Abstract
The development of compact and tunable room temperature sources of coherent THzfrequency signals would open a way for numerous new applications. The existing approaches to THzfrequency generation based on superconductor Josephson junctions (JJ), free electron lasers, and quantum cascades require cryogenic temperatures or/and complex setups, preventing the miniaturization and wide use of these devices. We demonstrate theoretically that a bilayer of a heavy metal (Pt) and a biaxial antiferromagnetic (AFM) dielectric (NiO) can be a source of a coherent THz signal. A spincurrent flowing from a DCcurrentdriven Pt layer and polarized along the hard AFM anisotropy axis excites a nonuniform in time precession of magnetizations sublattices in the AFM, due to the presence of a weak easyplane AFM anisotropy. The frequency of the AFM oscillations varies in the range of 0.1–2.0 THz with the driving current in the Pt layer from 10^{8} A/cm^{2} to 10^{9} A/cm^{2}. The THzfrequency signal from the AFM with the amplitude exceeding 1 V/cm is picked up by the inverse spinHall effect in Pt. The operation of a roomtemperature AFM THzfrequency oscillator is similar to that of a cryogenic JJ oscillator, with the energy of the easyplane magnetic anisotropy playing the role of the Josephson energy.
Introduction
An absence of compact and reliable generators and receivers of coherent signals in the frequency range 0.1–10 THz has been identified as a fundamental physical and technological problem^{1,2,3}. The existing approaches to THzfrequency generation, including superconductor Josephson junctions (JJ)^{4}, free electron lasers^{5}, and quantum cascades^{6} require complex setups, which limit wide use of these devices. At the same time, it has been demonstrated that ferromagnetic (FM) layered structures driven by a spintransfer torque (STT) created by a DC spin current^{7,8}, which compensates magnetic damping, can be used as spintorque or/and spinHall autooscillators in the frequency range of 1–30 GHz^{9,10,11,12,13,14,15}.
In order to increase the generation frequency it was proposed to use antiferromagnets (AFM) rather than FM films as active layers of spintronic autooscillators^{16,17}. Unfortunately, the traditional method of the STTinduced damping compensation in FM does not work for AFM. To compensate damping in a FM, the DC spin current must be polarized parallel to the direction of the static equilibrium magnetization. However, since AFMs have two magnetic sublattices with opposite magnetizations, the STT decreasing the damping in one of the sublattices increases it in the other sublattice, thus resulting in a zero net effect. Fortunately, the presence in an AFM of two magnetic sublattices coupled by a strong exchange interaction qualitatively changes the magnetization dynamics of AFM^{18}. In particular, it has been shown, that, in contrast with a FM, the STT acting on an AFM can lead to a dynamic instability in the magnetic sublattice orientation^{3,16,19,20}, which results in the rotation of the magnetizations of the AFM sublattices in the plane perpendicular to the direction of polarization of the applied spin current^{16,19,20}. This mechanism has been already used to experimentally switch the orientation of magnetic sublattices in AFM materials^{21,22}. However, the STTinduced rotation of the magnetic sublattices in an AFM has not been recognized so far as a possible mechanism of realization of THzfrequency AFM oscillators, since in a magnetically compensated AFM the steady rotation of sublattices does not create any AC spincurrent.
In this work we demonstrate theoretically that a simple structure consisting of a metallic layer with a strong spinorbit interaction (such as Pt) and a layer of a biaxial antiferromagnetic (AFM) dielectric (such as NiO) can be a base of a tunable roomtemperature THzfrequency signal generator. A DC spin current flowing from a currentdriven Pt layer and polarized along the hard anisotropy axis of the adjacent AFM layer can excite a rotation of the AFM sublattice magnetizations^{16,19,20} that is nonuniform in time due to the influence of a weak easyplane AFM anisotropy. This nonuniform rotation results in the THzfrequency spinpumping back into the Pt layer, creating an AC spin current that can be detected using the inverse spinHall effect. The generated signal amplitudes of 1 V/cm in the frequency range of 0.1–2.0 THz of the can be achieved for driving DC current densities of 1 × 10^{8} A/cm^{2} to 1.1 × 10^{9} A/cm^{2}, that have been experimentally achieved previously in spinHall nanooscillators^{12,13,14,15}.
