- Email: [email protected]
Evolutionary significance and regulated expression of Tdrd family genes in gynogenetic Japanese flounder (Paralichthys olivaceus)
Nanoscale spin-wave wake-up receiver Cite as: Appl. Phys. Lett. 115, 092401 (2019); https://doi.org/10.1063/1.5109623 Submitted: 10 May 2019 . Accepted: 10 August 2019 . Published Online: 26 August 2019 Q. Wang
, T. Brächer, M. Mohseni
, B. Hillebrands, V. I. Vasyuchka
, A. V. Chumak
, and P. Pirro
ARTICLES YOU MAY BE INTERESTED IN Enhancement of spin–orbit torque via interfacial hydrogen and oxygen ion manipulation Applied Physics Letters 115, 092402 (2019); https://doi.org/10.1063/1.5110206 Direct observation of magneto-Peltier effect in current-in-plane giant magnetoresistive spin valve Applied Physics Letters 115, 092406 (2019); https://doi.org/10.1063/1.5120569 Role of interface quality for the spin Hall magnetoresistance in nickel ferrite thin films with bulk-like magnetic properties Applied Physics Letters 115, 092403 (2019); https://doi.org/10.1063/1.5097600
Appl. Phys. Lett. 115, 092401 (2019); https://doi.org/10.1063/1.5109623 © 2019 Author(s).
115, 092401
Applied Physics Letters
ARTICLE
scitation.org/journal/apl
Nanoscale spin-wave wake-up receiver Cite as: Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Submitted: 10 May 2019 . Accepted: 10 August 2019 . Published Online: 26 August 2019 Q. Wang,a)
€ cher, M. Mohseni, T. Bra
B. Hillebrands, V. I. Vasyuchka,
A. V. Chumak,
and P. Pirro
AFFILIATIONS € t Kaiserslautern, 67663 Kaiserslautern, Germany Fachbereich Physik and Landesforschungszentrum OPTIMAS, Technische Universita a)
Author to whom correspondence should be addressed: [email protected]
ABSTRACT We present the concept of a passive spin-wave device which is able to distinguish different radio frequency pulse trains and validate its functionality using micromagnetic simulations. The information is coded in the phase of the individual pulses which are transformed into spinwave packets. The device splits every incoming packet into two arms, one of which is coupled to a magnonic ring which introduces a well-defined time delay and phase shift. Since the time delay is matched to the pulse repetition rate, adjacent packets interfere in a combiner, which makes it possible to distinguish simple pulse train patterns by the readout of the time-integrated spin-wave intensity in the output. Due to its passive construction, this device may serve as an energy-efficient wake-up receiver used to activate the main receiver circuit in power critical internet of things applications. Published under license by AIP Publishing. https://doi.org/10.1063/1.5109623
An ultralow power wireless communication capability is needed to realize wireless sensor networks in real-world scenarios. The most efficient way to reduce energy consumption is to leave the device off and wake it up asynchronously only when it is really needed. For this task, one needs a wake-up receiver which continuously monitors a channel, listens for a wake-up signal from other nodes, and activates the device on stimulus.1 Therefore, the power consumption as well as the footprint of the wake-up receiver is of crucial importance. Spin waves, a collective precessional motion of localized magnetic moments, can propagate at gigahertz frequencies in either conducting or insulating magnetic media with wavelength down to tens of nanometers and without any flow of electrons, reducing the energy consumption caused by Joule heating.2–6 In general, spin waves, and their quanta magnons, are considered as potential candidates as carriers of information in future devices due to their low losses,7–10 short wavelengths,11–15 high frequencies,16,17 and abundant nonlinear phenomena.18–20 Another advantage of spin waves is that due to their wave nature, they can be used for transfer and processing of information, which is coded in their amplitude and phase. Interference effects between different waves play a major role in harvesting the full potential of the wave nature for data processing. A prominent example for the advantages of wave-based data processing is the majority gate, whose output state represents the majority of input states,21,22 and which is an especially compact and readily scalable wave-based device. In this letter, we design a magnonic wake-up receiver based on spin-wave interference. The basic concept of the magnonic wake-up receiver consists of a Y-shaped splitter, a combiner, and a coupled
Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Published under license by AIP Publishing
magnonic circular ring structure, as depicted in Fig. 1(a). The device works in remanence. Radio frequency (RF) pulse trains from other wireless nodes will be converted into spin-wave pulse trains by a magnonic transducer in the input.23–25 The blue and red arrows show the flow path of the spin-wave packets in the magnonic wake-up receiver, which are split into two equal parts by a symmetric Y-shaped splitter.26 One of them is directly guided to the combiner region by the upper waveguide (blue channel). The other one is first transferred to the adjacent magnonic circular ring structure by dipolar coupling5,27 (the red dashed area), propagates along the ring, and then couples back to the waveguide with an additional time delay Dt and phase shift Du (red channel). The time delay and phase shift can be controlled by adjusting the delay length Ld [see Fig. 1(b)]. When the time delay Dt is equal to the pulse period sp of the spin-wave pulse train, the back coupled spin wave in the red channel will completely interfere with the next packet of the spin-wave pulse train in the blue channel in the combiner region. In this way, a large output amplitude due to constructive interference is obtained only for a suitable correlation of the spin-wave phases in neighboring pulses of the input signal. Thus, a particular pulse sequence has to be used to create a strong output signal which serves to wake up the next connected device. To investigate the spin-wave dynamics in the proposed structure, micromagnetic simulations were performed using Mumax3.28 The typical parameters of nanometer thick Yttrium Iron Garnet (YIG) were used in the simulations:10 saturation magnetization Ms ¼ 1.4 105 A/m, exchange constant A ¼ 3.5 pJ/m, and Gilbert damping a ¼ 2 104. The damping at the ends of the simulated
