- Email: [email protected]
CoO bilayers
Journal of Magnetism and Magnetic Materials 198}199 (1999) 534}536
Temperature dependent spin-wave behaviour in Co/CoO bilayers A. Ercole, E.T.M. Kernohan, G. Lauho!, J.A.C. Bland* Cavendish Laboratory, Madingley Road, Cambridge CB3 0HE, UK
Abstract We report Brillouin light scattering measurements of spin-wave frequencies in an exchange coupled ferromagnet (FM)/antiferromagnet (AF) epitaxial fcc Co/CoO bilayer structure. A striking temperature dependence of the measured frequencies in the cobalt layer in the range 77 to 300 K was observed and has been demonstrated to be due to exchange coupling to the antiferromagnetic layer. The existence of a uniaxial anisotropy "eld in the range 100}300 Oe within the AF layer along the FM magnetisation direction was demonstrated. The ratio of the interface to bulk AF exchange coupling strengths was found to lie in the range 0.75}1.1. The temperature dependence of the spin-wave line widths was also examined and used to show that locally ordered AF regions persist above the NeH el temperature and play a central role in determining the magnetic behaviour. 1999 Elsevier Science B.V. All rights reserved. Keywords: Brillouin light scattering; Exchange-bias; Interfacial magnetic properties
1. Introduction Exchange biasing in strongly coupled FM/AF structures is of considerable interest due to its relevance in magnetoresistive devices [1]. The e!ect has been attributed to the interfacial exchange interaction between the layers. In order to explain the observed coupling strength, which is 2 orders of magnitude less than that predicted by the earliest model [2], newer theories have attempted to describe the detailed AF spin structure at the interface [3}6]. However the e!ect of the uniaxial anisotropy K in the AF and the role of disorder at the interface are still poorly understood [7]. We have chosen to study a system with CoO (bulk NeH el temperature 295 K) as the AF layer since the antiferromagnetic order may be switched on or o! with temperature. A Co/CoO bilayer was grown at room temperature on a Cu/Si (0 0 1) template by molecular beam epitaxy under
* Corresponding author. Tel.: #44 1223 337284; #44 1223 350266. E-mail address: [email protected] (J.A.C. Bland)
fax:
an ultra high vacuum of order 10\ mbar [8]. The substrates were hydrogen passivated with aqueous HF to stabilise the surface [9]. After 30 As of Co was grown, a shutter was placed to cover half of the substrate and a 30 As Cu cap was deposited to form a control sample without an oxide layer. The sample was then oxidised for 30 min in 10\ mbar of high purity oxygen, and the shutter repositioned to enable the oxidised half to be capped, again with 30 As of Cu. RHEED images showed that the Co layer grows pseudomorphically with the Cu bu!er layer, and that although some disorder is introduced in oxidising the sample, it remains highly crystalline. Ex situ magnetooptical Kerr e!ect (MOKE) measurements showed the samples to have four-fold in-plane anisotropy. The oxidised sample showed an increased coercivity over the control sample as is to be expected from the increased surface disorder shown by the RHEED measurements. Polarised neutron re#ectometry (PNR) measurements were performed on both samples to access the magnetic moment and layer thicknesses. A "t to the control sample re#ectivity spectra yielded the Co magnetisation and the layer thicknesses. From these
0304-8853/99/$ } see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 8 8 5 3 ( 9 8 ) 0 1 2 3 4 - 7
A. Ercole et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 534}536
results, it could be deduced that 4pM "1.27 kOe, ! which is somewhat reduced from the bulk value, presumably due to the small amount of carbon contamination shown by Auger electron spectroscopy. The Co layer thickness was found to be 31 As . The other layer thicknesses were also found to be close to the nominal values. PNR from the oxidised sample gave the Co layer thickness as 27 As , and the CoO layer thickness as 7 As . This is equivalent to approximately 13 and 4 ML, respectively. The error on these "gures is approximately 1}2 ML. Our BLS system uses a Sandercock 3#3 pass tandem Fabry-Perot interferometer [10]. For the low temperature measurements, the sample was mounted in a customised gas #ow cryostat, in which the angle of the incident light was chosen to be 153. This ensured that the in-plane wavevector being probed was as small as possible, without moving to normal incidence which gives too much elastic signal, blinding the interferometer.