We also demonstrate, that the equations describing the operation of the proposed roomtemperature AFMbased oscillator are mathematically analogous to the ones describing the oscillators based on superconducting Josephson junctions (JJ)^{23}, with the energy of the easyplane magnetic anisotropy playing the role of the Josephson energy. Consequently, a number of effects studied previously in JJ oscillators at cryogenic temperatures can also be observed in the roomtemperature AFM oscillators. In particular, the inertial nature of the AFM dynamics^{18} leads to the hysteretic behavior of the AFM oscillator, which, therefore, have two different current thresholds: an “ignition” threshold, which is required to start the generation, and a lower “elimination” threshold which, in our case, is twice less then the “ignition” threshold, defining the minimum current density needed to support the generation, once it has been started.
Results
We consider a bilayer consisting of a layer of a heavy metal with strong spinorbital interaction (e.g., Pt) adjacent to a biaxial AFM layer (e.g., NiO), see Fig. 1. A DC electric current passing through the Pt layer creates, via the spinHall effect, a perpendicularlypolarized spin current flowing into the AFM layer^{24,25,26}. Spin current creates a nonconservative spintransfer torque (STT) on AFM sublattice magnetizations M_{j} (j = 1, 2): τ_{STT} = (τ/M_{s}) M_{j} × ( M_{j} × p )^{7,19,27}, where p is the direction of the spin current polarization, M_{s} =  M_{j} is the static magnetization of a sublattice, and τ is the amplitude of the spin current in the units of frequency^{28}. If the spin current is polarized perpendicularly to the AFM ground state (along the “hard” AFM axis; here we consider only such a configuration), it tilts the magnetizations from their equilibrium opposite orientation M_{1} = − M_{2}, which creates a strong effective field H = H_{ex}( M_{1} + M_{2})/(2M_{s}) (H_{ex} ~ 10^{3} T is the exchange field) leading to uniform rotation (in the absence of the inplane anisotropy) of the sublattice magnetizations in the plane perpendicular to the spincurrent polarization^{16,19,20}, see Fig. 1.
The rotation of the tilted sublattice magnetizations in an AFM induces the spin current flowing back, from the AFM to the Pt layer, via the spinpumping mechanism^{16,28,29}:
where g_{r} is the spinmixing conductance.
The exchange interaction is the strongest interaction in AFM, and, even under the action of an STT, the tilt angle of the sublattice magnetizations is small. Then, introducing the AFM Neel vector^{18,30} l = ( M_{1} − M_{2})/2M_{s} the spinpumping current (1) can be written as:
where ϕ is the azimuthal angle of the vector l . Equation (2) leads to an important conclusion: uniform rotation (with constant angular velocity ) of the Neel vector creates only a DC spin pumping signal in Pt, and creates no AC signal^{19,20}.
Equation (2) is fully analogous to the second Josephson equation connecting voltage bias in a Josephson junction (JJ) with the phase of the supercurrent^{23,31}. Similarly to JJ oscillators, to achieve the AC generation we need one more ingredient: a potential “force” that depends on ϕ. In Josephson junctions this potential comes from the tunneling Hamiltonian (or Josephson energy)^{23}. In AFM, the role of the Josephson energy is played by the energy of crystalline magnetic anisotropy W_{a} in the easy plane: , where H_{e} is the easyplane anisotropy field.
The presence of the anisotropy leads to a qualitative change in the dynamics of the magnetic sublattices: it creates an additional conservative torque τ_{a}(ϕ), which depends on the orientation of the magnetic sublattices, see Fig. 2(a,b). Thus the trajectory of each sublattice magnetization is not anymore a planar circle on the sphere’s equator^{16,17}, but is a more complicated curve (see Fig. 2(a,b)). In one part of the trajectory the anisotropy decreases the tilt angle between the magnetizations M_{1} and M_{2}, thus decelerating the rotation (Fig. 2(a)), while in another part the anisotropy increases the tilt and, therefore, accelerates the rotation (Fig. 2(b)).
At the same time, the trajectory of the Neel vector l remains a circle even in the presence of easyplane anisotropy (see Methods for the derivation details), and can be described with one scalar equation for the angle ϕ^{17,19,32}:
where ω_{e} = γH_{e}, γ is the gyromagnetic ratio, α is the effective Gilbert damping parameter, j is the electric current density in the Pt layer, and σ = τ/j is the torquecurrent proportionally coefficient^{16,19}. This equation is, essentially, a condition of the balance between four torques acting on the sublattice magnetizations: the exchange torque, created by the exchange field, the Gilbert damping torque, the anisotropy torque, and the STT.
