Ultra-high rate of temperature increment from superparamagnetic nanoparticles for highly efficient hyperthermia

The magneto-thermal effect, which represents the conversion of magnetostatic energy to heat from magnetic materials, has been spotlighted for potential therapeutic usage in hyperthermia treatments. However, the realization of its potential has been challenged owing to the limited heating from the magnetic nanoparticles. Here, we explored a new-concept of magneto-thermal modality marked by low-power-driven, fast resonant spin-excitation followed by consequent energy dissipation, which concept has yet to be realized for current hyperthermia applications. We investigated the effect of spin resonance-mediated heat dissipation using superparamagnetic Fe3O4 nanoparticles and achieved an extraordinary initial temperature increment rate of more than 150 K/s, which is a significant increase in comparison to that for the conventional magnetic heat induction of nanoparticles. This work would offer highly efficient heat generation and precision wireless controllability for realization of magnetic-hyperthermia-based medical treatment.

www.nature.com/scientificreports/ (IR) thermographic method. The novel mechanism offers exceedingly high-efficiency ultrafast local heating and consequently exceptionally high rates of temperature increment by adjustment of controllable field parameters such as the frequency and strength of AC magnetic fields and pulse width as well as DC field strength.

Results
High-efficiency heat generation based on resonant spin-excitation and dissipation. Figure 1 shows the underlying mechanism of resonant spin-excitation and consequent dissipation in magnetic nanoparticles. When a DC magnetic field of sufficient strength is applied to magnetic particles, individual magnetic moments (spins) inside them are aligned in the direction of the DC magnetic field. Then, a microwave magnetic field applied to the particles can make the magnetization (M) precess around the direction of the effective field (H eff ) inside the magnetic particles. This precessional motion of M can be effectively excited under its resonance condition, in which case the frequency of the microwave field is tuned to the intrinsic resonance frequency of the precession of M, as expressed by f R = γ 2π H eff , with γ the gyromagnetic ratio. This entire dynamic motion of M is expressed by Landau-Lifshitz-Gilbert (LLG) equation dM/dt = −γ M × H eff + (α/M S )M × dM/dt , where the terms correspond to the precession of M and its phenomenological damping, respectively. The precession motion excited by the microwave field is purely dissipative via damping; thus, the energy of the magnetic system is converted to heat through nonlinear spin relaxations introduced by various interactions such as spinorbit coupling, two-magnon scattering, and field inhomogeneity 19 . Since the time scale of spin dynamics is on the order of nano seconds 20 , ultra-fast time-scale local heating is possible, indeed faster than μs in principle, compared with that of Néel-Brownian relaxation mechanism 21 . In earlier work, we reported that the resonant spin dynamics of magnetic particles in the uniform magnetization state or vortex state according to particle size 17,18,22 can theoretically be used as a novel means to make nanoparticles a local heat source of exceedingly high heating power. Here, in the present study, we experimentally demonstrated the thermal effect of resonant spin excitation and relaxation using Fe 3 O 4 magnetic nanoparticles.

