Open Access
Add to BibSonomy
Abstract
We present an analysis of the optical and magnetorefractive properties of a series of Ni81Fe19/Au multilayers grown by magnetron sputtering as a function of the Au spacer thickness. The multilayers reach giant magneto resistance values of 4% with low magnetic fields for 2.3 nm Au spacers. The experimental results are well described taking only into account the contribution of the conduction electrons with spin dependent scattering times and different electron concentrations for the two spins. It is shown that the spectral response of the magnetorefractive effect depends strongly on the spin up vs. spin down ratio of both, scattering time and electron concentration. This dependence can be used to optimize mid infrared modulation of optical devices.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Metals play an important role in the development of mid- and far-infrared optical components and devices. For example, frequency selective surfaces, polarizers, and antennas, to name a few [1–5], all have in common that some of their elements are metals. To make these elements active, the use of metallic structures whose optical properties can be dynamically modified, for example with an external agent, is quite appealing. In this sense, it has recently been demonstrated [6] that metamaterial platforms based on Giant Magneto Resistance (GMR) Ni81Fe19/Au multilayers allow for mid-IR modulation under very weak magnetic fields. The physical mechanism responsible of this modulation is the Magneto Refractive Effect (MRE) [7], which consists on a magnetic field driven change in the optical properties of the system mediated by the spin related change in the electrical conductivity. Discovered in the nineties, the MRE has been observed in different types of structures and has also been used for contactless characterization of magnetotransport properties [8]. To name a few examples, the MRE effect has been studied in a wide variety of metallic and oxide materials systems in the form of thin films, multilayers or granular alloys [9–17].
From an applied point of view, not only the absolute MRE value is relevant, but also the fact that it occurs at moderate magnetic fields, allowing realistic device realization. In this sense, the Ni81Fe19/Au multilayer system is an excellent candidate, with reported values of MRE of up to 2.5% at 9 μm with magnetic fields as small as 50 Oe [18]. Like many other GMR multilayers, the Ni81Fe19/Au system exhibits an oscillatory dependence of GMR as a function of the Au spacer thickness, due to the oscillations in the interlayer exchange coupling [19]. Regarding the potential control of Mid-IR optical properties of metallic components based in Ni81Fe19/Au multilayers, the effect of the Au spacer thickness is in principle twofold, on one hand to tune the GMR value and as a consequence the MRE, and on the other to modify the carrier density of the whole multilayer, and as a consequence also the optical constants. In this scope, the purpose of the present work is to study the relevance of both effects on the magnetic modulation performance in the Mid-IR of this material by considering different Ni81Fe19/Au multilayers in which the amount of Au is a tuneable parameter.
2. Experimental and results
The multilayers were grown at 150°C on CaF2 (111) single crystals. To prevent island formation and improve film adhesion, a 3 nm Ti buffer layer was first grown by electron beam evaporation followed by 9 layers of Ni81Fe19 separated by 8 layers of Au, both deposited by magnetron sputtering from individual Ni81Fe19 and Au targets. The thickness of the individual Ni81Fe19 layers was 2.9 nm while the thickness of the individual Au layers ranged between 1.9 and 3.3 nm. Finally, a 5 nm Au cap layer was deposited at 150°C. X-ray diffraction (XRD) and reflectometry (XRR) measurements were performed in a multipurpose diffractometer with Cu kα radiation. The thickness of the individual layers was extracted from best fits to the XRR data using the X’Pert Reflectivity package. DC Magneto resistance measurements were performed at room temperature using 4 in-line probes with typically 2 mA input current and enough magnetic field to saturate the samples. Optical characterization was performed measuring both, the transmission at normal incidence, and the reflectivity at nearly normal incidence (10 degrees), using a FTIR spectrometer (VERTEX 70 from Bruker) equipped with a MCT nitrogen cooled detector, covering the range from 2 μm to 10 μm. Magnetic coils in the spectrometer samples enclosure, generating up to 300 Oe, allowed magnetic saturation of the multilayers when needed. In all the measurements, the magnetic field is applied in the plane of the samples.