115, 092401-1
Applied Physics Letters
structure was set to exponentially increase to 0.5 to prevent spin-wave reflection. In the real device, these waves could also be processed further in a larger magnonic network. The mesh was set to 10 10 50 nm3. The geometric sizes of the magnonic wake-up receiver are as follows: The width of the waveguides is w ¼ 100 nm, the thickness is h ¼ 50 nm, the minimum gap between the waveguide and the circular ring is 10 nm, and the center to center distance of the waveguides is d ¼ 200 nm. The angle of the Y-shaped splitter and combiner is h ¼ 20 , and the outer radius of the corner is R ¼ 200 nm as shown in Fig. 1(b). The length where the waveguide and the circular ring structure are coupled to each other is adjusted to Lw ¼ 500 nm as shown in Fig. 1(b). By choosing a proper spin-wave frequency, all the incident wave energy is transferred from the waveguide to the magnonic ring structure5 over this distance. Other frequencies can be used by changing Lw, the gap between the ring and the waveguide or the applied magnetic bias field. Figure 1(b) shows the simulated ground state in which the color code shows the orientation of the magnetization. Note that no external bias field is applied in our simulation. In the nanoscale waveguide, the static magnetization orients itself parallel to the x-direction spontaneously due to the strong shape anisotropy. In the circular ring structure, the static magnetization is the global vortex state where the magnetization orients itself along the ring. This global vortex state in the ring structure has already been reported experimentally in Permalloy.29,30 In our simulation, a similar behavior is found and the global vortex in the ring can be realized by applying a large external field along the xdirection to magnetize both the waveguides and the ring structure in the x-direction and, consequently, slowly decreasing this field to a small negative field (–20 mT in our case). The global vortex forms in the ring to minimize the total energy, and then, the external field is removed to obtain the remanence state. To determine the optimal parameters for the wake-up receiver, we first simulate the spin-wave dispersion in a single straight waveguide with the same parameters. A sinc function oscillation magnetic field by ¼ b0sinc(2pfct) (in the y-direction) was applied in the center of the waveguide over an area of 20 nm in length, with a maximum
FIG. 1. (a) Sketch of the spin-wave wake-up receiver consisting of a Y-shaped splitter, a combiner, and a coupled circular ring structure. (b) Simulated ground state of the spin-wave wake-up receiver. The color code wheel represents the in-plane orientations of magnetization. (c) The simulated (color map) and theoretical (white dashed line) dispersion curves and theoretical (blue line) spin-wave group velocity.
Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Published under license by AIP Publishing
ARTICLE
scitation.org/journal/apl
oscillation amplitude of b0 ¼ 1 mT and a cut-off frequency of fc ¼ 10 GHz. Figure 1(c) shows the dispersion curves of the first two lowest spin-wave width modes of an isolated straight waveguide. The color map was obtained by micromagnetic simulations, and the white dashed line for the lowest width mode was calculated using the analytical theory for fully unpinned boundary conditions.31 The blue line shows the analytical group velocity of the lowest width mode. A frequency of f ¼ 2.8 GHz (kx ¼ 0.03 rad/nm, k ¼ 210 nm, and group velocity vg ¼ 0.39 lm/ns) has been chosen as the working frequency. Then, only one wavenumber-frequency combination exists for the fundamental mode, avoiding the scattering between different width modes after the bent waveguides.32,33 All the working parameters are shown in Fig. 1(c). As we mentioned above, the phase of the spin waves coupled back from the ring in the red channel in Fig. 1(a) can be adjusted by changing the delay length Ld of the circular ring structure. Simulations with different delay lengths Ld were performed with a continuous spin-wave excitation at frequency f. The normalized output power as a function of the delay length Ld is shown in Fig. 2(b). A minimum is found at Ld ¼ 710 nm, which corresponds to a destructive interference between the two branches. Figure 2(a) shows a snapshot of the out-ofplane component of the dynamics magnetization mz, which is equivalent to the spin-wave amplitude, for a delay length of Ld ¼ 710 nm. The obvious destructive interference, which is observed in the combiner region, implies that the circular ring structure introduces an effective phase shift of p for the spin waves in the lower branch. Note that a destructive interference between the back coupled spin waves from the ring and the spin waves transferred to the ring has also been observed in the coupled region. However, due to the linear superposition of the two waves, this interference only affects the intensity profile in the coupled region and does not affect the spin-wave phase and amplitude outside of this region. Figure 2(c) shows the amplitudes of the spin-wave signal at point 1 and point 2 [marked in Fig. 2(a)] in the time range of 18 ns to 22 ns after excitation. From this, the phase shift of Du ¼ p is clearly visible. Note that the continuous spin waves are employed only to study the phase shift caused by the coupled ring, but the wake-up signal is a specifically designed pulse train.
FIG. 2. (a) Snapshot of the out-of-plane component spin-wave amplitude mz for a continuous excitation source with a frequency of f ¼ 2.8 GHz. (b) The normalized output power as a function of delay length Ld. (c) Spin-wave amplitudes at points 1 and 2 in the time range of 18 ns to 22 ns after excitation.
