Article | Open | Published:

# Ultra-low damping insulating magnetic thin films get perpendicular

## Abstract

A magnetic material combining low losses and large perpendicular magnetic anisotropy (PMA) is still a missing brick in the magnonic and spintronic fields. We report here on the growth of ultrathin Bismuth doped Y3Fe5O12 (BiYIG) films on Gd3Ga5O12 (GGG) and substituted GGG (sGGG) (111) oriented substrates. A fine tuning of the PMA is obtained using both epitaxial strain and growth-induced anisotropies. Both spontaneously in-plane and out-of-plane magnetized thin films can be elaborated. Ferromagnetic Resonance (FMR) measurements demonstrate the high-dynamic quality of these BiYIG ultrathin films; PMA films with Gilbert damping values as low as 3 × 10−4 and FMR linewidth of 0.3 mT at 8 GHz are achieved even for films that do not exceed 30 nm in thickness. Moreover, we measure inverse spin hall effect (ISHE) on Pt/BiYIG stacks showing that the magnetic insulator’s surface is transparent to spin current, making it appealing for spintronic applications.

## Introduction

Spintronics exploits the electron’s spin in ferromagnetic transition metals for data storage and data processing. Interestingly, as spintronics codes information in the angular momentum degrees of freedom, charge transport and therefore the use of conducting materials is not a requirement, opening thus electronics to insulators. In magnetic insulators (MI), pure spin currents are described using excitation states of the ferromagnetic background named magnons (or spin waves). Excitation, propagation and detection of magnons are at the confluent of the emerging concepts of magnonics1,2, caloritronics3, and spin-orbitronics4. Magnons, and their classical counterpart, the spin waves (SWs), can carry information over distances as large as millimeters in high-quality thick YIG films, with frequencies extending from the GHz to the THz regime5,6,7. The main figure of merit for magnonic materials is the Gilbert damping α1,5,8 which has to be as small as possible. This makes the number of relevant materials for SW propagation quite limited and none of them has yet been found to possess a large enough perpendicular magnetic anisotropy (PMA) to induce spontaneous out-of-plane magnetization. We report here on the Pulsed Laser Deposition (PLD) growth of ultra-low loss MI nanometers-thick films with large PMA: Bi substituted Yttrium Iron Garnet (BixY3-xFe5O12 or BiYIG) where tunability of the PMA is achieved through epitaxial strain and Bi doping level. The peak-to-peak FMR linewidth (that characterize the losses) can be as low as μ0ΔHpp = 0.3 mT at 8 GHz for 30 nm thick films. This material thus opens new perspectives for both spintronics and magnonics fields as the SW dispersion relation can now be easily tuned through magnetic anisotropy without the need of a large bias magnetic field. Moreover, energy efficient data storage devices based on magnetic textures existing in PMA materials like magnetic bubbles, chiral domain walls, and magnetic skyrmions would benefit from such a low loss material for efficient operation9.

The study of micron-thick YIG films grown by liquid phase epitaxy (LPE) was among the hottest topics in magnetism few decades ago. At this time, it has been already noticed that unlike rare earths (Thulium, Terbium, Dysprosium …) substitutions, Bi substitution does not overwhelmingly increase the magnetic losses10,11 even though it induces high uniaxial magnetic anisotropy12,13,14. Very recently, ultra-thin MI films showing PMA have been the subject of an increasing interest:15,16 Tm3Fe5O12 or BaFe12O19 (respectively a garnet and an hexaferrite) have been used to demonstrate spin-orbit-torque magnetization reversal using a Pt over-layer as a source of spin current4,17,18. However, their large magnetic losses prohibit their use as a spin-wave medium (reported value of μ0ΔHpp of TIG is 16.7 mT at 9.5 GHz)19. Hence, whether it is possible to fabricate ultra-low loss thin films with a large PMA that can be used for both magnonics and spintronics applications remains to be demonstrated. Indeed, not only low losses are important for long range spin wave propagation but they are also necessary for spin transfer torque oscillators (STNOs) as the threshold current scales with the Gilbert damping20.