Equation (3) is a wellknown equation describing dynamics of a massive particle in a tilted “washboard” potential. It also exactly coincides with the equation describing the superconducting phase in a resistively and capacitively shunted JJ under a current bias^{33}. Here, the role of the energy stored in a capacitor is played by the exchange energy, which accumulates the kinetic energy in the system, and the role of resistance is played by the Gilbert damping. For sufficiently large currents, equation (3) does not have stationary solutions, meaning that the AFM magnetization sublattices lose their stability, and the Neel vector starts to rotate. The electric current density needed to overcome this “ignition” threshold is proportional to the easyplane anisotropy:
What is more important, the rotation of the Neel vector, due to the influence of the easyplane magnetic anisotropy, is not uniform in time, . To illustrate this effect, we solved numerically a system of two coupled LandauLifshits equations describing the AFM dynamics for a bilayer NiO (5 nm)/Pt (20 nm) (see Methods for other calculation parameters). The time dependence of the azimuthal angle ϕ after a sudden application of the DC current is plotted in Fig. 3(a). The angle ϕ infinitely increases in time as the system makes revolutions around the direction of the spin current polarization^{20}, but its motion is not uniform in time due to the action of the easyplane anisotropy. This nonuniformity is further illustrated by the time dependence of the angular velocity , which oscillates in time with THz frequency (see Fig. 3(b)). Thus, the (easyplane) AFM magnetic anisotropy, existing in the plane perpendicular to the spin current polarization, leads to the generation of an AC output spincurrent (2), and turns a currentdriven Pt/AFM bilayer into a potential THzfrequency autooscillator.
Another important feature, which is evident from Fig. 3(b), is the existence of a transitional process from the stationary state to a steady oscillatory motion. In contrast to FM materials, the dynamics of AFM is inertial^{18,34}, so that it takes time for the STT to accelerate the sublattice magnetizations, and the magnetic system accumulates some kinetic energy during this transitional process. The inertial nature of the AFM dynamics implies that the AFM oscillator may exhibit a hysteretic behavior, and that two different threshold currents may exist in the AFM oscillator.
This process is illustrated by Fig. 4, which shows the dynamics of the AFM angle ϕ after application of electric current. In the first case (black curves) the current, first, overcomes the “ignition” threshold , and then is lowered to the “working” density, which is below . In the second case (red lines) the current is increased from zero directly to the “working” density. In the first case the oscillations continue even after the current was lowered below the “ignition” level, while in the second case the same “working” current cannot start any oscillations. The existence of the hysteresis in the AFM dynamics allows one to “ignite” the generation with a very short (<10 ps) largeamplitude current pulse, and then to reduce the current density for a continuous oscillator operation. The minimum current needed to sustain AFM oscillations (the “elimination” threshold) can be found analytically in the limit of a small damping from the condition that the work produced by the STT during one period of rotation equals the energy lost due to the Gilbert damping:
The existence of these two threshold currents is critically important for a practical implementation of the proposed oscillator. For the taken parameters of the bilayer the ignition current density is estimated as , which is larger than the working current for the already demonstrated FM spinHall oscillators^{13}, however, the elimination threshold current density is twice lower , which is lower, than the threshold current densities for the FM spinHall oscillators. We also note, that similar hysteretic transitions could occur in ferromagnetic spintorque nanooscillators, where a spin transfer torque provided by a spin current polarized in a certain particular direction can alter the energy landscape of a magnetic subsystem^{35,36}.
Equation (3) allows one to find an approximate analytical solution for supercritical currents :
where ω_{gen} is the AC generation frequency:
It is clear, that the output AC signal (2), proportional to the variable part of the angular velocity , depends on to the anisotropy field H_{e} in the AFM easy plane, and vanishes for an uniaxial AFM. Also the generation frequency ω_{gen} does not depend on the AFM resonance frequencies, but is determined only by the ratio of STT and the Gilbert damping. Similarly, in JJs, the generation frequency depends on current and not on the Josephson plasma frequency.
Figure 5(a) shows the dependence of the generated frequency ω_{gen} on the current density in the Pt layer, while Fig. 5(b) – dependence of the amplitudes of several harmonics of the output AC electric field as a function of the generated frequency (see Methods for the parameters of the oscillator). The frequency ω_{gen} can be continuously tuned from almost 0 to several THz with current densities j ~ 10^{8} A/cm^{2}. The first harmonic of the output AC electric field has a maximum at a relatively low frequency ≃0.1 THz, and, then, slowly decreases, remaining sufficiently large (>1 V/cm) up to the frequency of 2 THz. For small frequencies (small currents) the motion of the phase ϕ(t) is strongly nonlinear, and the output signal contains multiple higher harmonics, the amplitudes of which decrease rather fast with the increase of the generation frequency.
