Resonance-Based Sensing of Magnetic Nanoparticles Using Microfluidic Devices with Ferromagnetic Antidot Nanostructures

Abstract

We demonstrated resonance-based detection of magnetic nanoparticles employing novel designs based upon planar (on-chip) microresonators that may serve as alternatives to conventional magnetoresistive magnetic nanoparticle detectors. We detected 130 nm sized magnetic nanoparticle clusters immobilized on sensor surfaces after flowing through PDMS microfluidic channels molded using a 3D printed mold. Two detection schemes were investigated: (i) indirect detection incorporating ferromagnetic antidot nanostructures within microresonators, and (ii) direct detection of nanoparticles without an antidot lattice. Using scheme (i), magnetic nanoparticles noticeably downshifted the resonance fields of an antidot nanostructure by up to 207 G. In a similar antidot device in which nanoparticles were introduced via droplets rather than a microfluidic channel, the largest shift was only 44 G with a sensitivity of 7.57 G/ng. This indicated that introduction of the nanoparticles via microfluidics results in stronger responses from the ferromagnetic resonances. The results for both devices demonstrated that ferromagnetic antidot nanostructures incorporated within planar microresonators can detect nanoparticles captured from dispersions. Using detection scheme (ii), without the antidot array, we observed a strong resonance within the nanoparticles. The resonance's strength suggests that direct detection is more sensitive to magnetic nanoparticles than indirect detection using a nanostructure, in addition to being much simpler.

Keywords: ferromagnetic materials; microfluidics; soft lithography; superparamagnetic iron-oxide nanoparticles.

Figure A1

An image obtained using scanning electron microscopy of the first microresonator’s inductive ring after infusion of magnetic nanoparticles. Since the ring was misplaced near the edge of the microfluidic channel, the ring has been partially covered by a large agglomeration of nanoparticles.

Figure A2

An image obtained using scanning electron microscopy of the antidot nanostructure of the second device after infusion of magnetic nanoparticles. This ring was correctly placed near the center of the microfluidic channel, so it has not been covered in nanoparticles. Most of the particles have been caught within the antidots and along the two edges of the nanostructure perpendicular to the external magnetic field.

Figure A3

An image obtained using scanning electron microscopy of the antidot nanostructure in the third device after introduction of magnetic nanoparticles onto the surface via droplets. The MNPs have almost completely covered the antidot nanostructure. This image was obtained for an effective droplet concentration of 8 μg/mL.

Figure 1

(a) An optical microscopy image of a microresonator chip, depicting the microstripline, capacitive stubs C1 and C2, and the inductive ring. The inductive ring for this resonator contains an array of antidots etched into a sheet of Permalloy, shown in further detail in the inset. (b) Scanning electron microscopy (SEM) image of the antidot device’s inductive ring surrounding a 50 nm-thick square-shaped sheet of Permalloy. An 8 × 8 array of 400 nm-diameter circular antidots was defined into the 5 × 5 × 0.05 μm³ Permalloy sheet using electron beam lithography. The centers of the antidots are separated by 600 nm. The first microresonator device has an identical layout, but no Permalloy antidot nanostructure (AN) inside the ring.

Figure 2

The setup used to identify the resonance frequency of each microresonator and match the microresonators’ impedances to 50 Ω. Each microresonator is bonded to a coplanar stripline that is connected to one port of a vector network analyzer (VNA). A 50 Ω reference resistor is connected to the other port.

Figure 3

The molding process used to fabricate two open microfluidic channels from PDMS elastomer. (i) The U-shaped microchannel mold (red) is 3D printed with PLA thermoplastic. The channel is surrounded by a rectangular wall that contains the PDMS. The stub connected to the wall creates a space for the silicon piece used for impedance matching the microresonator. (ii) PDMS elastomer (blue) is poured into the mold and left for 24 h to set. (iii) The elastomer is removed from the mold. (iv) Luer stubs (orange) are connected to the inlet and outlet arms of the microchannel. This diagram is not to scale. (v) The microfluidic film is pressed against the surface of the microresonator film, closing the microfluidic channel. (vi) A photograph of the fabricated microchannel.

Figure 4

Diagram of the MNP detector showing the microresonator (orange) with two capacitive stubs C1 and C2. The inset shows the microresonator ring containing the antidot array nanostructure. The microfluidic channel (blue) is centered over the ring. The device is placed between two magnetic poles (green) providing an external magnetic dc field H along the plane of the detector and parallel to the microchannel. This diagram is not to scale.