In the quest for the optimal material platform, we explore here the growth of Bi doped YIG ultra-thin films using PLD with different substitution; BixY3-xIG (x = 0.7, 1, and 1.5) and having a thickness ranging between 8 and 50 nm. We demonstrate fine tuning of the magnetic anisotropy using epitaxial strain and measure ultra low Gilbert damping values (α = 3 × 10−4) on ultrathin films with PMA.

## Results

### Structural and magnetic characterizations

The two substrates that are used are gallium gadolinium garnet (GGG), which is best lattice matched to pristine YIG and substituted GGG (sGGG) which is traditionally used to accommodate substituted YIG films for photonics applications. The difference between Bi and Y ionic radii (rBi = 113 pm and rY = 102 pm)21 leads to a linear increase of the BixY3-xIG bulk lattice parameter with Bi content (Fig. 1a, b). In Fig. 1, we present the (2θω) X-ray diffraction patterns (Fig. 1c, d) and reciprocal space maps (RSM) (Fig. 1e, f) of BiYIG on sGGG(111) and GGG(111) substrates, respectively. The presence of (222) family peaks in the diffraction spectra shown in Fig. 1b, c is a signature of the films’ epitaxial quality and the presence of Laue fringes attests the coherent crystal structure existing over the whole thickness. As expected, all films on GGG are under compressive strain, whereas films grown on sGGG exhibit a transition from a tensile (for x = 0.7 and 1) towards a compressive (x = 1.5) strain. Reciprocal space mapping of these BiYIG samples shown in Fig. 1e, f evidences the pseudomorphic nature of the growth for all films, which confirms the good epitaxy.

The static magnetic properties of the films have been characterized using SQUID magnetometry, Faraday rotation measurements and Kerr microscopy. As the Bi doping has the effect of enhancing the magneto-optical response22,23,24, we measure on average a large Faraday rotation coefficients reaching up to θF = −3° μm−1 @ 632 nm for x = 1 Bi doping level and 15 nm film thickness. Chern et al.25 performed PLD growth of BixY3-xIG on GGG and reported an increase of $${\theta}_{\mathrm{F}} = - 1.9^{\circ} {\upmu}{\mathrm{m}}^{-1}$$ per Bi substitution x@ 632 nm. The Faraday rotation coefficients we find are slightly larger and may be due to the much lower thickness of our films as θF is also dependent on the film thickness26. The saturation magnetization (Ms) remains constant for all Bi content (see Table 1) within the 10% experimental errors. We observe a clear correlation between the strain and the shape of the in-plane and out-of-plane hysteresis loops reflecting changes in the magnetic anisotropy. While films under compressive strain exhibit in-plane anisotropy, those under tensile strain show a large out-of-plane anisotropy that can eventually lead to an out-of-plane easy axis for x = 0.7 and x = 1 grown on sGGG. The transition can be either induced by changing the substrate (Fig. 2a) or the Bi content (Fig. 2b) since both act on the misfit strain. We ascribe the anisotropy change in our films to a combination of magneto-elastic anisotropy and growth-induced anisotropy, this later term being the dominant one (see Supplementary Note 1).

In Fig. 2c, we show the magnetic domains structures at remanance observed using polar Kerr microscopy for Bi1Y2IG films after demagnetization: µm-wide maze-like magnetic domains demonstrates unambiguously that the magnetic easy axis is perpendicular to the film surface. We observe a decrease of the domain width (Dwidth) when the film thickness (tfilm) increases as expected from magnetostatic energy considerations. In fact, as Dwidth is several orders of magnitude larger than tfilm, a domain wall energy of σDW 0.7 and 0.65 mJ m−2 (for x = 0.7 and 1 Bi doping) can inferred using the Kaplan and Gerhing model27 (the fitting procedure is detailed in the Supplementary Note 2).