The linewidth of the generated output signal is an important characteristic of the AFM autooscillator (AFMO), determining the spectral purity and the phase noise level of the device. The generation linewidth of an autooscillator is mainly determined by the fluctuations of phase of the generated signal^{37}. In our model these phase fluctuations were taken into account by adding a stochastic term f(t) to the righthand side of (3). For simplicity, we assumed that the generated frequency ω_{gen} is substantially larger than the characteristic frequency of the antiferromagnetic resonance in the used antiferromagnet: . Under this assumption we were able to omit in (3) the term containing the inplane anisotropy of the AFM, and to solve the obtained stochastic equation:
for the case of a random walk of phase^{38}. The generation linewidth of the AFMO calculated by the above described method from (8) can be expressed as:
where V_{AFM} is the AFM volume, k_{B} is the Boltzmann constant and T is the absolute temperature of the AFMO. For an AFMO at room temperature and having the lateral dimensions of L = 10 μm (see below) the calculated linewidth is Δω/(2π) = 0.36 MHz, which corresponds to the effective quality factor (Qfactor) Q ≈ 3 × 10^{6} at the generation frequnecy ω_{gen}/(2π) = 1 THz.
The calculated Qfactor of the AFMO is substantially larger than the quality factor (Q = 3 × 10^{3} at the generation frequency of 40 GHz) of a typical spintorque nanooscillator (STNO) based on FM materials (see e.g. refs 11,37). The small value of the generation linewidth of the AFMO is related, mainly, to the large energy E_{0} of the excited AFM oscillations (E_{0}/(k_{B}T) = 10^{7} at ω_{gen}/(2\π) = 1 THz, see Eq. (95) in ref. 37 for details). This is happening for three reasons: (a) the AFMO has a larger volume of the active medium (AFM layer); (b) a maximum 90° precession angle; and (c) an order of magnitude larger precession frequency, compared to a typical FMbased STNO.
There are two other interesting peculiarities of the linewidth expression (9). First, the AFMO linewidth Δω is independent of the generation frequency (in the limit ), which is caused by the fact, that the average motion and fluctuations of phase, described by the equation (8), are independent. Secondly, contrary to naïve expectations, the linewidth (9) is inversely of proportional to the Gilbert damping constant α. Such a behavior of the AFMO linewidth in the limit , also, can be easily explained. Equation (8) describes a forced drag in a viscous medium, and with a larger damping (viscosity) each random “kick” will lead, on the average, to a smaller deviation of the phase, and, consequently, to a smaller generation linewidth.
Discussion
Above, we demonstrated that a thin layer of a biaxial AFM (e.g., NiO) driven by a DC spin current flowing in the adjacent Pt layer can work as a THzfrequency autooscillator. Below, we compare the proposed AFMO to other known sources of coherent microwave or THzfrequency radiation.
A working prototype of a spinHall nanooscillator (SHNO) based on a bilayer of a FM metal and Pt^{12,13,14} had a threshold current density of j ≈ 1.3 × 10^{8} A/cm^{2}, which is smaller than the ignition threshold current , however, this ignition current is needed for only a very short amount of time and can be reduced to lesser current densities above the elimination threshold , making the proposed AFM/Pt configuration promising for the practical implementation.
The frequency of the SHNO is determined mainly by the bias magnetic field and static magnetization of the FM layer, and was of the order of 5–12 GHz for the bias fields ranging between 400 Oe and 2000 Oe. In the proposed AFMO the generation frequency is determined by the driving electric current, and can be varied from 0.1 THz to 2.5 THz for experimentally achievable current densities (note, that the current densities of up to 1.1 × 10^{9} A/cm^{2} have been achieved in Py/Pt nanowires^{14}).
In ref. 17 it was proposed to achieve the THzfrequency generation in NiO/Pt bilayer via a nonlinear feedback mechanism, which is different from the AFMO generation mechanism described above. In ref. 17 the spincurrent is polarized along the easy axis of the AFM (easy plane anisotropy field in NiO H_{e} ≈ 628 Oe^{39,40}), which leads to a rather large threshold current density, because the currentinduced STT has to overcome both the large hard axis anisotropy (hard axis anisotropy field in NiO H_{h} ≈ 15.7 kOe^{39}) and damping. For the parameters used in this work, we estimate this current density (see Methods) as , which is more then H_{h}/H_{e} ≈ 25 times larger than the ignition threshold current for the mechanism proposed here, and, ≈50 times larger then the elimination threshold. Also, the nonlinear feedback mechanism, which stabilizes the AFM precession around the easy axis in ref. 17, works only in a very narrow range of the bias currents and generation frequencies, which severely limits the tunability of the oscillator proposed in ref. 17.