115, 092401-2
Applied Physics Letters
To study the delay time caused by the coupled magnonic ring, a single spin-wave pulse of 5 ns duration (sd ¼ 5 ns) is applied. Figure 3(a) shows a snapshot of mz in which the spin-wave energy transferring between the waveguide and the magnonic ring is clearly observed. The output spin-wave amplitude as a function of time is shown in Fig. 3(b). Two clear peaks are found, which correspond to the spin-wave packets exiting from the upper and the lower branch, respectively. The delay time of these two spin-wave packets is Dt ¼9.5 ns (peak-to-peak distance). This value is slightly larger than the value of 8.6 ns estimated from sp ¼ 2Lw þ2Lvgd þ2pRc , where Lw is the coupled length which strongly depends on the spin-wave frequency, Ld is the delay length, Rc ¼ 150 nm is the radius of the corner (center to center), and vg is the group velocity, which also strongly depends on the frequency as shown in Fig. 1(d) (blue line). This is due to the fact that the group velocity in the corner of the ring structure is smaller than the one in the straight waveguide.34,35 Moreover, the amplitude of the second peak localized around 22 ns is smaller than that of the first one due to its longer propagation length caused by the coupled magnonic ring. In this design, a RF pulse train that can be used to create a strong signal at the output is, for example, an alternating pulse train of logical 0 (phase 0) and logical 1 (phase p) as depicted in Fig. 4(a). Since the induced time delay Dt of the second pulse train [red wave in Fig. 4(a)] matches the period of the initial spin-wave pulse sp and shifts the phase of the wave packet by p, the logical bit number n þ 1 interferes always with the negated logical bit number n (n is an integer). For the alternating pulse train 1010101010, this results in a constructive interference of the central spin-wave pulses, leading to a large envelop of the spin-wave intensity shown in Fig. 4(c). The jitter of the output signal is only 0.1 ns, which is much smaller than the period of the output signal. Thus, it can be ignored in the proposed design. Other pulse sequences like the homogenous sequence 1111111111 create a much lower signal due to the destructive interference shown in Figs. 4(b) and 4(d). Note that the output, signal resulting from destructive interference, is not zero because of the difference in the amplitude of the
FIG. 3. (a) Snapshot of the out-of-plane component spin-wave amplitude mz. (b) The output spin-wave amplitude as a function of simulation time t.
Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Published under license by AIP Publishing
ARTICLE
scitation.org/journal/apl
FIG. 4. (a) and (b): RF pattern recognition scheme (1010101010 and 1111111111): The red pulse train has been delayed by one pulse period sp, and its phase is, consequently, shifted by Du ¼ p. (c) and (d): The simulated output spin-wave intensities which are extracted from a 500 nm long detection region at the output waveguide. (e) and (f) are similar to (c) and (d), but for a detection region with a length of 10 lm. The output spin-wave intensities are proportional to the output voltages.
two interfering spin waves. The output spin-wave signals can be converted back to a DC by placing a heavy metal layer (e.g., Pt) on top of the YIG waveguide using the inverse spin Hall effect. The advantages of the inverse spin Hall effect are the linear response of the spin-wave intensity in the linear region and the insensitivity of the spin-wave phase.13,36 Therefore, the shapes of the output DC currents also depend on the length of the detection region (i.e., the length of the Pt). Figures 4(c) and 4(d) show the spin-wave intensities which are extracted from the output waveguide with a length of 500 nm. This length is much smaller than the spacing between two output pulses (Ls ¼ vgsp ¼ 3.7 lm). Therefore, individual pulses can be detected and dips are observed between two adjacent pulses. The individual spinwave pulses can be converted into a wide DC pulse by increasing the detection length to detect several spin-wave pulses simultaneously. Figures 4(e) and 4(f) show the output signals extracted from a detection region with a length of 10 lm. A wide pulse with the full width at half maximum of around 85 ns is observed, and thus, if a threshold current is defined, this pulse allows us to trigger an action of the next connected device without any additional standby circuits. Note also that the absolute value of the phase is not essential in this approach. Only the phase difference between neighboring bits is of interest, which is important since only this information can be transmitted between an arbitrarily placed sender and receiver. Another advantage is the passive property that does not require additional RF or DC sources as the reference, which are hard to integrate on a chip. Also note that once the geometric sizes of the wake-up receiver are fixed, the wake-up signal of this device is unique. The unique wake-up signal significantly avoids the unnecessary wake-ups in real-world scenarios, which are fully filled with various signals. Furthermore, a different geometric design can be used to encode other pulse sequences. For instance, the homogenous sequence 11111111 pulse train can also be
115, 092401-3
Applied Physics Letters
used as a trigger signal by adjusting the delay length to meet the criterion in which the time delay Dt is equal to the pulse period sp and the phase shift is zero. In summary, we have designed a planar nanoscale spin-wave wake-up receiver based on a simple wave interference effect and tested it by utilizing micromagnetic simulations. The wake-up receiver consists of a symmetric Y-shaped splitter, a combiner, and a coupled ring structure which acts as a phase shifter and delay line. The operational principle of the spin-wave wake-up receiver is demonstrated. Relying on spin-wave interference, a dedicated RF pulse train can be used to create a strong signal at the output to wake up the next connected device. The presented device operates intrinsically passive and does not require any additional RF sources as a reference. Financial support by DFG within project B01 of the Transregional Collaborative Research Centre (SFB/TRR) 173 “SpinþX” and ERC Starting Grant 678309 MagnonCircuits and the EU Horizon 2020 research and innovation programme within the FET-OPEN project CHIRON (Contract No. 801055) is gratefully acknowledged. REFERENCES 1
G. U. Gamm, M. Sippel, M. Kostic, and L. M. Reindl, in Sensor Networks and Information Processing (ISSNIP), Brisbane, QLD, Australia, 7–10 December (2010), pp. 121–126. 2 A. Khitun, M. Bao, and K. L. Wang, J. Phys. D: Appl. Phys. 43, 264005 (2010). 3 B. Lenk, H. Ulrichs, F. Garbs, and M. M€ unzenberg, Phys. Rep. 507, 107 (2011). 4 A. V. Chumak, V. I. Vasyuchka, A. A. Serga, and B. Hillebrands, Nat. Phys. 11, 453 (2015). 5 Q. Wang, P. Pirro, R. Verba, A. Slavin, B. Hillebrands, and A. V. Chumak, Sci. Adv. 4, e1701517 (2018). 6 V. V. Kruglyak, S. O. Demokritov, and D. Grundler, J. Phys. D: Appl. Phys. 43, 264001 (2010). 7 Y. Kajiwara, K. Harii, S. Takahashi, J. Ohe, K. Uchida, M. Mizuguchi, H. Umezawa, H. Kawai, K. Ando, K. Takanashi, S. Maekawa, and E. Saitoh, Nature 464, 262 (2010). 8 P. Pirro, T. Br€acher, A. V. Chumak, B. L€agel, C. Dubs, O. Surzhenko, P. G€ornert, B. Leven, and B. Hillebrands, Appl. Phys. Lett. 104, 012402 (2014). 9 H. Yu, O. d’Allivy Kelly, V. Cros, R. Bernard, P. Bortolotti, A. Anane, F. Brandl, R. Huber, I. Stasinopoulos, and D. Grundler, Sci. Rep. 4, 6848 (2014). 10 C. Dubs, O. Surzhenko, R. Linke, A. Danilewsky, U. Br€ uckner, and J. Dellith, J. Phys. D: Appl. Phys. 50, 204005 (2017). 11 S. Wintz, V. Tiberkevich, M. Weigand, J. Raabe, J. Lindner, A. Erbe, A. Slavin, and J. Fassbender, Nat. Nanotechnol. 11, 948 (2016). 12 H. Yu, O. d’Allivy Kelly, V. Cros, R. Bernard, P. Bortolotti, A. Anane, F. Brandl, F. Heimbach, and D. Grundler, Nat. Commun. 7, 11255 (2016).
Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Published under license by AIP Publishing
ARTICLE
scitation.org/journal/apl
13
T. Br€acher, M. Fabre, T. Meyer, T. Fischer, S. Auffret, O. Boulle, U. Ebels, P. Pirro, and G. Gaudin, Nano Lett. 17, 7234 (2017). 14 S. J. H€am€al€ainen, F. Brandl, K. J. A. Franke, D. Grundler, and S. Van Dijken, Phys. Rev. Appl. 8, 014020 (2017). 15 C. S. Davies and V. V. Kruglyak, IEEE Trans. Magn. 52, 2300504 (2016). 16 T. Kampfrath, A. Sell, G. Klatt, A. Pashkin, S. M€ahrlein, T. Dekorsy, M. Wolf, M. Fiebig, A. Leitenstorfer, and R. Huber, Nat. Photonics 5, 31 (2011). 17 T. Balashov, P. Buczek, L. Sandratskii, A. Ernst, and W. Wulfhekel, J. Phys.: Condens. Matter 26, 394007 (2014). 18 A. V. Sadovnikov, E. N. Beginin, M. A. Morozova, Yu. P. Sharaevskii, S. V. Grishin, S. E. Sheshukova, and S. A. Nikitov, Appl. Phys. Lett. 109, 042407 (2016). 19 T. M. Spicer, P. S. Keatley, M. Dvornik, T. H. J. Loughran, A. A. Awad, P. D€ urrenfeld, A. Houshang, M. Ranjbar, J. Akerman, V. V. Kruglyak, and R. J. Hicken, Appl. Phys. Lett. 113, 192405 (2018). 20 Q. Wang, R. Verba, T. Br€acher, P. Pirro, and A. V. Chumak, preprint arXiv:1902.02855 (2019). 21 S. Klingler, P. Pirro, T. Br€acher, B. Leven, B. Hillebrands, and A. V. Chumak, Appl. Phys. Lett. 105, 152410 (2014). 22 T. Fischer, M. Kewenig, D. A. Bozhko, A. A. Serga, I. I. Syvorotka, F. Ciubotaru, C. Adelmann, B. Hillebrands, and A. V. Chumak, Appl. Phys. Lett. 110, 152401 (2017). 23 T. Kimura, Y. Otani, T. Sato, S. Takahashi, and S. Maekawa, Phys. Rev. Lett. 98, 156601 (2007). 24 S. Cherepov, P. K. Amiri, J. G. Alzate, K. Wong, M. Lewis, P. Upadhyaya, J. Nath, M. Bao, A. Bur, T. Wu, G. P. Garman, A. Khitun, and K. L. Wang, Appl. Phys. Lett. 104, 082403 (2014). 25 Y. Au, E. Ahmad, O. Dmytriiev, M. Dvornik, T. Davison, and V. V. Kruglyak, Appl. Phys. Lett. 100, 182404 (2012). 26 A. V. Sadovnikov, C. S. Davies, S. V. Grishin, V. V. Kruglyak, D. V. Romanenko, Y. P. Sharaevskii, and S. A. Nikitov, Appl. Phys. Lett. 106, 192406 (2015). 27 A. V. Sadovnikov, E. N. Beginin, S. E. Sheshukova, D. V. Romanenko, Y. P. Sharaevskii, and S. A. Nikitov, Appl. Phys. Lett. 107, 202405 (2015). 28 A. Vansteenkiste, J. Leliaert, M. Dvornik, M. Helsen, F. Garcia-Sanchez, and B. Van Waeyenberge, AIP Adv. 4, 107133 (2014). 29 M. Steiner and J. Nitta, Appl. Phys. Lett. 84, 939 (2004). 30 Y. Ren and A. O. Adeyeye, J. Appl. Phys. 105, 063901 (2009). 31 Q. Wang, B. Heinz, R. Verba, M. Kewenig, P. Pirro, M. Schneider, T. Meyer, B. L€agel, C. Dubs, T. Br€acher, and A. V. Chumak, Phys. Rev. Lett. 122, 247202 (2019). 32 P. Clausen, K. Vogt, H. Schultheiss, S. Sch€afer, B. Obry, G. Wolf, P. Pirro, B. Leven, and B. Hillebrands, Appl. Phys. Lett. 99, 162505 (2011). 33 A. V. Sadovnikov, C. S. Davies, V. V. Kruglyak, D. V. Romanenko, S. V. Grishin, E. N. Beginin, Y. P. Sharaevskii, and S. A. Nikitov, Phys. Rev. B 96, 060401 (2017). 34 N. Kumar, G. Venkat, and A. Prabhakar, IEEE Trans. Magn. 50, 1300306 (2014). 35 V. S. Tkachenko, A. N. Kuchko, M. Dvornik, and V. V. Kruglyak, Appl. Phys. Lett. 101, 152402 (2012). 36 A. V. Chumak, A. A. Serga, M. B. Jungfleisch, R. Neb, D. A. Bozhko, V. S. Tiberkevich, and B. Hillebrands, Appl. Phys. Lett. 100, 082405 (2012).
