Metastable magnetic bubble in [Co/Pd]4/Py multilayers
Yurui Wei, Chengkun Song, Yunxu Ma, Hongmei Feng, Chenbo Zhao, Xiaolei Li, Chengdong Jin, Jinshuai Wang, Chunlei Zhang, Jianbo Wang, Jiangwei Cao, Qingfang Liu
MMetastable magnetic bubble in [Co/Pd] /Py multilayers Yurui Wei Chengkun Song Yunxu Ma Hongmei Feng Chenbo Zhao Xiaolei Li Chendong Jin Jinshuai Wang Chunlei Zhang , Jianbo Wang Jiangwei Cao and QingfangLiu Corresponding author: Qingfang LiuE-mail address: [email protected]
Abstract
Magnetic bubbles are topologically spin textures that offering the interesting physics andgreat promise for next-generation information storage technologies. The main obstacles so farare that magnetic bubbles are generated with no field stimuli in new material systems at roomtemperature. Here, we report the observation of individual magnetic bubbles and its highfrequency measurement at room temperature in an exchange-coupled [Co/Pd] /Py multilayers.We demonstrate that the emergence of magnetic bubbles at remanence can be tuned by thein-plane tilted magnetic field (roughly 3°) along the film plane at room temperature. Highfrequency results indicate that the presence of magnetic bubbles leads to broadening of themagnetic permeability spectrum lines (due to the non-uniformity of the magnetic moments).Our findings open the door to the bubble-based spintronics at room temperature inexchange-coupled magnetic multilayers. Keywords
Magnetic bubble, Tilted magnetic field, At remanence, High frequency
Metastable states [1-3], the non-equilibrium state of free energy when the system is aboveequilibrium states, are vital for the materials application since some properties of themetastable material outperforms its equilibrium properties or even exhibit special properties.Therefore, the study of the metastable states of materials not only exhibits the theoreticalsignificance, but also has the important practical value. A vortex-like magnetization texture,called skyrmion, is one of the typical metastable state under room temperature when it wasfirst found in non-centrosymmetric bulk materials with B20 structure [4-7]. Its ancestor iscalled magnetic bubble, which exist in ferromagnetic compounds such as garnet film [8] andalloy films [9-10] with perpendicular anisotropy. A magnetic bubble is a rod-like domain without-of-plane magnetization embedded in a thin plate with antiparallel magnetizationbackground and separated with a cylindrical Bloch wall. Magnetic bubbles and skyrmions areclosely relative as they exhibit in the same topology [11-12]. However, their characteristicsizes are different [13-17]. Since magnetic bubbles present a long lifetime at roomtemperature, which made them attracts attention in recent years. Various imaging techniques,uch as Scanning Probe Microscopy [17], Lorentz transmission electron microscopy [14-15],and magnetic-force microscopy [6,9], et al [13,18-20], were used to the detection of magneticbubbles in ultrathin films.The formation of magnetic bubble depends on some factors [9,26], such as the qualityfactor su MKQ (the ratio of the perpendicular magnetic anisotropy and magnetostaticenergies). If Q>1, the stripe domains are formed in magnetic films. With applied magneticfield perpendicular to the film surface, the domain states transform to bubble domain statesand the bubble size strongly depends on the external field [21-23]. As an intermediate state,the stripe domain is important for formation of magnetic bubble. However, in some alloysingle-layer film (like permalloy film), the strong interfacial perpendicular anisotropy is noteasy to obtain and the stripe domain exists above the critical thickness (for Py film [24], thecritical thickness is larger than 80 nm), which prevents the existence of magnetic bubble inthis kind of film under the critic thickness.In this work, we fabricate [Co/Pd] /Py multilayers, and obtained the stripe domain state in Pylayer with 15 nm. After magnetized by an in-plane titled magnetic field then measured atremnant