One of the most important characteristics of an oscillator is its output power, which depends on the oscillator’s physical dimensions. The devices working at the frequencies of up to several THz must be smaller than the wavelength λ_{EM} of the electromagnetic radiation at these frequencies. To estimate the output power for the above proposed AFMO we assumed here that the working area of the AFMO has the characteristic dimension L = 10 μm ≪ λ_{EM}@1.5 THz ≈ 200 μm. For such a size the output power can be estimated as W = E^{2}ρ^{−1}L^{2}d_{Pt} (see Methods for parameters). Thus, for the proposed AFMO, the output voltage varies from 6 mV to 1 mV in the frequency range from 0.1 THz to 2.0 THz, which gives the power range from 1.5 μW to 40 nW.
The known oscillators working in the THzfrequency range are the currentbiased Josephson junction oscillators working at cryogenic temperatures and having a similar power range^{41}, but a much smaller output voltage (16 μV in ref. 42). The amplitude of the output voltage (and, consequently, the output power) of the above proposed AFMO working at room temperature depends on the magnitude of the inplane anisotropy in the AFM layer (see (6)). To increase the output voltage at higher generation frequencies one can use AFM materials with a stronger easyplane anisotropy, but this, obviously, will lead to a corresponding increase in the “ignition” threshold current density.
In the above presented calculations it was assumed that the driving spin current is polarized exactly along the hard anisotropy axis of the AFM, and it might be rather challenging to achieve in experiment that the AFM hard axis lies exactly in the plane of the interface between AFM and Pt, containing the direction of the spin polarization. A numerical solution of (11), however, reveals, that a finite misalignment angle Ψ ≲ 60° between the direction of the spincurrent polarization vector p and the AFM hard axis n_{h} does not destroy the generation effect, but only leads to an increase of the ignition threshold current as . Thus, the proposed THz generation mechanism can be realized even if the AFM hard axis does not exactly coincide with the plane of the AFM/Pt interface.
We also note, that although we took NiO as a model AFM in the above presented calculation, the proposed mechanism of the THzfrequency generation can be realized in other AFM materials, as long as the AFM material has a sufficiently strong easy plane anisotropy (i.e. a definite plane where the vector of antiferromagnetism l lies), and a weak additional anisotropy inside this easyplane.
In conclusion, we demonstrated theoretically, that a pure spin current in the AFM layer induced by a DC driving electric current flowing in the adjacent Pt layer can excite THzfrequency oscillations. We showed, that in the case of a NiO (5 nm) Pt (20 nm) bilayer it is possible to achieve the generation of 0.1–2.0 THz signals with reasonable current densities that were previously achieved in FM SHNO of a similar geometry. The estimated AC voltage of the proposed AFMO, which can be picked up using the inverse spinHall effect in the Pt layer can exceed 1 V/cm for the generation frequencies of up to 2 THz.
Methods
Spintransfer torque
In this work τ is the amplitude of the STT expressed in the units of frequency^{28}:
where j is the density of the driving DC electric current in the Pt layer, g_{r} is the spinmixing conductance at the PtAFM interface, λ is the spindiffusion length in the Pt, ρ is the Pt electric resistivity, M_{s} is the saturation magnetization of one of the AFM sublattices, and d_{AFM} and d_{Pt} are the thicknesses of the AFM and Pt layers, respectively.
Coupled LandauLifshitz equations
Dynamics of a thin antiferromagnetic film can be numerically simulated using two coupled LandauLifshitzGilbertSlonczewski equations^{7,8,16,43} for two magnetic sublattices M_{1} and M_{2} of the AFM:
where α is the effective Gilbert damping parameter, τ is the STT expressed in the frequency units (see equation (10)), p is a unit vector along the spin current polarization, and H_{1} and H_{2} are the effective magnetic fields acting on the sublattices M_{1} and M_{2}, respectively:
Here H_{ex} is the exchange field, H_{e} and H_{h} are the easyplane and hardaxis anisotropy fields, respectively, and n_{e} and n_{h} are the unit vectors along the hard and easy axes.