Measurement of temperature increments from Fe 3 O 4 nanoparticles.
In order to investigate heat generation via the energy dissipation associated with the above-noted mechanism, we devised an apparatus for measurement of Fe 3 O 4 particle temperature. The apparatus is composed mainly of a radio-frequency (RF) power pumping system to generate resonant spin-excitation and a thermal IR camera to directly measure the temperatures of 15 nm-size Fe 3 O 4 particles without silica shells (for details on the synthesis of Fe 3 O 4 nanoparticles, see Supplementary Section S1) via thermal-radiation detection means, as schematically represented in Fig. 2a (for details on the measurement system, see "Methods"). Figure 2b   Resonant spin-excitation and relaxation dynamics for dissipative local heating. RF magnetic fields were applied to excite magnetic nanoparticles under an external DC magnetic field. When the RF field frequency was tuned to an intrinsic resonant frequency of the precession of individual magnetic moments (indicated by red arrows), the precession on a specific angle occurred around the internal effective field (see blue cones). Simultaneously, the precession of the magnetic moments started to lose a certain amount of absorbed magnetic energy owing to intrinsic damping, resulting in the reorientation of magnetizations in the DC field direction. The energy due to magnetic loss was dissipated in the form of lattice vibrations through the various spin-lattice interactions, thereby increasing the temperature of the magnetic nanoparticles. www.nature.com/scientificreports/ ent duration times of 0.1, 10, and 30 s, respectively. In all of the cases, a DC field strength of H DC = 750 Oe was applied for measurement of the IR images. In the RF-off IR image, the purple color in the middle corresponds to the room temperature of the magnetic particles. Immediately after applying the RF field, at Δt = 0.1 s, the color of the Fe 3 O 4 particles turns to red, as observed only in the region of the particles. For the longer duration time of Δt = 10 s, the red color becomes yellow, indicating that the temperature increases further. For Δt = 30 s, the yellow color becomes brighter, corresponding to a 45 ℃ particle temperature. The colors of the sample stage also turned out to be red and then yellow, because the sample stage was also heated to 34 ℃ through thermal-conductionbased heat transfer. These IR images clearly reveal that the temperature increment of Fe 3 O 4 particles can be quantitatively measured via thermal radiation entailing heating of particles by resonant spin-excitation and the consequent continuous dissipation. The temperature measurements were carried out for different strengths, frequencies, and pulse widths of RF magnetic fields as well as DC field strengths.  In order to generate RF magnetic fields (by power pumping), RF currents of GHz frequencies were transmitted to the microstrip using a signal generator and an RF amplifier. A DC magnetic field was also applied on the axis of a microstrip line. The direction of the RF magnetic field was perpendicular to the DC field direction. The change of temperature of the particles was recorded by an IR camera through thermal radiation. A detailed explanation of the experimental procedure is given in Method. (b) Optical image (left) of sample stage and thermal IR images (right four) for different duration times ∆t of application of RF magnetic fields (i.e. ∆t = 0, 0.1, 10, and 30 s). The color bar on the right indicates the local temperature. This agreement was also confirmed by micromagnetic simulations (see Supplementary Section S4), evidencing that the temperature increments are caused purely by heat generation associated with the energy dissipation due to damping against the magnetization precession.
On the basis of the above experimental demonstration of temperature increments from nanoparticles, we further measured the ΔT variation as a function of time under the same resonance condition, i.e., the application of the RF field (f AC = 3.0 GHz and H AC = 2.37 Oe) in a single pulse of 1-s-duration time. Figure 3b illustrates the results of ΔT vs. t for different H DC values along with Fig. 3c's corresponding perspective view of the H DC -t plane. All of the shapes of ΔT vs. t have a similar trend, which represents an initially rapid increase followed by a slow further increase and then a sudden drop at RF field-off. The slow increase after the initial fast increase in the particle temperature is due to heat conduction from the particles in direct contact with the sample stage (a hydrocarbon/ceramic substrate). Remarkable differences between the individual H DC strengths are the initial   . S7d). Temperature cooling immediately occurs upon turning off of the RF field, showing classical exponential cooling trends similar to those of the temperature increments. To sum up, the largest peak in ΔT on the H DC -t plane shown in Fig. 3c indicates that the resonant RF pulse field allows for a sufficiently fast temperature increment and, consequently, possible targeted heating from the Fe 3 O 4 nanoparticles through the sustainable spin-excitation and relaxation process.
Initial temperature increment rate κ. On the other hand, in a non-adiabatic system, samples start to lose heat mainly through conduction to the environment when the temperature of the sample is higher than that of the surroundings. Measurement of the initial temperature increment rate dT/dt| ∆t=0 immediately upon application of the RF field can provide a good linear approximation to an ideal adiabatic system. Thus, heat loss can be assumed to be negligible at the initial stage of fast heat transfer, and κ = dT/dt| ∆t=0 can be expressed by a  Fig. 4(a). As noted earlier, under the resonance condition (red symbols), the temperature increment was observed to reach 19 K within 1 s. Contrastingly, under the off-resonance condition (blue symbols), the temperature increment was as small as 4 K. This remarkable difference between the resonance and off-resonance cases could be further clarified by the contrasting κ values of the time-derivative ΔT profiles, as shown in the inset of Fig. 4a. For resonance case, κ reached 93 K/s, which value is astonishingly high, especially as compared with κ = 12.5 K/s observed for off-resonance case. Most earlier studies have reported that κ is much less than ~ 1 K/s for conventional magnetic hyperthermia using a kHz-frequency oscillating field in liquid media 23,24 13 . Table 1 shows the comparison of heating powers SLP and κ obtained under different experimental conditions of AC magnetic fields for a variety of magnetic nanoparticles, as found in the literature. Our observation of such an extremely high value as κ = 93 K/s evidences that the resonant spin-excitation and relaxation mechanism is very promising for establishment of fast local heating for highly efficient hyperthermia treatment with extremely low-power RF magnetic field (e.g., H AC < 5 Oe) and DC magnetic field strengths (H DC < 1 kOe).
For comparison with other frequencies, we plotted κ versus H DC for f AC = 1.5, 2.0, 2.5 and 3.0 GHz, as shown in Fig. 4b (see Supplementary Fig. S7a-c for all of the ΔT − t plots). All of the κ values as a function of H DC exhibited a similar trend but different maximum values of κ = 40, 56, 72, 93 K/s at their corresponding DC fields for resonance, H DC = 321, 428, 535, and 750 Oe for f AC = 1.5, 2.0, 2.5, and 3.0 GHz, respectively. Those peak positions corresponded to the resonance static fields of the given frequency of RF field as previously noted (note that the intrinsic precession frequency, f R = 1.5, 2.0, 2.5, and 3.0, is given for H DC = 321, 428, 535, and 750 Oe, respectively).
Another feature we observed is that the κ values in the higher H DC range converged to a specific low value, about 10 K/s or less. This non-zero, small value of κ even under the off-resonance condition, could be ascribed to a dielectric origin of power loss in magnetic nanoparticles, as evidenced by the further electromagnetic numerical calculations shown in Supplementary Section S5. The results shown in Fig. 4b indicate that very high temperature increment rates can be achievable, and reliably controllable, only by adjusting the parameters of the externally applied RF fields as well as the DC field strength.
Delicate control of local heating by magnetic field parameters. In addition to the controllable amount of heat generated from the Fe 3 O 4 particles by the strength of DC magnetic fields, we also measured the maximum values of κ for different values of f AC and H AC . The maximum κ values for the indicated frequencies were estimated from Fig. 4b, which reveals a linear proportion to f AC , as plotted in Fig. 5a. On the other hand, the κ value as a function of H AC showed a quadratic dependence under the resonance condition  www.nature.com/scientificreports/ with decreasing σ for σ < 0.3 and turned out to be κ = 67 K/s for σ = 0.1 s. For smaller values of σ less than 0.1 s, we could not measure κ, due to the limited temporal resolution (30 Hz) of the IR camera used in our measurement (for the ΔT-t curve for σ = 0.05 s, see Supplementary Fig. S7e). The results in Fig. 5 reveal that the parameters of the externally applied RF fields, such as H AC , f AC , σ as well as H DC , allow for reliably delicate manipulation of κ, thereby enabling well-controllable local heating (temperature increment) around the Fe 3 O 4 particles. Furthermore, the delicate controls of these field parameters are readily available, and also, the strengths of the RF and DC magnetic fields used were relatively very low, as weak as a few Oe for the RF field and a few kOe for the DC field.