Figure 5

The setup used to observe FMR in the sensors via broadband FMR spectroscopy [50,51]. Each sensor is placed between the poles of an electromagnet producing a static magnetic field in the plane of the sensor. A waveform generator sends an ac current through modulation coils attached to each pole, adding an ac component to the static magnetic field. The sensor is bonded to a coplanar stripline, which is connected to a microwave interferometer (or receiver) via a circulator. A signal generator feeds a microwave signal into the interferometer at the microresonator’s resonance frequency. Attenuators (A), phase shifters (φ), and low noise amplifiers are used to amplify the microwave signal returning from the sensor while maintaining a suitable signal-to-noise ratio [52]. Tuning of the interferometer is monitored using a digital voltmeter (DVM). A homodyne mixer is used to rectify the microwave signal, which is fed into a lock-in amplifier. The amplifier measures the amplitude of the signal and records these data to a PC for later analysis.

Figure 6

Voltage from the first device with a bare microresonator during the flow of MNPs through the tubing and microfluidic channel. The double-headed arrow indicates the approximate period of time in which the MNP suspension was flowing through the microfluidic channel. During this time, any changes in the signal may indicate the presence of MNPs.

Figure 7

Microwave signal amplitude from the first sensor (microresonator without antidot array) before and after MNPs flowed through the microfluidic channel. The external magnetic field was swept from remanence to 6 kG to obtain the ferromagnetic resonance spectrum for this bare device. The resonance of the MNPs can be clearly observed in the signal obtained after pumping the MNPs through the channel.

Figure 8

(a) Ferromagnetic resonance spectrum for an 8 × 8 array of 400 nm-diameter circular antidots numerically modeled using MuMax³. The external static magnetic field was kept at 4 kG and directed along the x-axis. The frequency of the microwave excitation was swept up to 30 GHz to obtain a frequency-resolved ferromagnetic resonance spectrum for the ANs employed in these sensors. A single-side mode labeled A is present in the spectrum, along with two strong extended modes labeled B and C. The A, B, and C modes are analogous to the modes observed in the second device’s FMR spectrum shown in Figure 10. For the third device, only the A mode and one of the extended modes can be observed in the spectrum shown in Figure 11. The two edge modes labeled D and E on the far left were not visible in either experimental spectrum. (b) The spatial distributions of the ferromagnetic resonance modes that occur at each of the peaks labeled A, C, and E in the ferromagnetic spectrum of the 8 × 8 circular antidot array. Each resonance mode is localized within a different region of the antidot nanostructure.

Figure 9

Voltage from the second device (microresonator with AN and microfluidics) during the flow of MNPs through the tubing and microfluidic channel. The double-headed arrow indicates the approximate period of time in which the MNP suspension was flowing through the microfluidic channel. During this time, any changes in the signal may indicate the presence of MNPs.

Figure 10

Microwave signal amplitude from the second device (microresonator with AN and microfluidics) before and after MNPs flowed through the microfluidic channel. The external dc magnetic field was swept from remanence to 6 kG to obtain the field-resolved ferromagnetic resonance spectrum for the device. The spectrum contains three ferromagnetic resonance modes labeled A, B, and C. After flowing MNPs, the B and C resonances occur at lower magnetic fields. Note that, since the simulated spectrum was frequency-resolved, the order of the modes appears reversed in this field-resolved spectrum.

Figure 11

Microwave signal amplitude from the third device (microresonator with an AN, no microfluidics) before MNPs were introduced to the surface of the microresonator. The external dc magnetic field was swept from remanence to 6 kG to obtain the field-resolved ferromagnetic resonance spectrum for the device. The spectrum contains two prominent ferromagnetic resonance modes labeled A and B. Note that, since the simulated spectrum was frequency-resolved, the order of the modes appears reversed in this field-resolved spectrum.

Figure 12

The resonance field of the side mode, A, of the third device (microresonator with AN, no microfluidics) as the effective concentration of the MNP dispersion droplet increases from 0 to 12 μg/mL. Initially, the resonance field increases with the droplet concentration, reaching a maximum shift of 20 G at 8 μg/mL. The resonance field then begins to decrease slightly, the final shifts being only approximately 8 G. Each of the points in this plot was averaged over six measurements of the FMR spectrum, with the error bars indicating the standard deviation of these six measurements.

Figure 13

The resonance field of the extended mode, B, of the third device (microresonator with AN, no microfluidics) as the effective concentration of the MNP dispersion droplet increases from 0 to 12 μg/mL. The resonance field steadily decreases with the droplet concentration, reaching a maximum downwards shift of 44 G at 12 μg/mL. Each of the points in this plot was averaged over 6 measurements of the FMR spectrum, with the error bars indicating the standard deviation of these measurements.