In Fig. 1 we show experimental XRR measurements for the fabricated multilayers with the corresponding best fits. The obtained individual Au layer thickness for each multilayer appears close to the corresponding curves. As it can be seen, both multilayer and total thickness fringes are obtained in all cases, pointing to a good interface quality. This technique allows us to discern differences in Au thickness of 0.2 nm (for example between 2.6 and 2.8 nm), which roughly corresponds to a variation in average thickness of one gold atomic layer assuming (111) orientation. This (111) textured growth is actually confirmed by high angle XRD measurements, as it can be seen in the inset to Fig. 1 (Au(111) bulk diffraction peak position indicated as a reference. Colour code for each multilayer is maintained the same here and in subsequent figures).
Fig. 1 XRR measurements and best fits for all the fabricated multilayers. The Au spacer thickness extracted from fits appears close of the corresponding curve and with the same colour. Inset: high angle XRD measurements for the fabricated multilayers. (Colour code for each multilayer is maintained here and in subsequent figures). Right: sketch of the multilayer.
After accurate determination of the individual layer thickness within each multilayer, in Fig. 2 we show the corresponding magnetotransport measurements for these structures as a function of the Au spacer thickness. As it can be seen, the resistivity decreases as we increase the amount of Au, certainly related to the increased metallic character of the structure. On the other hand, GMR values of 4% and 0.8% (detail of the second one in the inset to Fig. 2) are obtained for 2.3 and 3.3 nm Au respectively, with magnetic fields as small as 25 Oe needed to reach saturation. Intermediate and thinner Au spacers yield no GMR, again in agreement with previous results [19].
Fig. 2 Experimental magnetic field dependence of the in-line 4-point resistivity for the fabricated multilayers. The inset shows a close-up of the magnetic field dependence of the resistivity for the multilayer with 3.3 nm Au spacer.
Driving now our attention to the corresponding optical properties, in the upper inset of Fig. 3 we present reflectivity spectra of the most representative multilayers. As it can be observed, for all the structures the reflectivity increases for higher wavelengths and, within the different samples, it raises as we increase the amount of Au. This reflects again the increased metallic character of the multilayers already observed in the magnetotransport measurements. On the other hand, in the transmission spectra (lower inset of Fig. 3) we observe a decrease of the intensity as we increase the wavelength, with a smaller transmission as the total thickness of the structure increases. From these transmission and reflectivity measurements, it is possible to determine the real and imaginary parts of the effective dielectric constant, whose values are shown in the main panels of Fig. 3. For all the samples the real part is negative and it decreases as we increase the wavelength, whereas the imaginary part is positive and it increases as we increase the wavelength. Moreover, although the values of the dielectric constant are very similar for all the samples, there is a tendency of the real part of the dielectric constant to decrease as we increase the amount of Au, whereas the imaginary part tends to increase. This is due to the increase in the number of carriers as the concentration of Au increases.
Fig. 3 Real (circles: experimental results, lines theoretical simulations) and imaginary (squares: experimental results, lines: theoretical simulations) part of the dielectric constants for representative multilayers obtained from corresponding reflectivity and transmission measurements (insets).