Antiferromagnetic dynamics
To study the dynamics of AFM analytically we use the standard σmodel^{18,30}. The coupled LandauLifshits equations (11) can be rewritten in terms of the l and m . Assuming that  m  ≪  l , which is valid when the exchange field H_{ex} is larger than any other field acting in the AFM, one can consider m as a slave variable:
In this approximation the dynamics of l is governed by one secondorder vectorial differential equation:
with an additional constraint  l  = 1. Here , and the symbol ⊗ denotes the direct vector product. Equation (14) is effectively twodimensional, so we can rewrite it in a spherical coordinate system. To simplify the analytical derivation we assume that = p (i.e., the spin current is polarized along the axis), n_{e} = , and n_{h} = :
where ω_{e} = γH_{e} and ω_{h} = γH_{h}. The ground state of the AFM corresponds to θ = π/2 and ϕ = 0. The solution θ = π/2 (i.e., vector l rotates in the xyplane) is stable for the considered geometry^{17,19} and automatically satisfies the equation (16). Using θ = π/2 in equation (15) it is possible to obtain a single equation (3) for the azimuthal angle ϕ.
Parameters of the system
In all the numerical simulations and estimations reported here we considered a Pt/NiO bilayer with the following parameters: thickness of the NiO and Pt layers d_{AFM} = 5 nm and d_{Pt} = 20 nm, spin diffusion length in Pt λ = 7.3 nm^{44}, electrical resistivity in Pt ρ = 4.8 × 10^{−7} Ω · m^{44}, spinmixing conductance at the PtNiO interface g_{r} = 6.9 × 10^{18} m^{−2}^{32}, magnetic saturation of one NiO sublattice M_{s} = 351 kA/m^{40}, spinHall angle in Pt θ_{SH} = 0.1^{44}, effective Gilbert damping is α = 3.5 × 10^{−3} (see below), exchange frequency ω_{ex} = 2π × 27.5 THz, γH_{e} = ω_{e} = 2π × 1.75 GHz and γH_{h} = ω_{h} = 2π × 43.9 GHz^{39}. For the chosen parameters the coefficient σ in (10) is σ/(2π) = 4.32 × 10^{−4} Hzm^{2}/A.
Output electric field
The rotation of the vector l in the AFM layer induces a spincurrent into the adjacent Pt layer, which, in turn, creates an electric field in the Pt layer via the inverse spinHall effect (ISHE). This AC electric field serves as the output signal of the AFMO. The ISHE electric field is calculated using the following analytic expression^{29}:
For the chosen parameters of the AFM oscillator the parameter κ is equal to κ ≈ 1.35 × 10^{−9} V/m ⋅ (rad/s)^{−1}.
Effective Gilbert damping
The intrinsic Gilbert damping constant α_{0} for NiO can be calculated from the experimentally measured linewidth Δω_{AFMR}/(2π) = 18 GHz of the AFM resonance^{39,45}. The linewidth Δω_{AFMR} is related to α_{0} by Δω_{AFMR} = α_{0}ω_{ex}, where ω_{ex} = γH_{ex} = 2π ⋅ 27.5 THz is the exchange frequency^{40}. One can see, that the intrinsic Gilbert damping in NiO is rather small: . However, the spin pumping from NiO to Pt layer can be described as an additional damping mechanism for the spin dynamics in the AFM^{28,29}, and the total effective damping constant can be written as:
For the chosen parameters of the AFMO the damping parameter is α = 3.5 × 10^{−3}. Thus, the effective damping in thin Pt/NiO bilayers is dominated by the spinpumping mechanism, and strongly depends on the NiO thickness.
Threshold current in the case when the spin current is polarized along the easy axis
Cheng et al. estimated the threshold STT needed to start the oscillations in the case when the spin current is polarized along the easy axis as (see eq. (3) in ref. 17):
For the parameters of our AFMO, the threshold electric current for the generation mechanism described in ref. 17 can be calculated as:
Additional Information
How to cite this article: Khymyn, R. et al. Antiferromagnetic THzfrequency Josephsonlike Oscillator Driven by Spin Current. Sci. Rep. 7, 43705; doi: 10.1038/srep43705 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
 1
Sirtori, C. Applied physics: Bridge for the terahertz gap. Nature 417, 132–133 (2002).
 2
Kleiner, R. Filling the terahertz gap. Science 318, 1254–1255 (2007).
 3
Gulyaev, Y. V., Zilberman, P. E., Mikhailov, G. M. & Chigarev, S. G. Generation of terahertz waves by a current in magnetic junctions. JETP Lett. 98, 742–752 (2014).