2. Results and discussion Measurements of the Co surface mode spin-wave frequency as a function of in-plane applied "eld angle and strength allowed the magnetic parameters of the samples to be deduced (see for instance Ref. [11]). For the Co layer, these were K /M"!500 Oe (making the Co [1 0 0] direction hard), g"2.30, K /M"35 Oe and K "!0.50 erg cm\ (i.e. favouring in-plane magnet isation). The angle scan from the CoO sample could be "tted using the layer thicknesses and Co layer magnetisation from the PNR measurements. K was found to be reduced to !0.29 erg cm\ due to the di!erence between the Co/CoO and Co/Cu interfaces. If it is assumed that the two Co/Cu interfaces in the control sample are equivalent then one can infer a value of K " !0.1 erg cm\ for the Co/CoO interface. The spin-wave excitations in magnetic materials corresponds to thermally excited spin-precessions and are described by solutions to the Landau}Lifshitz torque equation: *M "c(M;H ), *t
535
Ref. [12], which has contribution from exchange within the FM and AF layers (J and J ), an interface ex$+ $ change (J ), a uniaxial in-plane anisotropy "eld H and an out-of-plane anisotropy "eld H. In addition we have included a four-fold cubic anisotropy "eld H. To solve for the spin-wave mode frequencies, a harmonic time dependence is assumed for each S , and the resulting G matrix equation is solved. Fig. 1 shows the temperature dependence of the spinwave frequency as the sample was cooled in steps of 25 K with a 3 kOe "eld applied along either the easy or hard crystallographic axes. Similar measurements for the control sample showed no detectable variation in frequency. In "tting the data, only the CoO parameters have been varied. For the Co layer, the magnetic parameters from room temperature BLS angle and "eld scans have been used. Within the model, the CoO magnetic parameters modify the predicted magnitude and onset of the temperature dependence of the spin-wave frequency as follows. The critical temperature is governed almost entirely by the value of the exchange within the AF layer (found to be !365$5 Oe). The magnitude of the rise in frequency with temperature is governed by both the inplane (two-fold) anisotropy in the AF and the exchange strength at the interface: it is not possible to determine these parameters independently. However, as the interface exchange strength is assumed to be equal to that in the bulk AF layer then a value of H "120$10 Oe is deduced. Despite the interdependence of these parameters, by varying the interface exchange strength it was possible to determine that 100 Oe(H (300 Oe and that 0.75J (J (1.1J . It was not necessary to $ $ include H or a four-fold anisotropy in the CoO layer in order to obtain a "t to the data.
(1)
where M is the magnetisation, c the gyromagnetic ratio (proportional to the g-factor) and H is the sum of all "elds present in the sample, which may be external in origin, or internal due to exchange or magnetocyrstalline anisotropy. The structure is modeled as a number of atomic layers, and the magnetic atoms in the CoO are assumed to be layer-by-layer antiparallel, so that only vertical coupling between layers needs to be considered. For the e!ective "eld H experienced by spins S in the G G ith layer, we have used an ansatz similar to that of
Fig. 1. Temperature dependence of the Co spin-wave frequency for "elds directed along easy (Co[1 1 0] } top curve) and hard (Co[1 0 0] } lower curve) axes. The lines are "ts to the mean "eld model described in the text. By contrast, the unoxidised sample showed no detectable change in frequency with temperature. The inset shows the analogous data for the unoxidised sample on the same scale which shows only a 3% increase over the temperature range shown.
536
A. Ercole et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 534}536
re#ect statistical variations in surface roughness, oxide composition and/or bulk structural imperfections. The total exchange strength will increase to a limiting value at low temperature. The spin-wave line width, however, is sensitive not to the total exchange strength, but to the -uctuations in the exchange and is a!ected only by disordered regions of the AF layer.
3. Conclusions We have shown that the spin-wave frequencies in the Co layer of a Co/CoO bilayer structure show a striking temperature dependence due to exchange coupling to the antiferromagnetic CoO layer. This temperature dependence has been used to demonstrate the existence of a uniaxial anisotropy within the AF layer but along the FM magnetisation direction and to determine the interface exchange strength. The temperature dependence of the spin-wave line widths has been used to show that locally ordered AF regions persist above the NeH el temperature and play a central role in determining the magnetic behaviour. Fig. 2. The best "ts for Eq. (1) for the frequency and line width data as functions of temperature together with the best "t parameters.