The magnetic field induced changes in the dielectric constant due to MRE can be determined from reflectivity and transmission measurements under a magnetic field. The multilayers that show no GMR effect do not exhibit any magnetic field dependence in both, the reflectivity and transmission. On the contrary, the structures with GMR do show a magnetic field dependence of both magnitudes. As an example, in the upper inset to Fig. 4 we show the changes in the integrated transmission when a magnetic field is applied in the plane for the 4% GMR multilayer. As it can be seen, the shape of the curve and required saturation fields perfectly mimic the corresponding GMR curve of the same multilayer, showing the GMR origin of the MRE. Additionally, the spectral dependence of the reflection and transmission MRE spectra for the multilayers with 4% and 0.8% GMR are shown in the lower inset of Fig. 4. The curves correspond to the difference of the reflectivity or transmission between the two magnetic states of the sample, saturated and zero net magnetization, and normalized to the values at saturation. As it can be observed, the modulated transmission spectra show a broad band whose peak position and intensity depend on the GMR value. On the other hand, the modulated reflectivity spectra show a negative peak whose position and intensity also depend on the GMR value. The maximum change of the reflectivity and the transmission does not occur at the same wavelength due to the different spectral dependence of the sensitivity of the reflectivity and transmission to changes in the dielectric constant. These changes in the reflectivity (ΔR/R) and transmission (ΔT/T) induced by the magnetic field are due to changes in the real (Δεr) and imaginary (Δεi) part of the dielectric constant, which are presented in the main panels of Fig. 4. As it can be seen, in the 2 to 4-5 μm spectral range the modifications of the dielectric constants for both samples are small. On the other hand, above 5 μm the real part increases, whereas the imaginary part becomes negative and its absolute value shows a monotonic increase with wavelength.
Fig. 4 Magnetic modulation of the dielectric constants for two multilayers with 4% GMR (real part: black circles, imaginary part: black squares) and 0.8% GMR (real part: red circles, imaginary part: red squares), corresponding to 2.3 and 3.3 nm Au spacer thickness. Continuous, dashed and dotted lines correspond to theoretical simulations for different values of (β,α): doted black line (0.197,0), dashed black line (0.221,-.025), continuous black line (0.245,-.05). Upper inset: magnetic field dependence of the integrated modulation in transmission, mimicking the GMR curve of the multilayer. Lower inset: magnetic modulation of transmission and reflection spectra (back 4% sample, red 0.8% sample).
In this spectral range, the main contribution to the optical properties comes from the conduction electrons and the dielectric constant is determined by the electron conductivity. Therefore, when all the magnetic moments of the Ni81Fe19 layers are oriented parallel (P), the conductivity, and so the dielectric constant, can be expressed as the addition of the conductivity of the electrons whose spin is parallel ( + ) to the total magnetization (σ+(ω)), and the conductivity of those whose spin is antiparallel (-) to the total magnetization [20], (σ-(ω)). On the other hand, when the magnetic moments of the adjacent Ni81Fe19 layers are oriented antiparallel (AP), the electrons of both spin channels are indistinguishable and the conductivity is: σAP(ω) = (4σ-(ω)σ+(ω))/(σ-(ω) + σ+(ω)). The conductivity, σ+,-(ω),depends on the relaxation time, (τ+,-), and the ratio between the electron density, (n+,-), and effective mass, (m*+,-), of the conduction electrons: σ+,-(ω) = e2(n+,-/m*+,-)(τ+,-/(1-iωτ+,-). These parameters are spin dependent and can be expressed as: τ+,- = τ0/(1 ± β); n+,-/m*+,- = (n/2m*)(1/(1 ± α), with m*, τ0 and n being mean values of the effective mass, relaxation time and the total conduction electron density, respectively. β reflects the asymmetry in the relaxation times and α the differences in the electron densities and effective masses for the two spins states, respectively. If the electron effective masses are equal, the effective dielectric constant for the parallel and antiparallel state can be written as:
where ωp is the Plasmon frequency (ω2p = e2n/m*), and Γ0 = 1/τ0.These two expressions, εP(ω) and εAP(ω), clearly show the difference between the two magnetic sates for the sample: In the P state, the two terms between brackets in εP(ω) reflect the contribution of the two spin channels, which have different damping factors, Γ0(1 + β) and Γ0(1-β), and whose relative weight depends on their concentration. On the other hand, in the AP state, εAP(ω), the spin up and down electrons are indistinguishable, and the expression is reduced to that of a spin-less conduction electrons system, with an effective damping factor, Γeff = Γ0(1 + αβ). Besides, εAP(ω) also describes the frequency dependence of the dielectric constant of the structures that show no GMR effect, with Γeff = Γ0.