 4
Ozyuzer, L. et al. Emission of coherent THz radiation from superconductors. Science 318, 1291–1293 (2007).
 5
Nanni, E. A. et al. Terahertzdriven linear electron acceleration. Nat. Commun. 6, 8486 (2015).
 6
Hubers, H.W. Terahertz technology: Towards THz integrated photonics. Nat. Photon. 4, 503–504 (2010).
 7
Slonczewski, J. Currentdriven excitation of magnetic multilayers. J. Magn. Magn. Mater. 159, L1–L7 (1996).
 8
Berger, L. Emission of spin waves by a magnetic multilayer traversed by a current. Phys. Rev. B 54, 9353–9358 (1996).
 9
Tsoi, M. et al. Generation and detection of phasecoherent currentdriven magnons in magnetic multilayers. Nature 406, 46–48 (2000).
 10
Kiselev, S. I. et al. Microwave oscillations of a nanomagnet driven by a spinpolarized current. Nature 425, 380–383 (2003).
 11
Rippard, W., Pufall, M., Kaka, S., Russek, S. & Silva, T. Directcurrent induced dynamics in Co90Fe10/Ni80Fe20 point contacts. Phys. Rev. Lett. 92, 027201 (2004).
 12
Demidov, V. E. et al. Magnetic nanooscillator driven by pure spin current. Nat. Mater. 11, 1028–1031 (2012).
 13
Demidov, V. E., Urazhdin, S., Zholud, A., Sadovnikov, A. V. & Demokritov, S. O. Nanoconstrictionbased spinHall nanooscillator. Appl. Phys. Lett. 105, 172410 (2014).
 14
Duan, Z. et al. Nanowire spin torque oscillator driven by spin orbit torques. Nat. Commun. 5, 5616 (2014).
 15
Collet, M. et al. Generation of coherent spinwave modes in yttrium iron garnet microdiscs by spin–orbit torque. Nat. Commun. 7, 10377 (2016).
 16
Gomonay, E. V. & Loktev, V. M. Spintronics of antiferromagnetic systems (review article). Low Temp. Phys. 40, 17–35 (2014).
 17
Cheng, R., Xiao, D. & Brataas, A. Terahertz antiferromagnetic spin Hall nanooscillator. Phys. Rev. Lett. 116, 207603 (2016).
 18
Ivanov, B. A. Spin dynamics of antiferromagnets under action of femtosecond laser pulses (review article). Low Temp. Phys. 40, 91–105 (2014).
 19
Gomonay, H. V. & Loktev, V. M. Spin transfer and currentinduced switching in antiferromagnets. Phys. Rev. B 81, 144427 (2010).
 20
Cheng, R., Daniels, M. W., Zhu, J.G. & Xiao, D. Ultrafast switching of antiferromagnets via spintransfer torque. Phys. Rev. B 91, 064423 (2015).
 21
Wadley, P. et al. Electrical switching of an antiferromagnet. Science 351, 587–590 (2016).
 22
Kriegner, D. et al. Multiplestable anisotropic magnetoresistance memory in antiferromagnetic MnTe. Nat. Commun. 7, 11623 (2016).
 23
Barone, A. & Paterno, G. Physics and Applications of the Josephson Effect (WileyVCH, 1982).
 24
Jungwirth, T., Marti, X., Wadley, P. & Wunderlich, J. Antiferromagnetic spintronics. Nat. Nanotechnol. 11, 231–241 (2016).
 25
Moriyama, T. et al. Antidamping spin transfer torque through epitaxial nickel oxide. Appl. Phys. Lett. 106, 162406 (2015).
 26
Daniels, M. W., Guo, W., Stocks, G. M., Xiao, D. & Xiao, J. Spintransfer torque induced spin waves in antiferromagnetic insulators. New J. Phys. 17, 103039 (2015).
 27
Linder, J. Controllable spintransfer torque on an antiferromagnet in a dual spinvalve. Phys. Rev. B 84, 094404 (2011).
 28
Tserkovnyak, Y., Brataas, A., Bauer, G. E. W. & Halperin, B. I. Nonlocal magnetization dynamics in ferromagnetic heterostructures. Rev. Mod. Phys. 77, 1375–1421 (2005).
 29
Nakayama, H. et al. Geometry dependence on inverse spin Hall effect induced by spin pumping in Ni81Fe19/Pt films. Phys. Rev. B 85, 144408 (2012).
 30
Andreev, A. F. & Marchenko, V. I. Symmetry and the macroscopic dynamics of magnetic materials. Sov. Phys. Usp. 23, 21–34 (1980).
