated tuning ratio. The maximum tuning ratio obtained is larger than 243% for EMG901 at 100 MHz. For other ferrofluids, the tuning ratio obtained is from 20.22% to 243%, for a frequency range of a few MHz to 900 MHz. Appl. Sci. 2020, 10, x 8 of 15  Figure 6 shows the measurement result of the fabricated wire-wound inductor. As simulated, the maximum inductance is obtained for EMG901, while the vacuum presents the minimum achievable inductance. In Figure 6a, the measured inductance values are 55, 62.4, 68.8, 78.6 and 101 nH at 200 MHz, respectively, for vacuum, EMG911, EMG909, EMG905 and EMG901. The maximum quality factor is achieved during vacuum at 120 for 150 MHz. The quality factor slightly drops when the ferrofluids are injected. Figure 6c shows the tuning ratio of the fabricated wire-wound inductor.  Figure 6 shows the measurement result of the fabricated wire-wound inductor. As simulated, the maximum inductance is obtained for EMG901, while the vacuum presents the minimum achievable inductance. In Figure 6a, the measured inductance values are 55, 62.4, 68.8, 78.6 and 101 nH at 200 MHz, respectively, for vacuum, EMG911, EMG909, EMG905 and EMG901. The maximum quality factor is achieved during vacuum at 120 for 150 MHz. The quality factor slightly drops when the ferrofluids are injected. Figure 6c shows the tuning ratio of the fabricated wire-wound inductor. The measured tuning ratio for EMG901 at 200 MHz is 83.5%, while it increases to 90.6% at 300 MHz. As for the EMG911, EMG 909 and EMG905, the tuning ratios are 13.3%, 25% and 42.9%, respectively. Table 3 showed the comparison between the simulated and measurement result for the wire-wound inductor. As shown in the Table 3, the Q factor and tuning range for the simulated and measurement result at 100 MHz for EMG901 were 300%, 243%, 53.7% and 83.5%, respectively. The reason for the large difference between simulation and measurement data is because, in simulation, one only considers open space without close channel, thus, the effect of inner wall surface properties such as surface tensioning and meniscus force were not considered in the simulation model. Thus, the simulation was assumed to be a smooth movement of fluid in the channel. The simulation also assumes, as an ideal case, that parasitic capacitance is not included in this model. In the measurement result, the difficulties of movement fluidic and parasitic capacitance were affecting the value of inductance, and this contributed to the low tuning ratio and Q factor. The tuning ratio percentage is calculated by where L 1 and L 2 are the minimum and maximum achieved inductances.
Appl. Sci. 2020, 10, x 9 of 15 The measured tuning ratio for EMG901 at 200 MHz is 83.5%, while it increases to 90.6% at 300 MHz. As for the EMG911, EMG 909 and EMG905, the tuning ratios are 13.3%, 25% and 42.9%, respectively. Table 3 showed the comparison between the simulated and measurement result for the wire-wound inductor. As shown in the table 3, the Q factor and tuning range for the simulated and measurement result at 100 MHz for EMG901 were 300%, 243%, 53.7% and 83.5%, respectively. The reason for the large difference between simulation and measurement data is because, in simulation, one only considers open space without close channel, thus, the effect of inner wall surface properties such as surface tensioning and meniscus force were not considered in the simulation model. Thus, the simulation was assumed to be a smooth movement of fluid in the channel. The simulation also assumes, as an ideal case, that parasitic capacitance is not included in this model. In the measurement result, the difficulties of movement fluidic and parasitic capacitance were affecting the value of inductance, and this contributed to the low tuning ratio and Q factor. The tuning ratio percentage is calculated by where L1 and L2 are the minimum and maximum achieved inductances.   Figure 7a shows the simulated result of the wire bond inductor that varies from 47 to 181 at 200 MHz. For vacuum, EMG911, EMG909, EMG905 and EMG901, the simulated inductances are 47.8, 61.5, 85.5, 107.8 and 286.3 nH at 300 MHz, respectively. The contra result observed is compared to the wire-wound inductor, where, according to Figure 7b, the maximum quality factor obtained is 70 at 350 MHz for EMG901. Figure 7c shows the simulated tuning ratio, where the maximum tuning ratio belongs to the EMG901, equal to 498% at 300 MHz. According to the inductance variations, the minimum and the maximum tuning ratio correspond to EMG911 and EMG901, respectively.   