Therefore, εAP(ω) can be used to simulate the experimental values of the effective dielectric constant presented in Fig. 3 obtained at 0 field. The values of ωp thus obtained vary between 5.3 eV and 5.8 eV, with a tendency to increase as we increase the amount of Au, whereas Γeff is in the range of 75 - 70 meV. In Fig. 3 we present the theoretical curves for the 1.9nm Au (cyan curve), 2.3nm Au (black curve), and 3.3 nm Au (red curve) samples, which have a ωp, of 5.3 eV, 5.5 eV and 5.8 eV, respectively, and a damping factor Γeff of 70 meV, 75 meV, and 75 meV, respectively. Regarding the evolution of the plasmon frequency, it has been suggested [21] that confinement effects may reduce ωp. In this case, the metallic Ni81Fe19 layer should act as a barrier for the conduction electrons, which is not likely to be the case. Thefore, by analyzing the evolution of all the fabricated multilayers, we can conclude that the increase of ωp with the amount of Au can be related to the increase of the electron concentration, whereas the increase of the damping factor is consistent with the decrease of the interface quality along the series due to accumulated roughness for thicker Au layers. On the other hand, the MRE-induced changes of the dielectric constant are given by the difference εP(ω)-εAP(ω), which depend, not only on ωp and Γ0, but also on the spin related parameters, β, α. As example, in Fig. 4 we present this difference for ωp = 5.5 eV, Γ0 = 75 meV, and for different values of (β,α). These combinations of β and α give a value of GMR of 4%. As it can be observed, the spectral dependence of the MRE induced changes depends strongly on the values of α and β, being very sensitive to the difference in the electron concentration. For example, a difference of 5% in the electron concentration (α = −0.05) gives rise to a strong increase and a red shift of the real part of the dielectric constant, but with similar changes in the imaginary part. Moreover, the experimental results of the 4% sample (black circles and black squares), are similar to the theoretical curves with β = 0.245 and α = −0.05. In that figure, we also present theoretical curves of the difference εP(ω)-εAP(ω) (continuous red line) for ωp = 5.8 eV, Γ0 = 75 meV, β = 0.119, and α = −0.03. As it can be seen, the theoretical curves fit the experimental results of the 0.8% GMR sample.
The dependence on α and β of the spectral shape of MRE induced changes of the optical properties has important consequences on the magnetic modulation of nanostructures. As an example, we present in Fig. 5 simulation of the transmission (T) and magnetic modulation of the transmission (ΔT/T) of an array of rectangular prism antennas, with antennas concentration of 2% and prism dimensions of 2x0.05x0.05 μm3 located on top of a substrate with refractive index 1.4.
Fig. 5 Simulated spectra of the transmission (a) and magnetic modulation of the transmission (b) at normal incidence for light polarized parallel to the long axis of the antenna prism of an array of GMR antennas for different values of (β,α): doted black line (0.197,0), dashed black line (0.221,-.025), continuous black line (0.245,-.05). The inset represents a schema of the antenna layer.
The antenna material is a multilayer with a GMR value of 4%, whose optical properties of the P and AP magnetic states are described by the dielectric constants of Eq. (1) and (2), respectively, with ωp = 5.5 eV, Γ0 = 75 meV. On the other hand, regarding the α and β parameters, three different pairs of values have been considered, all of them giving the same GMR value. The antennas layer has been simulated using a Maxwell Garnet approximation, being the incident light polarized along the long antenna axis. In Fig. 5(a) we present the transmission spectra at normal incidence for the three different pairs of α and β values. As it can be observed, the transmission spectra are virtually identical for the three pairs of values, showing a dip that corresponds to the resonance of the antenna. On the contrary, the shape of the magnetic modulated transmission spectra (difference between the transmission when the antennas are saturated, P magnetic state, and at zero field, AP magnetic sate, normalized to the transmission at zero field) strongly depends on the values of α and β (Fig. 5(b)), highlighting the sensitivity of this magnitude, ΔT/T, to the spin asymmetry of the relaxation time and electron concentrations. Therefore, exploring ways to modify the asymmetry of these two magnitudes such as, for example, by doping the Ni81Fe19/Au multilayers with sub monolayers of different elements [22], may give rise to structures with similar GMR values but with different MRE induced changes in the optical properties, allowing an optimization of the mid- and far-infrared magnetic modulation of optical devices.