 31
Savel’ev, S., Yampol’skii, V. A., Rakhmanov, A. L. & Nori, F. Terahertz Josephson plasma waves in layered superconductors: spectrum, generation, nonlinear and quantum phenomena. Rep. Prog. Phys. 73, 026501 (2010).
 32
Cheng, R., Xiao, J., Niu, Q. & Brataas, A. Spin pumping and spintransfer torques in antiferromagnets. Phys. Rev. Lett. 113, 057601 (2014).
 33
Stewart, W. C. Currentvoltage characteristics of Josephson junctions. Appl. Phys. Lett. 12, 277 (1968).
 34
Kimel, A. V. et al. Inertiadriven spin switching in antiferromagnets. Nat. Phys. 5, 727–731 (2009).
 35
Pinna, D., Stein, D. L. & Kent, A. D. Spintorque oscillators with thermal noise: A constant energy orbit approach. Phys. Rev. B 90 (2014).
 36
Ye, L., Wolf, G., Pinna, D., ChavesO’Flynn, G. D. & Kent, A. D. State diagram of an orthogonal spin transfer spin valve device. J. Appl. Phys. 117, 193902 (2015).
 37
Slavin, A. & Tiberkevich, V. Nonlinear autooscillator theory of microwave generation by spinpolarized current. IEEE Transactions on Magnetics 45, 1875–1918 (2009).
 38
Gomonay & Loktev. Spin torque antiferromagnetic nanooscillator in the presence of magnetic noise. Condens. Matter Phys. 15, 43703 (2012).
 39
Sievers, A. J. & Tinkham, M. Far infrared antiferromagnetic resonance in MnO and NiO. Phys. Rev. 129, 1566–1571 (1963).
 40
Hutchings, M. T. & Samuelsen, E. J. Measurement of spinwave dispersion in NiO by inelastic neutron scattering and its relation to magnetic properties. Phys. Rev. B 6, 3447–3461 (1972).
 41
Darula, M., Doderer, T. & Beuven, S. Millimetre and submm wavelength radiation sources based on discrete Josephson junction arrays. Supercond. Sci. Technol. 12, R1–R25 (1999).
 42
Ulrich, B. & Kluth, E. Josephson junction millimeter microwave source and homodyne detector. Proc. IEEE 61, 51–54 (1973).
 43
Gurevich, A. G. & Melkov, G. A. Magnetization Oscillations and Waves (CRC Press, NY, 1996).
 44
Wang, H. L. et al. Scaling of spin Hall angle in 3d, 4d, and 5d metals from Y3Fe5O12/metal spin pumping. Phys. Rev. Lett. 112, 197201 (2014).
 45
Nishitani, J., Kozuki, K., Nagashima, T. & Hangyo, M. Terahertz radiation from coherent antiferromagnetic magnons excited by femtosecond laser pulses. Appl. Phys. Lett. 96, 221906 (2010).
Acknowledgements
This work was supported in part by Grant No. ECCS 1305586 from the National Science Foundation of the USA, by the contract from the US Army TARDEC, RDECOM, and by the Center for NanoFerroic Devices (CNFD) and the Nanoelectronics Research Initiative (NRI). BAI was partly supported by the National Academy of Sciences of Ukraine, project No 1/16N. IL was partly supported by the Russian Scientific Foundation under the Grant #141900760.
Author information
Affiliations
Contributions
R.K., I.L. and V.T. developed an analytical model of AFM excitation under the action of STT. R.K. performed the numerical simulations. A.S. and B.A.I. formulated the problem and supervised the work. All the authors contributed to the text.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Khymyn, R., Lisenkov, I., Tiberkevich, V. et al. Antiferromagnetic THzfrequency Josephsonlike Oscillator Driven by Spin Current. Sci Rep 7, 43705 (2017). https://doi.org/10.1038/srep43705
Received:
Accepted:
Published:
Further reading

Observation of currentinduced switching in noncollinear antiferromagnetic IrMn3 by differential voltage measurements
Nature Communications (2021)

Butterflyshaped magnetoresistance in triangularlattice antiferromagnet Ag2CrO2
Scientific Reports (2020)

Electrical manipulation of the magnetic order in antiferromagnetic PtMn pillars
Nature Electronics (2020)

Chiralinduced switching of antiferromagnet spins in a confined nanowire
Communications Physics (2019)

How a ferromagnet drives an antiferromagnet in exchange biased CoO/Fe(110) bilayers
Scientific Reports (2019)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.