state, the stripe domain state changed to magnetic bubble. The stability and chiralityas well as the high-frequency properties of these magnetic bubbles are then investigated withmagnetic force microscope (MFM) and vector network analyzer ferromagnetic resonance(VNA-FMR) techniques. The magnetic multilayers with the composition of Ta(10 nm)/Py(15 nm)/Ta(3 nm), Ta(10nm)/Pd (5 nm)/[Co(0.3 nm)/Pd(0.8 nm)] /Ta(3 nm) and Ta(10 nm)/Pd (5 nm)/[Co(0.3nm)/Pd(0.8 nm)] / Py(15 nm)/Ta(3 nm), where Py denotes permalloy (i.e., Ni Fe ), weredeposited onto a thermal oxidized Si substrate by magnetron sputtering with the base pressurebelow 1.9×10 -7 Torr [Figure 1(a)]. The Pd and Ta layers were deposited by radio-frequency(RF) sputtering, while the Py and Co layers were deposited by DC magnetron sputtering (thesputtering pressure of Ar gas was kept at 5 mTorr for all layers). A 3-nm Ta cap was depositedto prevent oxidation of the stack. The thin amorphous Ta (10 nm) seed layer allows for greatermobility of the deposited atoms and an improved fcc-(111) orientation of the Pd layerdeposition, thus improving the perpendicular anisotropy of Co/Pd multilayers [26].M–H hysteresis loops were measured using vibrating sample magnetometry (VSM, ADEtechnologies, EV9) and focused magneto-optical Kerr microscopy (NanoMOKE2). Magneticdomain structure was observed by a commercial MFP-3D AFM/MFM (Asylum Research).The MFM tips (ASYHMFM) were ~84 nm in radius and kept at a distance h of 20 nm fromthe sample surface. Ferromagnetic resonance measurements (FMR) were fulfilled usingvector network analyzer (VNA) as a function of the applied microwave magnetic field underdifferent dc magnetic field. All measurements were performed at room temperature.
3. Results and discussion
Figure 1(c) shows the hysteresis loops of the Ta(10 nm)/Pd(5 nm)/[Co(0.3 nm)/Pd(0.8nm)] / Py(15 nm)/Ta(3 nm) film with the applied field perpendicular and parallel to the filmlane. For comparison, the result of film without Py layer (that is Ta(10 nm)/Pd(5nm)/[Co(0.3 nm)/Pd(0.8 nm)] /Ta(3 nm)) is also shown in Figure 1(b). The hysteresis loop of[Co/Pd] multilayer shows obvious out-of-plane anisotropy with the effective anisotropyconstant K = 8.1 × J/m , which is smaller than that obtained in reference 25. However, theout-of-plane hysteresis loop of [Co/Pd] /Py exhibits a two-steps reversal, because the film iscomposed of a soft magnetic layer Py and exchange-coupled (Co/Pd) magnetic layers whichthe soft magnetic layer is first reversed and then the exchange-coupled layer flipped. Due tothe existence of Py, the magnetic moment is easy to tilt an angle from the out-of-planedirection, which is due to the competition between the in-plane anisotropy of Py and theout-of plane perpendicular anisotropy of (Co/Pd) . That leads to a lower saturated field inin-plane hysteresis loops of [Co/Pd] /Py film. In addition, the in-plane hysteresis loop of[Co/Pd] /Py film shows a little horizontal shift along the magnetic field with a bias field ofabout 4 Oe, which may due to the formation of un-equally oriented closure domains [27] orvortex cores [28] at the [Co/Pt] n /NiFe interface after in-plane saturation. If the out-of-planeanisotropy increase (realized by increase of stack number n in (Co/Pd) n to n=6 in ourexperiment), the exchange-bias became more obvious. The MOKE loop of Py layer [Figure1(d) and 1(e)] in our [Co/Pd] /Py film shows that the easy magnetized axis is still lying inplane, although the out-of-plane MOKE hysteresis loops show a high remanence ratio,indicating perpendicular component of magnetic moment in Py layer due to the strongmagnetostatic and exchange interaction between Py and (Co/Pd) layer.Figure 2(a)-2(d) show the image of magnetic domain of our [Co/Pd] /Py film, measured byMFM under a vertical magnetic field to the film plane at room temperature. In Figure 2(a), wepresent the MFM imaging at zero fields. One can notice that micron-sized domains exist inthe image. After the magnetic vertical field B=10 mT is applied [Figure 2(b)], the stripedomains with magnetization perpendicular to the film plane grow at the expense of domainsopposite to the film plane. As the vertical field increasing to B =18 mT, the labyrinth domainswith longer sizes are observed. When the field increase to 24 mT, all the stripes and labyrinthdomains completely transformed into random magnetic bubble with the average diameter ofabout 420 nm, as shown in Figure 2(d). When the vertical field exceed 25 mT, the magneticbubble annihilated and the film was saturated to a ferromagnetic state.The above results show that the labyrinth domain is sensitive to magnetic field. Next, wechanged the magnetized method and analysis the domain structure of our (Co/Pd) /Py filmagain. Before MFM observation, the sample was magnetized under an in-plane tilted field(roughly 3° to the plane as shown in Figure 3) with a small out-of plane component, then themagnetic field is removed, the whole process spend about 30 seconds. Figure 3(a)-(i) showthe MFM images after magnetized under different tilted field (from 0 mT to 1000 mT). EachMFM image was obtained from the same region of the sample, but not the same specific areaof the sample surface. Figure 3(a) shows the snap-shot of the MFM imaging at zero fields.One can notice that the stripe domains, the micron-sized domains (large white areas), andmagnetic bubbles coexist in the image. The images reveal that the out-of-plane magnetizationorientation averaged over the thickness of the magnetic domains through the variations ofdark/bright intensity contrast. After magnetized under 100-500 mT [Figure 3(b)-(d)], thelarger white domain disappeared due to the perpendicular component of tilted field, only leftthe labyrinth domain with a size of 2 μm. At the magnetized field of 800 mT [Figure 3(e)], allabyrinth domains completely transformed into disordered magnetic bubbles (white) with thediameter about 580 nm.The reason we didn’t show the labyrinth patterns is because the mazedomain still occurs in the range of 500 –
800 mT . As the magnetized field further increase,the size and number of bubbles gradually becomes smaller. At 950 mT, the white bubbleshave the smallest size of ~230 nm [Figure 3(f)]. After the magnetized field increase to 1000mT, the bubbles completely disappear [Figure 3(i)]. During our experiment, we also observedthe black bubbles [Figure 3(g), 900 mT] if we change the magnetized field to -3° direction tothe film plane (-z component field) and remove it before MFM observation. Further increaseof magnetized filed leads to the disappearance of some bubbles as shown in Figure 3(h). Thisresult indicates that the tilted field with small vertical magnetic field component can indeedtrigger bubbles in opposite polarity[20].The above results show that after magnetized at certain tilted magnetic field, magneticbubbles state in Py layer is triggered. To verify the thermal stability [29-32], we excited themagnetic bubble state at the tilted field of 650 mT then removed the magnetic field. At theremnant state, the MFM was used to observe the film again. After first observation, thesample is not taken away from the holder to ensure the next observation in the same area afterwaiting different time (1, 3, 5, 7, 9 hours). The domain structure under different relaxationtime is shown in Figure 4(a-f). No changes in the domain structure were observed after 9hours, indicating the bubble state is stable at zero fields within 9 hours. However, after threedays, the bubble state disappears (not show). It means that the magnetic bubble state in our(Co/Pd) /Py film is a metastable state. In addition, the labyrinth magnetic domains in our filmmay have chirality because we obtained the non-zero DMI coefficient D=0.38 mJ/m in our(Co/Pd) /Py film by using momentum-resolved BLS measurements. Since the thickness of Pyis 15 nm, we think that the interfacial DMI mainly comes from the interface of Co/Pd [seen atSupplementary Material S1] .To figure out the chirality of our magnetic