### Dynamical characterization and spin transparency

The most striking feature of these large PMA films is their extremely low magnetic losses that we characterize using Ferromagnetic Resonance (FMR) measurements. First of all, we quantify by in-plane FMR the anisotropy field HKU deduced from the effective magnetization (Meff): HKU=MS–Meff (the procedure to derive Meff from in-plane FMR is presented in Supplementary Note 3). HKU values for BiYIG films with different doping levels grown on various substrates are summarized in Table 1. As expected from out-of-plane hysteresis curves, we observe different signs for HKU. For spontaneously out-of-plane magnetized samples, HKU is positive and large enough to fully compensate the demagnetizing field while it is negative for in-plane magnetized films. From these results, one can expect that fine tuning of the Bi content allows fine tuning of the effective magnetization and consequently of the FMR resonance conditions. We measure magnetic losses on a 30 nm thick Bi1Y2IG//sGGG film under tensile strain with PMA (Fig. 3a). We use the FMR absorption line shape to extract the peak-to-peak linewidth (ΔHpp) at different out-of-plane angle for a 30 nm thick perpendicularly magnetized Bi1Y2IG//sGGG film at 8 GHz (Fig. 3b). This yields an optimal value of μ0ΔHpp as low as 0.3 mT (Fig. 3c) for 27° out-of-plane polar angle. We stress here that state-of-the-art PLD grown YIG//GGG films exhibit similar values for ΔHpp at such resonant conditions28. This angular dependence of ΔHpp that shows pronounced variations at specific angle is characteristic of a two magnons scattering relaxation process with few inhomogeneities29. The value of this angle is sample dependent as it is related to the distribution of the magnetic inhomogeneities. The dominance in our films of those two intrinsic relaxation processes (Gilbert damping and two-magnons scattering) confirms the high films quality. We also derive the damping value of this film (Fig. 3d) by selecting the lowest linewidth (corresponding to a specific out-of-plane angle) at each frequency, the spread of the out-of-plane angle is ±3.5° around 30.5°. The obtained Gilbert damping value α = 3 × 10−4 and the peak-to-peak extrinsic linewidth μ0ΔH0 = 0.23 mT are comparable to the one obtained for the best PLD grown YIG//GGG nanometer thick films28 (α = 2 × 10−4). For x = 0.7 Bi doping, the smallest observed FMR linewidth is 0.5 mT at 8 GHz.

The low magnetic losses of BiYIG films could open new perspectives for magnetization dynamics control using spin-orbit torques20,30,31. For such phenomenon interface transparency to spin current is then the critical parameter which is defined using the effective spin-mixing conductance (G↑↓). We use spin pumping experiments to estimate the increase of the Gilbert damping due to Pt deposition on Bi1Y2IG films. The spin-mixing conductance can thereafter be calculated using $$G_{ \uparrow \downarrow } = \frac{{4\pi M_{\mathrm{s}}t_{{\mathrm{film}}}}}{{g_{{\mathrm{eff}}}\mu _{\mathrm{B}}}}\left( {{\it{\Delta }}\alpha } \right)$$ where Ms and tfilm are the BiYIG magnetization and thickness, geff is the effective Landé factor (geff = 2), μB is the Bohr magneton and Δα is the increase in the Gilbert damping constant induced by the Pt top layer. We obtain G↑↓ = 3.9 × 1018 m−2 which is comparable to what is obtained on PLD grown YIG//GGG systems28,32,33. Consequently, the doping in Bi should not alter the spin orbit-torque efficiency and spin-torque devices made out of BiYIG will be as energy efficient as their YIG counterpart. To further confirm that spin current crosses the Pt/BiYIG interface, we measure Inverse Spin Hall Effect (ISHE) in Pt for a Pt/Bi1.5Y1.5IG(20 nm)//sGGG in-plane magnetized film (to fulfill the ISHE geometry requirements the magnetization needs to be in-plane and perpendicular to the measured voltage). We measure a characteristic voltage peak due to ISHE that reverses its sign when the static in-plane magnetic field is reversed (Fig. 4). We emphasize here that the amplitude of the signal is similar to that of Pt/YIG//GGG in the same experimental conditions.