Discussion
In summary, we experimentally demonstrated a novel concept of high-efficiency heat generation based on resonant spin-excitation and dissipation, which has never been demonstrated in the context of magnetic hyperthermia. In particular, we observed an unprecedented value of initial temperature increment rate (κ ~ 150 K/s), even with very low power consumption, as small as a few Oe of RF magnetic field. The extremely high value of κ allows for local heating as fast as 0.1 s. Furthermore, this mechanism enables a delicate control of local heating by adjustment of several field parameters, i.e., the strengths of DC/AC magnetic fields, the frequency of the AC field, and its pulse width. Our approach can provide key insights into a new direction toward successful magnetic heat-treatment solutions to overcome the current limitations of conventional magnetic hyperthermia. Its rapid temperature increment would allow a remarkably fast response 31 ; and the dosage of magnetic nanoparticles required for sufficient heat generation/dissipation, therefore, can be much reduced. On the other hand, the frequency of AC magnetic fields for this new technique is higher than the Brezovich limit (f•H = 4.85 × 10 8 A/ m•s) 32 , which is the maximum of the product of AC magnetic field strength and frequency required for human safety from high-frequency field exposure. However, clinically available MRI using RF magnetic field strengths www.nature.com/scientificreports/ of 10-100 μT and frequencies up to 500 MHz (at 11.7 T MRI) 33 give rise to f•H = 4 × 10 10 A/m•s, which far exceeds the Brezovich's acceptable limit. Our proposed technique has f•H < 7 × 10 11 A/m•s, which would be clinically acceptable by using short pulse field exposure and also AC field frequency of less than GHz with larger particle sizes 18 . For successful clinical implementations in the near future, however, it is necessary to conduct in-vitro and in-vivo experimental demonstrations based on resonant spin-excitation and consequent dissipation in magnetic nanoparticles injected into tumors. In the meantime, this work offers a first step towards a paradigmatic shift in magnetic hyperthermia that would make possible successful medical treatment based on utilization of FDAapproved magnetic nanoparticles.

Methods
Measurements of temperature increment from Fe 3 O 4 particles. Using a drop-casting method, the Fe 3 O 4 nanocrystals (about 1 mg) were placed on the surface of a 400-μm-length Cu line in a microstrip consisting of a central waveguide on the top layer, a ground plane on the bottom layer, and a low-loss hydrocarbon/ ceramic substrate (RO4003) between them. To allow spin excitation in the particles, AC currents of different GHz frequencies using a signal generator (E8257D, Agilent) were applied to the microstrip on which the nanoparticles had been placed, and were amplified to several watts by an RF power amplifier (5170FT, Ophir) to generate sufficient strengths of AC magnetic fields around the signal line. The experimental setup also includes an electromagnet for application of DC magnetic fields along the microstrip line. The AC magnetic fields were applied in the direction perpendicular to the DC field direction. An infrared (IR) camera (T650sc, FLIR) was used to measure real-time temperature increments in the samples from thermal radiation at an accuracy of about ± 1 K and a maximum temporal resolution of 30 Hz. The numerical values of temperature were determined by averaging local temperatures in an area of 200 μm × 200 μm at the center of the signal line. In order to read the exact temperature of the magnetic nanoparticles, the parameter values for the calculation are as follows: infrared emissivity of Fe 3 O 4 nanoparticles 34 ε Fe 3 O 4 = 0.97 (very close to emissivity of black body, ε = 1 ), temperature of atmosphere T atm = 25 °C, distance from IR camera to sample D = 50 cm. We also measured the temperatures of ice and boiling water in order to confirm the calibration and accuracy (see Supplementary Section S2).