bubble, micromagentic simulations wereperformed with the MuMax3 code. The simulated model includes a Py layer placed at the topof [Co/Pd] layers. The dimension of the sample is 1200 nm×1200 nm×20 nm, in which theNiFe film is 16 nm thick and Co is 4 nm thick. We take the following material parameters forthe soft NiFe and hard [Co/Pd] layers [25-26]: exchange constant mJA Py /1013 , mJA PdCo /106 . The perpendicular magnetic anisotropy constant of[Co/Pd], /101.8 mJK PdCo and saturation magnetization mAM Py /108 , mAM PdCo /105 were obtained from VSM measurement. To ensure the labyrinthdomain in Py layer, an effective non-zero perpendicular magnetic anisotropy K is taken /1032 mJ which originates from the dipolar interaction from the Co/Pd layers. Inthe simulation, we have to approximate the results of the phenomenological model. Thegeneral trend can be seen although the difference with the real system. The system isdiscretized into a mesh of dimensions 4 nm × × . Due to the non-zero DMI coefficient in our multilayers, the DM interaction isonsidering in the Co/Pd layers.Followed the experimental result, we first simulate the field-driven evolution of labyrinthdomain, as shown in Figure 5(a)-(h), to obtain the magnetic bubble state. At 200 mT verticalfield along +z direction, most labyrinth domain changed to bubble. The random isolatedbubble state absolutely emerged at the magnetic field of 250 mT. By increasing the magneticfield further to 300 mT, the size of bubble is reduced [Figure 5(g)]. If the magnetic field keepsincreasing further, the bubble annihilated gradually and the film was finally saturated to aferromagnetic state.By analyzing the magnetic moment distribution, we find there exists magnetic bubbles withtopological number 1, and -1, which is labeled by colour circle shown in figure 5(g). Theenlarged images of these skyrmion are also shown in figure 5(h). The existence of thesebubbles with topological number of 1 and -1 in Py layer may due to the influence of Co/Pdlayer with a non-zero DM interaction, which prefer to host a Neel skyrmion. To verify thisfact, the domain structure of Co layer is analyzed, which shows that, under the verticalmagnetic field, there exists Neel skyrmion. Due to the exchange interaction between Co andPy at the interface, the Neel skyrmion in Co layer changes to Bloch-like skyrmion (or bubble)in Py layer by orienting the magnetic moment out of the radial direction. Figure 5(i) showsthe magnetization distributions of several bubbles in upper Py , interface Py and interface Colayers. One can see that the magnetic moment transitional process of the bubble from Co toPy. We can see Neel skyrmions corresponding to the interface of Py and the Co layer. Frominterface Py to upper Py, Bloch skyrmions occurs in the picture. Our simulation resultsconfirm that bubble domains appearing in Co/Pd multilayer penetrate through the wholemultilayer system.Recently, understanding of the dynamics of magnetic bubble with non-zero topologicalnumber (skyrmions) under external fields is an important issue for their manipulation[33-35].Besides the lowest energy translation mode [34], breathing, clockwise and anticlockwiseradial modes have already been investigated [35]. However, due to the sensitivity of signal inultrathin film with perpendicular magnetic anisotropy, the microwave magnetic property isdifficult to obtain by vector network analyzer ferromagnetic resonance (VNA-FMR). In our[Co/Pd] /Py film, Py has a thickness of 15 nm, which makes it possible to measure thehigh-frequency properties of magnetic bubble in Py layer.Figure 6 shows the imaginary part of the permeability spectra of our [Co/Pd] /Py film underdifferent magnetic states. During the measurement, the static field and microwave field areapplied in the plane as indicated in Figure 6(a) and 6(b). Figure 6(a) shows the imaginary partof permeability spectra for the bubble state, which is obtained by magnetized the sampleunder titled field of 800 mT and then removed the field. During the