115, 092401-4
, T. Brächer, M. Mohseni
, B. Hillebrands, V. I. Vasyuchka
, A. V. Chumak
, and P. Pirro
ARTICLES YOU MAY BE INTERESTED IN Enhancement of spin–orbit torque via interfacial hydrogen and oxygen ion manipulation Applied Physics Letters 115, 092402 (2019); https://doi.org/10.1063/1.5110206 Direct observation of magneto-Peltier effect in current-in-plane giant magnetoresistive spin valve Applied Physics Letters 115, 092406 (2019); https://doi.org/10.1063/1.5120569 Role of interface quality for the spin Hall magnetoresistance in nickel ferrite thin films with bulk-like magnetic properties Applied Physics Letters 115, 092403 (2019); https://doi.org/10.1063/1.5097600
Appl. Phys. Lett. 115, 092401 (2019); https://doi.org/10.1063/1.5109623 © 2019 Author(s).
115, 092401
Applied Physics Letters
ARTICLE
scitation.org/journal/apl
Nanoscale spin-wave wake-up receiver Cite as: Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Submitted: 10 May 2019 . Accepted: 10 August 2019 . Published Online: 26 August 2019 Q. Wang,a)
€ cher, M. Mohseni, T. Bra
B. Hillebrands, V. I. Vasyuchka,
A. V. Chumak,
and P. Pirro
AFFILIATIONS € t Kaiserslautern, 67663 Kaiserslautern, Germany Fachbereich Physik and Landesforschungszentrum OPTIMAS, Technische Universita a)
Author to whom correspondence should be addressed: [email protected]
ABSTRACT We present the concept of a passive spin-wave device which is able to distinguish different radio frequency pulse trains and validate its functionality using micromagnetic simulations. The information is coded in the phase of the individual pulses which are transformed into spinwave packets. The device splits every incoming packet into two arms, one of which is coupled to a magnonic ring which introduces a well-defined time delay and phase shift. Since the time delay is matched to the pulse repetition rate, adjacent packets interfere in a combiner, which makes it possible to distinguish simple pulse train patterns by the readout of the time-integrated spin-wave intensity in the output. Due to its passive construction, this device may serve as an energy-efficient wake-up receiver used to activate the main receiver circuit in power critical internet of things applications. Published under license by AIP Publishing. https://doi.org/10.1063/1.5109623
An ultralow power wireless communication capability is needed to realize wireless sensor networks in real-world scenarios. The most efficient way to reduce energy consumption is to leave the device off and wake it up asynchronously only when it is really needed. For this task, one needs a wake-up receiver which continuously monitors a channel, listens for a wake-up signal from other nodes, and activates the device on stimulus.1 Therefore, the power consumption as well as the footprint of the wake-up receiver is of crucial importance. Spin waves, a collective precessional motion of localized magnetic moments, can propagate at gigahertz frequencies in either conducting or insulating magnetic media with wavelength down to tens of nanometers and without any flow of electrons, reducing the energy consumption caused by Joule heating.2–6 In general, spin waves, and their quanta magnons, are considered as potential candidates as carriers of information in future devices due to their low losses,7–10 short wavelengths,11–15 high frequencies,16,17 and abundant nonlinear phenomena.18–20 Another advantage of spin waves is that due to their wave nature, they can be used for transfer and processing of information, which is coded in their amplitude and phase. Interference effects between different waves play a major role in harvesting the full potential of the wave nature for data processing. A prominent example for the advantages of wave-based data processing is the majority gate, whose output state represents the majority of input states,21,22 and which is an especially compact and readily scalable wave-based device. In this letter, we design a magnonic wake-up receiver based on spin-wave interference. The basic concept of the magnonic wake-up receiver consists of a Y-shaped splitter, a combiner, and a coupled
Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Published under license by AIP Publishing
magnonic circular ring structure, as depicted in Fig. 1(a). The device works in remanence. Radio frequency (RF) pulse trains from other wireless nodes will be converted into spin-wave pulse trains by a magnonic transducer in the input.23–25 The blue and red arrows show the flow path of the spin-wave packets in the magnonic wake-up receiver, which are split into two equal parts by a symmetric Y-shaped splitter.26 One of them is directly guided to the combiner region by the upper waveguide (blue channel). The other one is first transferred to the adjacent magnonic circular ring structure by dipolar coupling5,27 (the red dashed area), propagates along the ring, and then couples back to the waveguide with an additional time delay Dt and phase shift Du (red channel). The time delay and phase shift can be controlled by adjusting the delay length Ld [see Fig. 1(b)]. When the time delay Dt is equal to the pulse period sp of the spin-wave pulse train, the back coupled spin wave in the red channel will completely interfere with the next packet of the spin-wave pulse train in the blue channel in the combiner region. In this way, a large output amplitude due to constructive interference is obtained only for a suitable correlation of the spin-wave phases in neighboring pulses of the input signal. Thus, a particular pulse sequence has to be used to create a strong output signal which serves to wake up the next connected device. To investigate the spin-wave dynamics in the proposed structure, micromagnetic simulations were performed using Mumax3.28 The typical parameters of nanometer thick Yttrium Iron Garnet (YIG) were used in the simulations:10 saturation magnetization Ms ¼ 1.4 105 A/m, exchange constant A ¼ 3.5 pJ/m, and Gilbert damping a ¼ 2 104. The damping at the ends of the simulated
115, 092401-1
Applied Physics Letters
structure was set to exponentially increase to 0.5 to prevent spin-wave reflection. In the real device, these waves could also be processed further in a larger magnonic network. The mesh was set to 10 10 50 nm3. The geometric sizes of the magnonic wake-up receiver are as follows: The width of the waveguides is w ¼ 100 nm, the thickness is h ¼ 50 nm, the minimum gap between the waveguide and the circular ring is 10 nm, and the center to center distance of the waveguides is d ¼ 200 nm. The angle of the Y-shaped splitter and combiner is h ¼ 20 , and the outer radius of the corner is R ¼ 200 nm as shown in Fig. 1(b). The length where the waveguide and the circular ring structure are coupled to each other is adjusted to Lw ¼ 500 nm as shown in Fig. 1(b). By choosing a proper spin-wave frequency, all the incident wave energy is transferred from the waveguide to the magnonic ring structure5 over this distance. Other frequencies can be used by changing Lw, the gap between the ring and the waveguide or the applied magnetic bias field. Figure 1(b) shows the simulated ground state in which the color