## Conclusion

In summary, this new material platform will be highly beneficial for magnon-spintronics and related research fields like caloritronics. In many aspects, ultra-thin BiYIG films offer new leverages for fine tuning of the magnetic properties with no drawbacks compared to the reference materials of these fields: YIG. BiYIG with its higher Faraday rotation coefficient (almost two orders of magnitude more than that of YIG) will increase the sensitivity of light based detection techniques that can be used (Brillouin light spectroscopy (BLS) or time resolved Kerr microscopy34). Innovative schemes for on-chip magnon-light coupler could be now developed bridging the field of magnonics to the one of photonics. From a practical point of view, the design of future active devices will be much more flexible as it is possible to easily engineer the spin waves dispersion relation through magnetic anisotropy tuning without the need of large bias magnetic fields. For instance, working in the forward volume waves configuration comes now cost free, whereas in standard in-plane magnetized media one has to overcome the demagnetizing field. As the development of PMA tunnel junctions was key in developing today scalable MRAM technology, likewise, we believe that PMA in nanometer-thick low loss insulators paves the path to new approaches where the magnonic medium material could also be used to store information locally combining therefore the memory and computational functions, a most desirable feature for the brain-inspired neuromorphic paradigm.

## Methods

### Pulsed laser deposition (PLD) growth

The PLD growth of BiYIG films is realized using stoichiometric BiYIG target. The laser used is a frequency tripled Nd:YAG laser (λ= 355 nm), of a 2.5 Hz repetition rate and a fluency varying from 0.95 to 1.43 J cm−2 depending upon the Bi doping in the target. The distance between target and substrate is fixed at 44 mm. Prior to the deposition the substrate is annealed at 700 °C under 0.4 mbar of O2. For the growth, the pressure is set at 0.25 mbar O2 pressure. The optimum growth temperature varies with the Bi content from 400 to 550 °C. At the end of the growth, the sample is cooled down under 300 mbar of O2.

### Structural characterization

An Empyrean diffractometer with Kα1 monochromator is used for measurement in Bragg-Brentano reflection mode to derive the (111) interatomic plan distance. Reciprocal Space Mapping is performed on the same diffractometer and we used the diffraction along the (642) plane direction which allow to gain information on the in-plane epitaxy relation along [20-2] direction.

### Magnetic characterization

A quantum design SQUID magnetometer was used to measure the films’ magnetic moment (Ms) by performing hysteresis curves along the easy magnetic direction at room temperature. The linear contribution of the paramagnetic (sGGG or GGG) substrate is linearly subtracted.

Kerr microscope (Evico Magnetics) is used in the polar mode to measure out-of-plane hysteresis curves at room temperature. The same microscope is also used to image the magnetic domains structure after a demagnetization procedure. The spatial resolution of the system is 300 nm.

A broadband FMR setup with a motorized rotation stage was used. Frequencies from 1 to 20 GHz have been explored. The FMR is measured as the derivative of microwave power absorption via a low frequency modulation of the DC magnetic field. Resonance spectra were recorded with the applied static magnetic field oriented in different geometries (in-plane or tilted of an angle θ out-of the stripline plane). For out-of-plane magnetized samples the Gilbert damping parameter has been obtained by studying the linewidth angular dependence. The procedure assumes that close to the minimum linewidth (Fig. 3a) most of the linewidth angular dependence is dominated by the inhomogeneous broadening, thus optimizing the angle for each frequency within few degrees allows to estimate better the intrinsic contribution. To do so we varied the out-of-plane angle of the static field from 27° to 34° for each frequency and we select the lowest value of ΔHpp.

For Inverse spin Hall effect measurements, the same FMR setup was used, however here the modulation is no longer applied to the magnetic field but to the RF power at a frequency of 5 kHz. A Stanford Research SR860 lock-in was used a signal demodulator.

### Data availability

The data that support the findings of this study are available within the article or from the corresponding author upon reasonable request.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Change history

• ### 05 September 2018

This Article was originally published without the accompanying Peer Review File. This file is now available in the HTML version of the Article; the PDF was correct from the time of publication.

## References

1. 1.

Karenowska, A. D., Chumak, A. V., Serga, A. A. & Hillebrands, B. Magnon spintronics. Handb. Spintron. 11, 1505–1549 (2015).

2. 2.