3. Conclusion
We have analysed the optical properties and MRE effect in the mid infrared range of a series of Ni81Fe19/Au multilayers grown by magnetron sputtering. In this spectral range the optical properties are determined by conduction electrons with an increase of the plasmon frequency as we increase the amount of Au. On the other hand, the spectral dependence of the magnetic field induced changes in the dielectric constant suggests that both, the spin asymmetry of the relaxation time and the different electron concentration, contribute to MRE.
Funding
MINECO (MAT2014-58860-P, FIS2015-72035-EXP, MAT2017-84009-R); MiNa Laboratory at IMN funding from MINECO (CSIC13-4E-1794); CM (S2013/ICE-2822 (Space-Tec)), both with support from EU (FEDER, FSE).
References
1. E. C. Kinzel, J. C. Ginn, R. L. Olmon, D. J. Shelton, B. A. Lail, I. Brener, M. B. Sinclair, M. B. Raschke, and G. D. Boreman, “Phase resolved near-field mode imaging for the design of frequency-selective surfaces,” Opt. Express 20(11), 11986–11993 (2012). [CrossRef] [PubMed]
2. J. A. Mason, S. Smith, and D. Wasserman, “Strong absorption and selective thermal emission from a midinfrared metamaterial,” Appl. Phys. Lett. 98(24), 241105 (2011). [CrossRef]
3. S. Ogawa, K. Okada, N. Fukushima, and M. Kimata, “Wavelength selective uncooled infrared sensor by plasmonics,” Appl. Phys. Lett. 100(2), 021111 (2012). [CrossRef]
4. J. K. Gansel, M. Thiel, M. S. Rill, M. Decker, K. Bade, V. Saile, G. von Freymann, S. Linden, and M. Wegener, “Gold helix photonic metamaterial as broadband circular polarizer,” Science 325(5947), 1513–1515 (2009). [CrossRef] [PubMed]
5. F. Neubrech, A. Pucci, T. W. Cornelius, S. Karim, A. García-Etxarri, and J. Aizpurua, “Resonant plasmonic and vibrational coupling in a tailored nanoantenna for infrared detection,” Phys. Rev. Lett. 101(15), 157403 (2008). [CrossRef] [PubMed]
6. G. Armelles, L. Bergamini, N. Zabala, F. García, M. L. Dotor, L. Torné, R. Alvaro, A. Griol, A. Martínez, J. Aizpurua, and A. Cebollada, “Metamaterial platforms for spintronic modulation of mid-Infrared response under very weak magnetic field,” ACS Photonics 5(10), 3956–3961 (2018). [CrossRef]
7. J. C. Jacquet and T. A. Valet, “A new magnetooptical effect discovered on magnetic multilayers: the magnetorefractive effect,” Proc. MRS 384, 477–490 (1995). [CrossRef]
8. M. Vopsaroiu, T. Stanton, O. Thomas, M. Cain, and S. M. Thompson, “Infrared metrology for spintronic materials and devices,” Meas. Sci. Technol. 20(4), 045109 (2009). [CrossRef]
9. R. J. Baxter, D. G. Pettifor, E. Y. Tsymbal, D. Bozec, J. A. D. Matthew, and S. M. Thompson, “Importance of the interband contribution to the magneto-refractive effect in Co/Cu multilayers,” J. Phys. Cond. Mat. 15(45), L695–L702 (2003). [CrossRef]
10. M. Vopsaroiu, D. Bozec, J. A. D. Matthew, S. M. Thompson, C. H. Marrows, and M. Perez, “Contactless magnetoresistance studies of Co/Cu multilayers using the infrared magnetorefractive effect,” Phys. Rev. B Condens. Matter Mater. Phys. 70(21), 214423 (2004). [CrossRef]