measurement ofpermeability, the in-plane field is in the range of 0 to 200 Oe to ensure the bubble state doesnot destroyed by in-plane field. As comparison, no magnetic bubble state (after magnetizedthe sample under 1000 mT titled field) was also measured. The results are shown in Figure6(b). It can be seen that, for no bubble state, a sharp resonance peaks occurs at near 1 GHzunder zero magnetic field, which is in accordance with the in-plane uniform precession mode.As the magnetic field gradually increases, the resonance peak moves to the higher frequency.At 200 Oe field, the maximum frequency of 4.2 GHz is obtained. However, for multilayerswith bubble state in Py, the permeability spectrum shows an asymmetric broadening peak at 0agnetic fields. With increasing the magnetic field, the resonance peak also moves to thehigher frequency. At higher magnetic field (like larger than 100 Oe), the asymmetry of spectradisappear. The above results indicate that the in-plane uniform precession mode and otherresonance modes originated from the non-trivial spin texture of magnetic bubble coexist. Thesuperposition of two kinds of resonance modes broadens the line width.To analyse the resonance mode carefully, the frequency of in-plane uniform precessionmode (Kittel model) versus the magnetic field is fitting, which is shown in Figure 7(a) and 7(e)for bubble state and saturated state, respectively. The main resonance frequency f r can bedetermined by the applied field H app according to the Kittel equation [36] as follows, ))(4(2 appkappkr HHHHMf (1)Where γ/2π=2.94 GHz/kOe is the gyromagnetic ratio, 4π M is the magnetization of the filmsalong the applied field H app . Generally, 4π M is much bigger than H app and H k . Thus the aboveequation (1) can be simplified to be appkkkr HHMHHMf )24()4()2( (2)Based on equation (2), the experimental results can be qualitatively fitting [Figure 7(a) and7(e)]. We can obtain that 4π M eff = 9.9 kGs, H k = 4.5 Oe for bubble states and 4π M eff =10.2 kGs, H k =10 Oe for in-plane saturated state, respectively. Due to the spin non-uniform precession atbubble states, the resonance field H k = 4.5 Oe is smaller than the resonance field H k =10.2 Oeof no bubble states. Since other resonance mode exist besides the kittel mode, so the mainpeak (peak1) is fitted with Gaussian formula. Except the Kittel peak, another small peak(peak2) is obtained. Figure 7(c) and 7(d) show the resonance absorption peak of bubble statesat the field H app =100 Oe and H app =200 Oe. It can be seen that the linewidth of the main peakbecomes narrower and the small peak is not obvious when the field increases. We concludethat the main peak and other mode peak do not move at the same speed with the field. Inaddition, from the results reported in the literature, we judge that this small peak maycorrespond to the precession mode of the clockwise (or counterclockwise) of the chiral bubblestate. Figure 7(f)-7(h) show resonance absorption peak of in-plane saturated state. Thecorresponding sweeping fields are 0 Oe, 100 Oe and 200 Oe. One can notice that the singlepeak Lorentz fitting well corresponding to the experimental results. Our results demonstrate that the emergence of metastable bubbles at remnant can be tunedby the tilted field (roughly 3°) along the thin film plane at room temperature in anexchange-coupled Py/[Co/Pd] multilayer. Micromagnetic simulation shows that the magneticmoment distribution of magnetic bubbles is same as the Bloch type with the topologicalnumber of 1 and -1. Moreover, the high frequency properties of bubble state was studied byVNA-FMR techniques. As the dc magnetic field changes from 0 to 200 Oe, the resonancepeak include the Kittel mode and in-plane clockwise or anticlockwise mode of chiral bubblestate. We thus provide an effective and promising method for high frequency technologicalexplorations of bubble-based devices. cknowledgments This work was supported by the National Natural Science Foundation of China (Nos. 11574121,51771086).