Chumak, A. V., Serga, A. A. & Hillebrands, B. Magnonic crystals for data processing. J. Phys. D Appl. Phys. 50, 244001–20 (2017).

3. 3.

Bauer, G. E. W., Saitoh, E. & Van Wees, B. J. Spin caloritronics. Nat. Mater. 11, 391–399 (2012).

4. 4.

Li, P. et al. Spin-orbit torque-assisted switching in magnetic insulator thin films with perpendicular magnetic anisotropy. Nat. Commun. 7, 12688 (2016).

5. 5.

Serga, A. A., Chumak, A. V. & Hillebrands, B. YIG magnonics. J. Phys. D Appl. Phys. 43, 264002–16 (2010).

6. 6.

Seifert, T. et al. Launching magnons at the terahertz speed of the spin Seebeck effect. Prepr. http//arxiv.org/abs/1709.00768 (2017).

7. 7.

Onbasli, M. C. et al. Pulsed laser deposition of epitaxial yttrium iron garnet films with low Gilbert damping and bulk-like magnetization. APL Mater. 2, 106102 (2014).

8. 8.

Zakeri, K. et al. Spin dynamics in ferromagnets: Gilbert damping and two-magnon scattering. Phys. Rev. B 76, 104416 (2007).

9. 9.

Fert, A., Cros, V. & Sampaio, J. Skyrmions on the track. Nat. Nanotechnol. 8, 152–156 (2013).

10. 10.

Vittoria, C., Lubitz, P., Hansen, P. & Tolksdorf, W. FMR linewidth measurements in bismuth‐substituted YIG. J. Appl. Phys. 57, 3699–3700 (1985).

11. 11.

Sposito, A., Gregory, S. A., de Groot, P. A. J. & Eason, R. W. Combinatorial pulsed laser deposition of doped yttrium iron garnet films on yttrium aluminium garnet. J. Appl. Phys. 115, 53102 (2014).

12. 12.

Fratello, V. J., Slusky, S. E. G., Brandle, C. D. & Norelli, M. P. Growth-induced anisotropy in bismuth: rare-earth iron garnets. J. Appl. Phys. 60, 2488–2497 (1986).

13. 13.

Popova, E. et al. Magnetic anisotropies in ultrathin bismuth iron garnet films. J. Magn. Magn. Mater. 335, 139–143 (2013).

14. 14.

J. Ben Youssef Ph.D Thesis. Characterisation and physical study of bismuth substituted thin garnet films grown by liquid phase epitaxy (LPE). (1989).

15. 15.

Fu, J. et al. Epitaxial growth of Y3Fe5O12 thin films with perpendicular magnetic anisotropy. Appl. Phys. Lett. 110, 202403 (2017).

16. 16.

Wang, C. T. et al. Controlling the magnetic anisotropy in epitaxial Y3Fe5O12 films by manganese doping. Phys. Rev. B 96, 224403 (2017).

17. 17.

Avci, C. O. et al. Fast switching and signature of efficient domain wall motion driven by spin-orbit torques in a perpendicular anisotropy magnetic insulator/Pt bilayer. Appl. Phys. Lett. 111, 072406 (2017).

18. 18.

Quindeau, A. et al. Tm3Fe5O12/Pt heterostructures with perpendicular magnetic anisotropy for spintronic applications. Adv. Electron. Mater. 3, 1600376 (2017).

19. 19.

Tang, C. et al. Anomalous hall hysteresis in Tm3Fe5O12/Pt with strain-induced perpendicular magnetic anisotropy. Phys. Rev. B 94, 1–5 (2016).

20. 20.

Collet, M. et al. Generation of coherent spin-wave modes in yttrium iron garnet microdiscs by spin–orbit torque. Nat. Commun. 7, 10377 (2016).

21. 21.

Hansen, P., Klages, C.‐P., & Witter, K. Magnetic and magneto‐optic properties of praseodymium‐ and bismuth‐substituted yttrium iron garnet films. J. Appl. Phys. 60, 721–727 (1986).

22. 22.