Fig. 2 shows the temperature tempendence of the spin-wave frequencies f (¹) and linewidths *f (¹). The critical temperatures at which these quantities reach their high temperature limits, ¹ and ¹ respectively, are , * substantially di!erent, indicating that the increase in f (¹) and the increase in *f (¹) with reducing temperature are due to ordering on di!erent lengthscales. We may speculate on the microscopic origin of these e!ects. That the value of ¹ is lower than ¹ suggests , * that whilst f (¹) is governed by the magnetic ordering within the &bulk' of the AF layer, *f (¹) is instead controlled by small regions of the AF (possibly at the interface) which are able to maintain their magnetic order up to high temperatures than the bulk. Such regions correspond to locally AF ordered spins, but the random relative arrangement of these regions gives rise to the increase in line width. That a whole spectrum of ordering temperatures can exist in such FM/AF bilayer systems has been demonstrated for sputtered FeNi/FeMn [13]. The individual contributions to the exchange strength
References [1] B. Dieny, V.S. Spersiosu, S.S.S. Parkin, B.A. Gurney, D.R. Wilhout, D. Mauri, Phys. Rev. B 43 (1991) 1297. [2] W.H. Meiklejohn, C.P. Bean, Phys. Rev. 102 (1956) 1413. [3] A.P. Malozemo!, J. Appl. Phys. 63 (1988) 3874. [4] N.C. Koon, Phys. Rev. Lett. 78 (1977) 4865. [5] Y. Ijiri, J.A. Borchers, R.W. Erwin, S.-H. Lee, P.J. van der Zaag, R.M. Wolf, Phys. Rev. Lett. 80 (1998) 608. [6] B.H. Miller, E. Dan Dahlberg, Appl. Phys. Lett. 69 (1996) 3932. [7] B. Raquet, R. Mamy, J.C. Ousset, N. Negre, M. Goiran, C. Guerret Piecourt, J. Magn. Magn. Mater. 184 (1998) 41. [8] J. Lee, G. Lauho!, M. Tselepi, S. Hope, P. Rosenbusch, J.A.C. Bland, H.A. DuK rr, G. van der Laan, J.Ph. SchilleH , J.A.D. Matthew, Phys. Rev. B 55 (1997) 15103. [9] T. Ambrose, K. Leifer, K.J. Hemker, C.L. Chien, J. Appl. Phys. 81 (1997) 5007. [10] R. Mock, B. Hillebrands, J.R. Sandercock, J. Phys. E 20 (1987) 656. [11] R.J. Hicken, D.E.P. Eley, M. Gester, C. Daboo, A.J.R. Ives, J.A.C. Bland, J. Magn. Magn. Mater. 145 (1995) 278. [12] R.L. Stamps, R.E. Camley, R.J. Hicken, J. Appl. Phys. 81 (1997) 4485. [13] C. Tsang, K. Lee, J. Appl. Phys. 53 (1982) 2605.
Temperature dependent spin-wave behaviour in Co/CoO bilayers A. Ercole, E.T.M. Kernohan, G. Lauho!, J.A.C. Bland* Cavendish Laboratory, Madingley Road, Cambridge CB3 0HE, UK
Abstract We report Brillouin light scattering measurements of spin-wave frequencies in an exchange coupled ferromagnet (FM)/antiferromagnet (AF) epitaxial fcc Co/CoO bilayer structure. A striking temperature dependence of the measured frequencies in the cobalt layer in the range 77 to 300 K was observed and has been demonstrated to be due to exchange coupling to the antiferromagnetic layer. The existence of a uniaxial anisotropy "eld in the range 100}300 Oe within the AF layer along the FM magnetisation direction was demonstrated. The ratio of the interface to bulk AF exchange coupling strengths was found to lie in the range 0.75}1.1. The temperature dependence of the spin-wave line widths was also examined and used to show that locally ordered AF regions persist above the NeH el temperature and play a central role in determining the magnetic behaviour. 1999 Elsevier Science B.V. All rights reserved. Keywords: Brillouin light scattering; Exchange-bias; Interfacial magnetic properties
1. Introduction Exchange biasing in strongly coupled FM/AF structures is of considerable interest due to its relevance in magnetoresistive devices [1]. The e!ect has been attributed to the interfacial exchange interaction between the layers. In order to explain the observed coupling strength, which is 2 orders of magnitude less than that predicted by the earliest model [2], newer theories have attempted to describe the detailed AF spin structure at the interface [3}6]. However the e!ect of the uniaxial anisotropy K in the AF and the role of disorder at the interface are still poorly understood [7]. We have chosen to study a system with CoO (bulk NeH el temperature 295 K) as the AF layer since the antiferromagnetic order may be switched on or o! with temperature. A Co/CoO bilayer was grown at room temperature on a Cu/Si (0 0 1) template by molecular beam epitaxy under
* Corresponding author. Tel.: #44 1223 337284; #44 1223 350266. E-mail address: [email protected] (J.A.C. Bland)
fax:
an ultra high vacuum of order 10\ mbar [8]. The substrates were hydrogen passivated with aqueous HF to stabilise the surface [9]. After 30 As of Co was grown, a shutter was placed to cover half of the substrate and a 30 As Cu cap was deposited to form a control sample without an oxide layer. The sample was then oxidised for 30 min in 10\ mbar of high purity oxygen, and the shutter repositioned to enable the oxidised half to be capped, again with 30 As of Cu. RHEED images showed that the Co layer grows pseudomorphically with the Cu bu!er layer, and that although some disorder is introduced in oxidising the sample, it remains highly crystalline. Ex situ magnetooptical Kerr e!ect (MOKE) measurements showed the samples to have four-fold in-plane anisotropy. The oxidised sample showed an increased coercivity over the control sample as is to be expected from the increased surface disorder shown by the RHEED measurements. Polarised neutron re#ectometry (PNR) measurements were performed on both samples to access the magnetic moment and layer thicknesses. A "t to the control sample re#ectivity spectra yielded the Co magnetisation and the layer thicknesses. From these
0304-8853/99/$ } see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 8 8 5 3 ( 9 8 ) 0 1 2 3 4 - 7
A. Ercole et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 534}536
results, it could be deduced that 4pM "1.27 kOe, ! which is somewhat reduced from the bulk value, presumably due to the small amount of carbon contamination shown by Auger electron spectroscopy. The Co layer thickness was found to be 31 As . The other layer thicknesses were also found to be close to the nominal values. PNR from the oxidised sample gave the Co layer thickness as 27 As , and the CoO layer thickness as 7 As . This is equivalent to approximately 13 and 4 ML, respectively. The error on these "gures is approximately 1}2 ML. Our BLS system uses a Sandercock 3#3 pass tandem Fabry-Perot interferometer [10]. For the low temperature measurements, the sample was mounted in a customised gas #ow cryostat, in which the angle of the incident light was chosen to be 153. This ensured that the in-plane wavevector being probed was as small as possible, without moving to normal incidence which gives too much elastic signal, blinding the interferometer.
2. Results and discussion Measurements of the Co surface mode spin-wave frequency as a function of in-plane applied "eld angle and strength allowed the magnetic parameters of the samples to be deduced (see for instance Ref. [11]). For the Co layer, these were K /M"!500 Oe (making the Co [1 0 0] direction hard), g"2.30, K /M"35 Oe and K "!0.50 erg cm\ (i.e. favouring in-plane magnet isation). The angle scan from the CoO sample could be "tted using the layer thicknesses and Co layer magnetisation from the PNR measurements. K was found to be reduced to !0.29 erg cm\ due to the di!erence between the Co/CoO and Co/Cu interfaces. If it is assumed that the two Co/Cu interfaces in the control sample are equivalent then one can infer a value of K " !0.1 erg cm\ for the Co/CoO interface. The spin-wave excitations in magnetic materials corresponds to thermally excited spin-precessions and are described by solutions to the Landau}Lifshitz torque equation: *M "c(M;H ), *t
535
Ref. [12], which has contribution from exchange within the FM and AF layers (J and J ), an interface ex$+ $ change (J ), a uniaxial in-plane anisotropy "eld H and an out-of-plane anisotropy "eld H. In addition we have included a four-fold cubic anisotropy "eld H. To solve for the spin-wave mode frequencies, a harmonic time dependence is assumed for each S , and the resulting G matrix equation is solved. Fig. 1 shows the temperature dependence of the spinwave frequency as the sample was cooled in steps of 25 K with a 3 kOe "eld applied along either the easy or hard crystallographic axes. Similar measurements for the control sample showed no detectable variation in frequency. In "tting the data, only the CoO parameters have been varied. For the Co layer, the magnetic parameters from room temperature BLS angle and "eld scans have been used. Within the model, the CoO magnetic parameters modify the predicted magnitude and onset of the temperature dependence of the spin-wave frequency as follows. The critical temperature is governed almost entirely by the value of the exchange within the AF layer (found to be !365$5 Oe). The magnitude of the rise in frequency with temperature is governed by both the inplane (two-fold) anisotropy in the AF and the exchange strength at the interface: it is not possible to determine these parameters independently. However, as the interface exchange strength is assumed to be equal to that in the bulk AF layer then a value of H "120$10 Oe is deduced. Despite the interdependence of these parameters, by varying the interface exchange strength it was possible to determine that 100 Oe(H (300 Oe and that 0.75J (J (1.1J . It was not necessary to $ $ include H or a four-fold anisotropy in the CoO layer in order to obtain a "t to the data.