code shows the orientation of the magnetization. Note that no external bias field is applied in our simulation. In the nanoscale waveguide, the static magnetization orients itself parallel to the x-direction spontaneously due to the strong shape anisotropy. In the circular ring structure, the static magnetization is the global vortex state where the magnetization orients itself along the ring. This global vortex state in the ring structure has already been reported experimentally in Permalloy.29,30 In our simulation, a similar behavior is found and the global vortex in the ring can be realized by applying a large external field along the xdirection to magnetize both the waveguides and the ring structure in the x-direction and, consequently, slowly decreasing this field to a small negative field (–20 mT in our case). The global vortex forms in the ring to minimize the total energy, and then, the external field is removed to obtain the remanence state. To determine the optimal parameters for the wake-up receiver, we first simulate the spin-wave dispersion in a single straight waveguide with the same parameters. A sinc function oscillation magnetic field by ¼ b0sinc(2pfct) (in the y-direction) was applied in the center of the waveguide over an area of 20 nm in length, with a maximum
FIG. 1. (a) Sketch of the spin-wave wake-up receiver consisting of a Y-shaped splitter, a combiner, and a coupled circular ring structure. (b) Simulated ground state of the spin-wave wake-up receiver. The color code wheel represents the in-plane orientations of magnetization. (c) The simulated (color map) and theoretical (white dashed line) dispersion curves and theoretical (blue line) spin-wave group velocity.
Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Published under license by AIP Publishing
ARTICLE
scitation.org/journal/apl
oscillation amplitude of b0 ¼ 1 mT and a cut-off frequency of fc ¼ 10 GHz. Figure 1(c) shows the dispersion curves of the first two lowest spin-wave width modes of an isolated straight waveguide. The color map was obtained by micromagnetic simulations, and the white dashed line for the lowest width mode was calculated using the analytical theory for fully unpinned boundary conditions.31 The blue line shows the analytical group velocity of the lowest width mode. A frequency of f ¼ 2.8 GHz (kx ¼ 0.03 rad/nm, k ¼ 210 nm, and group velocity vg ¼ 0.39 lm/ns) has been chosen as the working frequency. Then, only one wavenumber-frequency combination exists for the fundamental mode, avoiding the scattering between different width modes after the bent waveguides.32,33 All the working parameters are shown in Fig. 1(c). As we mentioned above, the phase of the spin waves coupled back from the ring in the red channel in Fig. 1(a) can be adjusted by changing the delay length Ld of the circular ring structure. Simulations with different delay lengths Ld were performed with a continuous spin-wave excitation at frequency f. The normalized output power as a function of the delay length Ld is shown in Fig. 2(b). A minimum is found at Ld ¼ 710 nm, which corresponds to a destructive interference between the two branches. Figure 2(a) shows a snapshot of the out-ofplane component of the dynamics magnetization mz, which is equivalent to the spin-wave amplitude, for a delay length of Ld ¼ 710 nm. The obvious destructive interference, which is observed in the combiner region, implies that the circular ring structure introduces an effective phase shift of p for the spin waves in the lower branch. Note that a destructive interference between the back coupled spin waves from the ring and the spin waves transferred to the ring has also been observed in the coupled region. However, due to the linear superposition of the two waves, this interference only affects the intensity profile in the coupled region and does not affect the spin-wave phase and amplitude outside of this region. Figure 2(c) shows the amplitudes of the spin-wave signal at point 1 and point 2 [marked in Fig. 2(a)] in the time range of 18 ns to 22 ns after excitation. From this, the phase shift of Du ¼ p is clearly visible. Note that the continuous spin waves are employed only to study the phase shift caused by the coupled ring, but the wake-up signal is a specifically designed pulse train.
FIG. 2. (a) Snapshot of the out-of-plane component spin-wave amplitude mz for a continuous excitation source with a frequency of f ¼ 2.8 GHz. (b) The normalized output power as a function of delay length Ld. (c) Spin-wave amplitudes at points 1 and 2 in the time range of 18 ns to 22 ns after excitation.
115, 092401-2
Applied Physics Letters
To study the delay time caused by the coupled magnonic ring, a single spin-wave pulse of 5 ns duration (sd ¼ 5 ns) is applied. Figure 3(a) shows a snapshot of mz in which the spin-wave energy transferring between the waveguide and the magnonic ring is clearly observed. The output spin-wave amplitude as a function of time is shown in Fig. 3(b). Two clear peaks are found, which correspond to the spin-wave packets exiting from the upper and the lower branch, respectively. The delay time of these two spin-wave packets is Dt ¼9.5 ns (peak-to-peak distance). This value is slightly larger than the value of 8.6 ns estimated from sp ¼ 2Lw þ2Lvgd þ2pRc , where Lw is the coupled length which strongly depends on the spin-wave frequency, Ld is the delay length, Rc ¼ 150 nm is the radius of the corner (center to center), and vg is the group velocity, which also strongly depends on the frequency as shown in Fig. 1(d) (blue line). This is due to the fact that the group velocity in the corner of the ring structure is smaller than the one in the straight waveguide.34,35 Moreover, the amplitude of the second peak localized around 22 ns is smaller than that of the first one due to its longer propagation length caused by the coupled magnonic ring. In this design, a RF pulse train that can be used to create a strong signal at the output is, for example, an alternating pulse train of logical 0 (phase 0) and logical 1 (phase p) as depicted in Fig. 4(a). Since the induced time delay Dt of the second pulse train [red wave in Fig. 4(a)] matches the period of the initial spin-wave pulse sp and shifts the phase of the wave packet by p, the logical bit number n þ 1 interferes always with the negated logical bit number n (n is an integer). For the alternating pulse train 1010101010, this results in a constructive interference of the central spin-wave pulses, leading to a large envelop of the spin-wave intensity shown in Fig. 4(c). The jitter of the output signal is only 0.1 ns, which is much smaller than the period of the output signal. Thus, it can be ignored in the proposed design. Other pulse sequences like the homogenous sequence 1111111111 create a much lower signal due to the destructive interference shown in Figs. 4(b) and 4(d). Note that the output, signal resulting from destructive interference, is not zero because of the difference in the amplitude of the
FIG. 3. (a) Snapshot of the out-of-plane component spin-wave amplitude mz. (b) The output spin-wave amplitude as a function of simulation time t.
Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Published under license by AIP Publishing
ARTICLE
scitation.org/journal/apl
FIG. 4. (a) and (b): RF pattern recognition scheme (1010101010 and 1111111111): The red pulse train has been delayed by one pulse period sp, and its phase is, consequently, shifted by Du ¼ p. (c) and (d): The simulated output spin-wave intensities which are extracted from a 500 nm long detection region at the output waveguide. (e) and (f) are similar to (c) and (d), but for a detection region with a length of 10 lm. The output spin-wave intensities are proportional to the output voltages.
two interfering spin waves. The output spin-wave signals can be converted back to a DC by placing a heavy metal layer (e.g., Pt) on top of the YIG waveguide using the inverse spin Hall effect. The advantages of the inverse spin Hall effect are the linear response of the spin-wave intensity in the linear region and the insensitivity of the spin-wave phase.13,36 Therefore, the shapes of the output DC currents also depend on the length of the detection region (i.e., the length of the Pt). Figures 4(c) and 4(d) show the spin-wave intensities which are extracted from the output waveguide with a length of 500 nm. This length is much smaller than the spacing between two output pulses (Ls ¼ vgsp ¼ 3.7 lm). Therefore, individual pulses can be detected and dips are observed between two adjacent pulses. The individual spinwave pulses can be converted into a wide DC pulse by increasing the detection length to detect several spin-wave pulses simultaneously. Figures 4(e) and 4(f) show the output signals extracted from a detection region with a length of 10 lm. A wide pulse with the full width at half maximum of around 85 ns is observed, and thus, if a threshold current is defined, this pulse allows us to trigger an action of the next connected device without any additional standby circuits. Note also that the absolute value of the phase is not essential in this approach. Only the phase difference between neighboring bits is of interest, which is important since only this information can be transmitted between an arbitrarily placed sender and receiver. Another advantage is the passive property that does not require additional RF or DC sources as the reference, which are hard to integrate on a chip. Also note that once the geometric sizes of the wake-up receiver are fixed, the wake-up signal of this device is unique. The unique wake-up signal significantly avoids the unnecessary wake-ups in real-world scenarios, which are fully filled with various signals. Furthermore, a different geometric design can be used to encode other pulse sequences. For instance, the homogenous sequence 11111111 pulse train can also be
115, 092401-3
Applied Physics Letters
used as a trigger signal by adjusting the delay length to meet the criterion in which the time delay Dt is equal to the pulse period sp and the phase shift is zero. In summary, we have designed a planar nanoscale spin-wave wake-up receiver based on a simple wave interference effect and tested it by utilizing micromagnetic simulations. The wake-up receiver consists of a symmetric Y-shaped splitter, a combiner, and a coupled ring structure which acts as a phase shifter and delay line. The operational principle of the spin-wave wake-up receiver is demonstrated. Relying on spin-wave interference, a dedicated RF pulse train can be used to create a strong signal at the output to wake up the next connected device. The presented device operates intrinsically passive and does not require any additional RF sources as a reference. Financial support by DFG within project B01 of the Transregional Collaborative Research Centre (SFB/TRR) 173 “SpinþX” and ERC Starting Grant 678309 MagnonCircuits and the EU Horizon 2020 research and innovation programme within the FET-OPEN project CHIRON (Contract No. 801055) is gratefully acknowledged. REFERENCES 1
G. U. Gamm, M. Sippel, M. Kostic, and L. M. Reindl, in Sensor Networks and Information Processing (ISSNIP), Brisbane, QLD, Australia, 7–10 December (2010), pp. 121–126. 2 A. Khitun, M. Bao, and K. L. Wang, J. Phys. D: Appl. Phys. 43, 264005 (2010). 3 B. Lenk, H. Ulrichs, F. Garbs, and M. M€ unzenberg, Phys. Rep. 507, 107 (2011). 4 A. V. Chumak, V. I. Vasyuchka, A. A. Serga, and B. Hillebrands, Nat. Phys. 11, 453 (2015). 5 Q. Wang, P. Pirro, R. Verba, A. Slavin, B. Hillebrands, and A. V. Chumak, Sci. Adv. 4, e1701517 (2018). 6 V. V. Kruglyak, S. O. Demokritov, and D. Grundler, J. Phys. D: Appl. Phys. 43, 264001 (2010). 7 Y. Kajiwara, K. Harii, S. Takahashi, J. Ohe, K. Uchida, M. Mizuguchi, H. Umezawa, H. Kawai, K. Ando, K. Takanashi, S. Maekawa, and E. Saitoh, Nature 464, 262 (2010). 8 P. Pirro, T. Br€acher, A. V. Chumak, B. L€agel, C. Dubs, O. Surzhenko, P. G€ornert, B. Leven, and B. Hillebrands, Appl. Phys. Lett. 104, 012402 (2014). 9 H. Yu, O. d’Allivy Kelly, V. Cros, R. Bernard, P. Bortolotti, A. Anane, F. Brandl, R. Huber, I. Stasinopoulos, and D. Grundler, Sci. Rep. 4, 6848 (2014). 10 C. Dubs, O. Surzhenko, R. Linke, A. Danilewsky, U. Br€ uckner, and J. Dellith, J. Phys. D: Appl. Phys. 50, 204005 (2017). 11 S. Wintz, V. Tiberkevich, M. Weigand, J. Raabe, J. Lindner, A. Erbe, A. Slavin, and J. Fassbender, Nat. Nanotechnol. 11, 948 (2016). 12 H. Yu, O. d’Allivy Kelly, V. Cros, R. Bernard, P. Bortolotti, A. Anane, F. Brandl, F. Heimbach, and D. Grundler, Nat. Commun. 7, 11255 (2016).