References [1] Langer J. S, Annals of physics:54,258-275 (1969).[2] Langer J. S, Physica 73, 61-72 (1974).[3] Rößler, U. K., Bogdanov A. N, and Pfleiderer C, Nature 442, 797–801 (2006).[4] Muhlbauer, Binz B, Jonietz F, Pfleiderer C, Rosch A, Neubauer A, Georgii R, Böni P,Science 323,915–919 (2009).[5] Yu X. Z, Onose Y, Kanazawa N, Park J. H, Han J. H, Matsui Y, Nagaosa N and Tokura Y,Nature 465, 901–904 (2010).[6] Milde P, Köhler D, Seidel J, Eng L, Bauer A, Chacon A, Science 2013;340:1076-80.[7] Yu X. Z, Kanazawa N, Onose Y, Kimoto K, Zhang W, Ishiwata S, Nature materials2011;10.[8] Slonczewski J. C, Malozemoff A. P, and Giess E. A, Appl. Phys. Lett. 24, 396 (1974).[9] Moutafis C , Komineas S, Vaz C. A. F , Bland J. A. C , Shima T, Seki T, and Takanashi K,Phys. Rev. B , 104426 (2007).[10] Masaaki Tanaka, Hiroki Kanazawa, Sho Sumitomo, Syuta Honda, Ko Mibu, andHiroyuki Awano, Applied Physics Express, 8, 073002 (2015).[11] Naoto Nagaosa and Yoshinori Tokura, Nature nanotechnology. 8, 899–911 (2013).[12] Sampaio J, Cros V, Rohart S, Thiaville A and Fert A, Nature Nanotechnology8, 839–844 (2013).[13] Wanjun Jiang, Pramey Upadhyaya, Wei Zhang, Guoqiang Yu, M. Benjamin Jungfleisch,Frank Y. Fradin, John E. Pearson, Yaroslav Tserkovnyak, Kang L. Wang, Olle Heinonen,Suzanne G. E. te Velthuis, Axel Hoffmann1, Science 349, 283 (2015).[14] Xiuzhen Yu, Maxim Mostovoy , Yusuke Tokunaga , Weizhu Zhang , Koji Kimoto ,Yoshio Matsui, Yoshio Kanekod , Naoto Nagaosa , and Yoshinori Tokura,PNAS 109 (23) 8856-8860 (2012).[15] Nakajima H, Kawase H, Kurushima K, Kotani A, Kimura T, and Mori S, Phys. Rev. B ,024431 (2017).[16] Xiuzhen Yu, Yusuke Tokunaga, Yasujiro Taguchi, and Yoshinori Tokura, Advanced.Materials, 29, 1603958 (2017).[17] Fukumura T, Sugawara H, Hasegawa T, Tanaka, Sakaki H, Kimura T, Tokura Y,Science 284:1969–1971(1999).[18] Rougemaille N, Schmid A. K, Eur. Phys. J. Appl. Phys. 50, 20101 (2010)[19] Wu J, Carlton D, Park J.S, Meng Y, Arenholz E, Doran A, Young A.T, Scholl A, HwangC, Zhao H.W, Bokor J, Qiu Z.Q, Nature Physics, 7 303-306 (2011) .[20] Zhaogang Qin, Ying Wang, Shimeng Zhu, Chendong Jin, Jiecai Fu, Qingfang Liu, andJiangwei Cao. ACS Appl. Mater. Interfaces,Just Accepted Manuscript (2018).[21] Suzuki T, J Magn Magn Mater 31-34:1009–1010 (1983).[22] Shawn D. Pollard, Joseph A. Garlow, Jiawei Yu, Zhen Wang, Yimei Zhu and Hyunsooang, Nature Communications 8, 14761 (2017).[23] Anjan Soumyanarayanan, Raju M, Gonzalez Oyarce A. L, Anthony K. C. Tan, Mi-YoungIm, Petrović A. P, Pin Ho, Khoo K. H, Tran M, Gan C. K, Ernult F and Panagopoulos C,Nature Materials 16, 898–904 (2017).[24] Xiaoyu Li, Xiaojun Sun, Jianbo Wang, Qingfang Liu, J Magn Magn Mater,377 :142-146(2015).[25] Bollero A, Buda-Prejbeanu L. D, Baltz V, Sort J, Rodmacq B, and Dieny B, Phys. Rev.B,73, 144407 (2006).[26] Larysa Tryputen, Feng Guo, Frank Liu, T. N. Anh Nguyen, Majid S. Mohseni, SunjaeChung, Yeyu Fang, Johan Åkerman, R. D. McMichael, and Caroline A. Ross, Phys. Rev. B,91, 014407 (2015).[27] Sort J, Popa A, Rodmacq B, and Dieny B, Phys. Rev. B 70, 174431 (2004).[28] Bollero A, Baltz V, Buda-PrejbeanuL. D, Rodmacq B, and Dieny B, Phys. Rev. B 84,094423 (2011).[29] Hiroshi Oike, Akiko Kikkawa, Naoya Kanazawa, Yasujiro Taguchi, Masashi Kawasaki,Yoshinori Tokura and Fumitaka Kagawa, Nature Physics 12, 62–66 (2016).