(1)
where M is the magnetisation, c the gyromagnetic ratio (proportional to the g-factor) and H is the sum of all "elds present in the sample, which may be external in origin, or internal due to exchange or magnetocyrstalline anisotropy. The structure is modeled as a number of atomic layers, and the magnetic atoms in the CoO are assumed to be layer-by-layer antiparallel, so that only vertical coupling between layers needs to be considered. For the e!ective "eld H experienced by spins S in the G G ith layer, we have used an ansatz similar to that of
Fig. 1. Temperature dependence of the Co spin-wave frequency for "elds directed along easy (Co[1 1 0] } top curve) and hard (Co[1 0 0] } lower curve) axes. The lines are "ts to the mean "eld model described in the text. By contrast, the unoxidised sample showed no detectable change in frequency with temperature. The inset shows the analogous data for the unoxidised sample on the same scale which shows only a 3% increase over the temperature range shown.
536
A. Ercole et al. / Journal of Magnetism and Magnetic Materials 198}199 (1999) 534}536
re#ect statistical variations in surface roughness, oxide composition and/or bulk structural imperfections. The total exchange strength will increase to a limiting value at low temperature. The spin-wave line width, however, is sensitive not to the total exchange strength, but to the -uctuations in the exchange and is a!ected only by disordered regions of the AF layer.
3. Conclusions We have shown that the spin-wave frequencies in the Co layer of a Co/CoO bilayer structure show a striking temperature dependence due to exchange coupling to the antiferromagnetic CoO layer. This temperature dependence has been used to demonstrate the existence of a uniaxial anisotropy within the AF layer but along the FM magnetisation direction and to determine the interface exchange strength. The temperature dependence of the spin-wave line widths has been used to show that locally ordered AF regions persist above the NeH el temperature and play a central role in determining the magnetic behaviour. Fig. 2. The best "ts for Eq. (1) for the frequency and line width data as functions of temperature together with the best "t parameters.
Fig. 2 shows the temperature tempendence of the spin-wave frequencies f (¹) and linewidths *f (¹). The critical temperatures at which these quantities reach their high temperature limits, ¹ and ¹ respectively, are , * substantially di!erent, indicating that the increase in f (¹) and the increase in *f (¹) with reducing temperature are due to ordering on di!erent lengthscales. We may speculate on the microscopic origin of these e!ects. That the value of ¹ is lower than ¹ suggests , * that whilst f (¹) is governed by the magnetic ordering within the &bulk' of the AF layer, *f (¹) is instead controlled by small regions of the AF (possibly at the interface) which are able to maintain their magnetic order up to high temperatures than the bulk. Such regions correspond to locally AF ordered spins, but the random relative arrangement of these regions gives rise to the increase in line width. That a whole spectrum of ordering temperatures can exist in such FM/AF bilayer systems has been demonstrated for sputtered FeNi/FeMn [13]. The individual contributions to the exchange strength
References [1] B. Dieny, V.S. Spersiosu, S.S.S. Parkin, B.A. Gurney, D.R. Wilhout, D. Mauri, Phys. Rev. B 43 (1991) 1297. [2] W.H. Meiklejohn, C.P. Bean, Phys. Rev. 102 (1956) 1413. [3] A.P. Malozemo!, J. Appl. Phys. 63 (1988) 3874. [4] N.C. Koon, Phys. Rev. Lett. 78 (1977) 4865. [5] Y. Ijiri, J.A. Borchers, R.W. Erwin, S.-H. Lee, P.J. van der Zaag, R.M. Wolf, Phys. Rev. Lett. 80 (1998) 608. [6] B.H. Miller, E. Dan Dahlberg, Appl. Phys. Lett. 69 (1996) 3932. [7] B. Raquet, R. Mamy, J.C. Ousset, N. Negre, M. Goiran, C. Guerret Piecourt, J. Magn. Magn. Mater. 184 (1998) 41. [8] J. Lee, G. Lauho!, M. Tselepi, S. Hope, P. Rosenbusch, J.A.C. Bland, H.A. DuK rr, G. van der Laan, J.Ph. SchilleH , J.A.D. Matthew, Phys. Rev. B 55 (1997) 15103. [9] T. Ambrose, K. Leifer, K.J. Hemker, C.L. Chien, J. Appl. Phys. 81 (1997) 5007. [10] R. Mock, B. Hillebrands, J.R. Sandercock, J. Phys. E 20 (1987) 656. [11] R.J. Hicken, D.E.P. Eley, M. Gester, C. Daboo, A.J.R. Ives, J.A.C. Bland, J. Magn. Magn. Mater. 145 (1995) 278. [12] R.L. Stamps, R.E. Camley, R.J. Hicken, J. Appl. Phys. 81 (1997) 4485. [13] C. Tsang, K. Lee, J. Appl. Phys. 53 (1982) 2605.