Appl. Phys. Lett. 115, 092401 (2019); doi: 10.1063/1.5109623 Published under license by AIP Publishing
ARTICLE
scitation.org/journal/apl
13
T. Br€acher, M. Fabre, T. Meyer, T. Fischer, S. Auffret, O. Boulle, U. Ebels, P. Pirro, and G. Gaudin, Nano Lett. 17, 7234 (2017). 14 S. J. H€am€al€ainen, F. Brandl, K. J. A. Franke, D. Grundler, and S. Van Dijken, Phys. Rev. Appl. 8, 014020 (2017). 15 C. S. Davies and V. V. Kruglyak, IEEE Trans. Magn. 52, 2300504 (2016). 16 T. Kampfrath, A. Sell, G. Klatt, A. Pashkin, S. M€ahrlein, T. Dekorsy, M. Wolf, M. Fiebig, A. Leitenstorfer, and R. Huber, Nat. Photonics 5, 31 (2011). 17 T. Balashov, P. Buczek, L. Sandratskii, A. Ernst, and W. Wulfhekel, J. Phys.: Condens. Matter 26, 394007 (2014). 18 A. V. Sadovnikov, E. N. Beginin, M. A. Morozova, Yu. P. Sharaevskii, S. V. Grishin, S. E. Sheshukova, and S. A. Nikitov, Appl. Phys. Lett. 109, 042407 (2016). 19 T. M. Spicer, P. S. Keatley, M. Dvornik, T. H. J. Loughran, A. A. Awad, P. D€ urrenfeld, A. Houshang, M. Ranjbar, J. Akerman, V. V. Kruglyak, and R. J. Hicken, Appl. Phys. Lett. 113, 192405 (2018). 20 Q. Wang, R. Verba, T. Br€acher, P. Pirro, and A. V. Chumak, preprint arXiv:1902.02855 (2019). 21 S. Klingler, P. Pirro, T. Br€acher, B. Leven, B. Hillebrands, and A. V. Chumak, Appl. Phys. Lett. 105, 152410 (2014). 22 T. Fischer, M. Kewenig, D. A. Bozhko, A. A. Serga, I. I. Syvorotka, F. Ciubotaru, C. Adelmann, B. Hillebrands, and A. V. Chumak, Appl. Phys. Lett. 110, 152401 (2017). 23 T. Kimura, Y. Otani, T. Sato, S. Takahashi, and S. Maekawa, Phys. Rev. Lett. 98, 156601 (2007). 24 S. Cherepov, P. K. Amiri, J. G. Alzate, K. Wong, M. Lewis, P. Upadhyaya, J. Nath, M. Bao, A. Bur, T. Wu, G. P. Garman, A. Khitun, and K. L. Wang, Appl. Phys. Lett. 104, 082403 (2014). 25 Y. Au, E. Ahmad, O. Dmytriiev, M. Dvornik, T. Davison, and V. V. Kruglyak, Appl. Phys. Lett. 100, 182404 (2012). 26 A. V. Sadovnikov, C. S. Davies, S. V. Grishin, V. V. Kruglyak, D. V. Romanenko, Y. P. Sharaevskii, and S. A. Nikitov, Appl. Phys. Lett. 106, 192406 (2015). 27 A. V. Sadovnikov, E. N. Beginin, S. E. Sheshukova, D. V. Romanenko, Y. P. Sharaevskii, and S. A. Nikitov, Appl. Phys. Lett. 107, 202405 (2015). 28 A. Vansteenkiste, J. Leliaert, M. Dvornik, M. Helsen, F. Garcia-Sanchez, and B. Van Waeyenberge, AIP Adv. 4, 107133 (2014). 29 M. Steiner and J. Nitta, Appl. Phys. Lett. 84, 939 (2004). 30 Y. Ren and A. O. Adeyeye, J. Appl. Phys. 105, 063901 (2009). 31 Q. Wang, B. Heinz, R. Verba, M. Kewenig, P. Pirro, M. Schneider, T. Meyer, B. L€agel, C. Dubs, T. Br€acher, and A. V. Chumak, Phys. Rev. Lett. 122, 247202 (2019). 32 P. Clausen, K. Vogt, H. Schultheiss, S. Sch€afer, B. Obry, G. Wolf, P. Pirro, B. Leven, and B. Hillebrands, Appl. Phys. Lett. 99, 162505 (2011). 33 A. V. Sadovnikov, C. S. Davies, V. V. Kruglyak, D. V. Romanenko, S. V. Grishin, E. N. Beginin, Y. P. Sharaevskii, and S. A. Nikitov, Phys. Rev. B 96, 060401 (2017). 34 N. Kumar, G. Venkat, and A. Prabhakar, IEEE Trans. Magn. 50, 1300306 (2014). 35 V. S. Tkachenko, A. N. Kuchko, M. Dvornik, and V. V. Kruglyak, Appl. Phys. Lett. 101, 152402 (2012). 36 A. V. Chumak, A. A. Serga, M. B. Jungfleisch, R. Neb, D. A. Bozhko, V. S. Tiberkevich, and B. Hillebrands, Appl. Phys. Lett. 100, 082405 (2012).
115, 092401-4