[30] David Cortés-Ortuño, Weiwei Wang, Marijan Beg, Ryan A. Pepper, Marc-AntonioBisotti, Rebecca Carey, Mark Vousden, Thomas Kluyver, Ondrej Hovorka and HansFangohr,Scientific Reports 7, 4060 (2017).[31] Koya Makino, Johannes D. Reim, Daiki Higashi, Daisuke Okuyama, Taku J. Sato,Yusuke Nambu, Elliot P. Gilbert, Norman Booth, Shinichiro Seki, and Yoshinori Tokura,Phys. Rev. B 95, 134412 (2017).[32] Lingyao Kong and Jiadong Zang, Phys. Rev. Lett. 111, 067203 (2013).[33] Olga Petrova and Oleg Tchernyshyov, Phys. Rev. B 84, 214433 (2011).[34] Masahito Mochizuki, Phys. Rev. Lett. 108, 017601 (2012).[35] Montoya S. A, Couture S, ChessJ. J, Lee J. C. T, Kent N, M.-Y. Im, Kevan S. D, FischerP, McMorran B. J, Roy S, Lomakin V, and Fullerton E. E, Phys. Rev. B , 224405 (2017).[36] Charles Kittel, Phys. Rev. 73, 155 (1948). Figure captions
Figure 1. (a) Schematic illustration of exchange-coupled Ta/Pd/[Co/Pd] /NiFe/Ta multilayersstructure. The arrows represent the direction of magnetic moment arrangement. (b) Out-ofplane and in-plane hysteresis loop of (Co/Pd) multilayers measured by VSM. (c)Out-of-plane and in-plane hysteresis loop of [Co/Pd] /Py multilayers measured by VSM. Theinsert graph shows the amplification of in-plane hysteresis loop of [Co/Pd] /Py multilayers. (d)polar MOKE M–H loops for [Co/Pd] , Py(15 nm) and [Co/Pd] /Py(15 nm). (e) transverseMOKE M–H loops for [Co/Pd] , Py(15 nm) and [Co/Pd] /Py(15 nm).Figure 2. (a)-(d) MFM images for [Co/Pd] /Py(15 nm) multilayers at different vertical fields.(a) 0 mT, (b) 10 mT, (c) 18 mT, (d) 24 mT. Especially, the red and green circles have nospecial significance in the Figure 2d. They only represent bubble domains in order to seesome clearly. All the scale bar corresponds to 1 m .igure 3. (a)-(i) MFM imaging of the field-dependent magnetic domain morphology of our[Co/Pd] /Py(15nm) multilayers at room temperature. The images are captured aftermagnetized the sample under titled field then removed the field before MFM observation.From 3(a) to 3(i) , the values of the magnetic tilted field were respectively: (a) 0 mT ,(b)100mT, (c) 300 mT ,(d) 500 mT,(e) 800 mT, (f) 950 mT , (g) -900 mT, (h) -950 mT ,(i) 1000 mT.All the scale bar corresponds to 2 m .Figure 4. Bubble stability at room temperature. MFM imaging of domain states when waiteda serial of times after removing the magnetic field. (a) 0h, (b) 1h ,(c) 3h, (d) 5h , (e) 7h, (f) 9h.All the scale bar corresponds to 2 m .Figure 5. Micromagnetic simulation of (Co/Pd)/Py film. (a)-(h) magnetization configurationsunder different vertical magnetic field. (a) the initial labyrinth domain states at B z = 0 mT, (b)50 mT, (c) 100 mT, (d) 150 mT , (e) 200 mT, (f) 250 mT, (g) 300 mT. (h) the magneticmoment distribution of specific magnetic bubbles. (i) the magnetization distributions of theupper Py (left), interface Py (middle) and interface Co(right). All the scale bar corresponds to200 nm.Figure 6. The imaginary part of the permeability spectra as a function of frequency underdifferent dc magnetic field for (a) bubble state in Py (b) in-pane saturated state in Py.Figure 7. Imaginary part of the permeability spectra analysis. (a)-(d) Bubble state (e)-(h)in-plane saturated state. The red solid lines are linear fitting curves. (a) and (e) the appliedmagnetic field H app dependence on the square of resonance frequency, (b) and (f) DC field H app =0 Oe, (c) and (g) H app = 100 Oe, (d) and (h) H appapp