Figure 7a shows the simulated result of the wire bond inductor that varies from 47 to 181 at 200 MHz. For vacuum, EMG911, EMG909, EMG905 and EMG901, the simulated inductances are 47.8, 61.5, 85.5, 107.8 and 286.3 nH at 300 MHz, respectively. The contra result observed is compared to the wire-wound inductor, where, according to Figure 7b, the maximum quality factor obtained is 70 at 350 MHz for EMG901. Figure 7c shows the simulated tuning ratio, where the maximum tuning ratio belongs to the EMG901, equal to 498% at 300 MHz. According to the inductance variations, the minimum and the maximum tuning ratio correspond to EMG911 and EMG901, respectively. Figure 8 shows the measurement result for the wire bond inductor. As shown in Figure 7a, a minimum inductance value of 42.6 nH at 150 MHz showed at vacuum. The maximum inductance obtained is 66.6 nH at 150 MHz, with EMG901 ferrofluids. The measured inductances for EMG911, EMG909 and EMG905 are 46.2, 49.8, 54.2 nH at 150 MHz, respectively. The measured quality factors are shown in Figure 8b. The maximum quality factor obtained with a vacuum at 39.1 for 150 MHz, while EMG901-filled channel presents the minimum achieved quality factor of 17.2 at 150 MHz. Figure 8c shows the tuning ratio. The maximum achieved tuning ratio is 81% for EMG901 at 310 MHz. Table 2 showed the summary of performance for the wire-bonded structure tunable inductor. Table 4 shows the characterization of wire-bonded structure tunable inductor with different type of ferrofluid. Table 5 shows the summary of the fabricated tunable inductor compared to the other work using solenoid structure technology. As shown in Table 5, the fabricated tunable inductor proved to be a promising and alternative tunable inductor. 61.5, 85.5, 107.8 and 286.3 nH at 300 MHz, respectively. The contra result observed is compared to the wire-wound inductor, where, according to Figure 7b, the maximum quality factor obtained is 70 at 350 MHz for EMG901. Figure 7c shows the simulated tuning ratio, where the maximum tuning ratio belongs to the EMG901, equal to 498% at 300 MHz. According to the inductance variations, the minimum and the maximum tuning ratio correspond to EMG911 and EMG901, respectively.

Comparison Fabricated Inductor to Previous Work
A comparison table to compare the proposed techniques, wire-bonded and wire-wound inductors with two other solenoid inductors is provided in Table 5. The wire-bonded and wire-wound inductors show a maximum tuning ratio of 81% and 90.6% at 310-and 300 MHz with EMG901, respectively, which is quite a bit higher than the one achieved by (Yoon et al. 1999b) at 5 MHz, but comparable to the one achieved by (Ning et al. 2006) at 100 MHz. However, the quality factor of the proposed inductor (29.88 and 45.2 at 310 and 300 MHz) are much higher than the one measured by (Vroubel et al. 2004). This shows that the proposed techniques maintain a high quality factor, which is necessary for RF applications.
The measured peak quality factor for wire-bond and wire-wound solenoid inductors are 18.4 at 108 MHz and 69 at 30 MHz, respectively, when EMG901 is fully injected. However, depending on the application and the desired tuning range, the quality factor can be increased in lower-permeability EMG liquids are selected. In this case, a lower tuning ratio is achieved.

Conclusions
The tunability of two microfluidic inductor structures were investigated; wire-wound inductor and wire bond inductor. Oil-based ferrofluids as high permeable liquids were used as microfluid. The liquid is being injected through microchannel created for bot inductor structure to alter its permeability property, thus giving a tunable inductance. For the wire-wound structure, the highest tuning range was achieved with EMG901 ferrofluid equal to 90.6% at 300MHz, while the best quality factor was maintained. The wire-bonded structure, which has a high potential to use in the MEMS process, achieved a tuning ratio of 81% at 310 MHz with EMG901 ferrofluid. As for a conclusion, a tunable inductor with high quality factor can be realized using microfluidic techniques.