Robertson, J. M., Wittekoek, S., Popma, T. J. A., & Bongers, P. F. Preparation and optical properties of single crystal thin films of bismuth substituted iron garnets for magneto-optic applications. Appl. Phys. 2, 219–228 (1973).

23. 23.

Hansen, P., Witter, K., & Tolksdorf, W. Magnetic and magneto-optic properties of lead- and bismuth-substituted yttrium iron garnet films. Phys. Rev. B 27, 6608 (1983).

24. 24.

Matsumoto, K. et al. Enhancement of magneto-optical Faraday rotation by bismuth substitution in bismuth and aluminum substituted yttrium–iron–garnet single-crystal films grown by coating gels. J. Appl. Phys. 71, 2467 (1992).

25. 25.

Chern, M.-Y. & Liaw, J.-S. Study of BixY3-xFe5O12 thin films grown by pulsed laser deposition. Jpn. J. Appl. Phys. 36, 1049 (1997).

26. 26.

Kahl, S., Popov, V. & Grishin, A. M. Optical transmission and Faraday rotation spectra of a bismuth iron garnet film. J. Appl. Phys. 94, 5688–5694 (2003).

27. 27.

Kaplan, B. & Gehring, G. A. The domain structure in ultrathin magnetic films. J. Magn. Magn. Mater. 128, 111–116 (1993).

28. 28.

Hamadeh, A. et al. Full control of the spin-wave damping in a magnetic insulator using spin-orbit torque. Phys. Rev. Lett. 113, 197203 (2014).

29. 29.

Hurben, M. J. & Patton, C. E. Theory of two magnon scattering microwave relaxation and ferromagnetic resonance linewidth in magnetic thin films. J. Appl. Phys. 83, 4344–4365 (1998).

30. 30.

Xiao, J. & Bauer, G. E. W. Spin-wave excitation in magnetic insulators by spin-transfer torque. Phys. Rev. Lett. 108, 217204 (2012).

31. 31.

Safranski, C. et al. Spin caloritronic nano-oscillator. Nat. Commun. 8, 117 (2017).

32. 32.

Takahashi, R. et al. Electrical determination of spin mixing conductance at metal/insulator interface using inverse spin Hall effect. J. Appl. Phys. 111, 07C307 (2012).

33. 33.

Heinrich, B. et al. Spin pumping at the magnetic insulator (YIG)/normal metal (Au) interfaces. Phys. Rev. Lett. 107, 66604 (2011).

34. 34.

Stigloher, J. et al. Snell’s law for spin waves. Phys. Rev. Lett. 117, 1–5 (2016).

## Acknowledgements

We acknowledge J. Sampaio for preliminary Faraday rotation measurements and N. Reyren and A. Barthélémy for fruitful discussions. This research was supported by the ANR Grant ISOLYIG (ref 15-CE08-0030-01). L.S. is partially supported by G.I.E III-V Lab. France.

## Author information

### Affiliations

1. #### Unité Mixte de Physique CNRS, Thales, Univ. Paris-Sud, Université Paris Saclay, 91767, Palaiseau, France

• Lucile Soumah
• , Cécile Carrétéro
• , Eric Jacquet
• , Paolo Bortolotti
• , Vincent Cros
2. #### LABSTICC, UMR 6285 CNRS, Université de Bretagne Occidentale, 29238, Brest, France

• Nathan Beaulieu
•  & Jamal Ben Youssef
3. #### Thales Research and Technology, Thales, 91767, Palaiseau, France

• Lilia Qassym
•  & Richard Lebourgeois

### Contributions

L.S. performed the growth, all the measurements, the data analysis and wrote the manuscript with A.A., N.B. and J.B.Y. conducted the quantitative Faraday Rotation measurements and participated in the FMR data analysis. L.Q. and fabricated the PLD targets. R.L. supervised the target fabrication and participated in the design of the study. E.J. participated in the optimization of the film growth conditions. C.C. supervised the structural characterization experiments. A.A. conceived the study and was in charge of overall direction. P.B. and V.C. contributed to the design and implementation of the research. All authors discussed the results and commented on the manuscript.

### Competing interests

The authors declare no competing interests.