11. R. T. Mennicke, D. Bozec, V. G. Kravets, M. Vopsaroiu, J. A. D. Matthew, and S. M. Thompson, “Modelling the magnetorefractive effect in giant magnetoresistive granular and layered materials,” J. Magn. Magn. Mater. 303(1), 92–110 (2006). [CrossRef]
12. I. D. Lobov, M. M. Kirillova, A. A. Makhnev, L. N. Romashev, and V. V. Ustinov, “Parameters of Fe/Cr interfacial electron scattering from infrared magnetoreflection,” Phys. Rev. B Condens. Matter Mater. Phys. 81(13), 134436 (2010). [CrossRef]
13. T. Ogasawara, H. Kuwatsuka, T. Hasama, and H. Ishikawa, “Proposal of an optical nonvolatile switch utilizing surface plasmon antenna resonance controlled by giant magnetoresistance,” Appl. Phys. Lett. 100(25), 251112 (2012). [CrossRef]
14. Z. Jim, A. Tkach, F. Casper, V. Spetter, H. Grimm, A. Thomas, T. Kampfrath, M. Bonn, M. Kläui, and D. Turchinovich, “Accessing the fundamentals of magnetotransport in metals with terahertz probes,” Nat. Phys. 11(9), 761–766 (2015). [CrossRef]
15. V. G. Kravets, D. Bozec, J. A. D. Matthew, S. M. Thompson, H. Menard, A. B. Horn, and A. F. Kravets, “Correlation between the magnetorefractive effect, giant magnetoresistance, and optical properties of Co-Ag granular magnetic films,” Phys. Rev. B Condens. Matter Mater. Phys. 65(5), 054415 (2002). [CrossRef]
16. J. van Driel, F. R. de Boer, R. Coehoorn, G. H. Rietjens, and E. S. J. Heuvelmans-Wijdenes, “Magnetorefractive and magnetic-linear-dichroism effect in exchange-biased spin valves,” Phys. Rev. B Condens. Matter Mater. Phys. 61(22), 15321–15326 (2000). [CrossRef]
17. S. M. Strutner, A. Garcia, S. Ula, C. Adamo, W. L. Richards, K. Wang, D. Schlom, and G. P. Carman, “Index of refraction changes under magnetic field observed in La0.66Sr0.33MnO3 correlated to the magnetorefractive effect,” Opt. Mater. Expr. 7(2), 468–476 (2017).
18. G. Armelles, A. Cebollada, F. García, and C. Pecharromán, “Magnetic modulation of mid-infrared plasmons using Giant Magnetoresistance,” Opt. Express 25(16), 18784–18796 (2017). [CrossRef] [PubMed]
19. S. S. P. Parkin, R. F. C. Farrow, R. F. Marks, A. Cebollada, G. R. Harp, and R. J. Savoy, “Oscillations of interlayer exchange coupling and giant magnetoresistance in (111) oriented permalloy/Au multilayers,” Phys. Rev. Lett. 72(23), 3718–3721 (1994). [CrossRef] [PubMed]
20. A. Fert, A. Barthelemy, and F. Petroff, “Nanomagnetism: Ultrathin films, multilayers and nanostructures,” Contemporary Concepts of Condensed Matter Science. Editors: D. L. Mills and J. A. C. Bland. Elsevier Amsterdam 2006.
21. I. V. Bondarev and V. M. Shalaev, “Universal features of the optical properties of ultrathin plasmonic films,” Opt. Mater. Express 7(10), 3731 (2017). [CrossRef]
22. C. H. Marrows and B. J. Hickey, “Impurity scattering from δ-layers in giant magnetoresistance systems,” Phys. Rev. B Condens. Matter Mater. Phys. 63(22), 220405 (2001). [CrossRef]
