- Email: [email protected]
MBE-grown spin valves
Journal of Magnetism and Magnetic Materials 156 (1996) 63-64
~ H Journalof magnetism and magnetic ~ i materials
N
ELSEVIER
MBE-grown spin valves K.-M.H. Lenssen *, J.C.S. Kools, A.E.M. De Veirman, J.J.T.M. Donkers, M.T. Johnson, A. Reinders, R. C o e h o o r n Philips Research lxlboratories. Prq~essor Holstlaan 4. NL-5656 AA Eindhot'en. The Netherland.~
Abstract Spin valves were grown by vapour deposition in an MBE system, instead of the usual sputter deposition. Compared to sputtered samples, the GMR ratio A R / R was slightly lower and the exchange-bias field was halved. Also, the (1 I1) texture was much less pronounced. The Ta buffer layer was studied by TEM analysis, providing new information about the required structure to initiate the (l I l) texture.
For the first time, we report on spin-valve multilayers that were grown by vapour deposition in a molecular beam epitaxy (MBE) system. It might be expected that MBEgrown films are superior, e.g. with respect to the interface quality, as compared to films made by means of sputter deposition. Both techniques have different characteristics, e.g. concerning the kinetic energy of the particles, possibly resulting in different film microstructures. Therefore, it is interesting to compare MBE-grown spin valves with sputtered samples. A series of spin valves with varying Cu thickness tcu was grown, consisting of S i ( 1 0 0 ) / 3 n m T a / S n m NisoF%o/tcu C u / 6 nm NisoFe2o/8 nm F%0Mn50/3 nm Ya (to, = 2.0, 2.5. 3.0 and 3.5 nm), using (co-)evaporation from elemental sources. During growth a magnetic field of 20 k A / m was applied along the long axis of the rectangular 4 × 12 mm 2 substrates; the base pressure was ~ 10-t0 mbar, and the substrate temperature was kept at 20°C. The magnetoresistance of the multilayers was measured in a four-terminal configuration (current and field both ahmg the long axis). The data are shown in Fig. 1. The largest effect of 3.9% was found for the spin valve with to, , = 2.5 rim. In order to determine the effective magnetic dead-layer thickness in MBE-grown Nis0Fe=o (Py), multilayers of the type Si(100)/3nm Ta/tpy NisoFez0/2nm C u / 3 n m Ya (tp>. = 2.0. 4.0, 6.0 and 8.0 nm) were deposited. The magnetic moment of these films was determined in a SQU1D magnetometer. A good linear relationship between magnetic moment and /py was found (slope of 625 k A / m ) , yielding a value of 0.5 nm as an estimation of the sum of
the effective dead-layer thicknesses at both interfaces. This value is smaller than the value obtained for comparable sputter-deposited systems ( ~ 0 . 8 n m [1]), which is in agreement with the expected sharper interfaces associated with a low-energetic deposition technique like MBE. The ratio A R / R of the giant magneto-resistance (GMR) effect was somewhat ( ~ 1%) less than that of sputtered spin valves [2,3]. In Fig. I it can be seen that this is not the result of insufficient exchange biasing, although the exchange-bias field is only half the value fi)und for sputtered samples. X-ray diffraction showed that the degree of (111) texture of the films was much lower, as indicated by the large width of rocking curves around the (111) peak of more than 15° (as compared to typically 5 ° for sputtered samples [4]). This can explain the reduced exchange-bias field [5], and possibly also the lower GMR ratio. That (l 1 l) texture is also directly correlated with
32
30 2~ 2~
24 2~
~
-5
I 0
i 5
1 10
I
I 15
20
tt (kA/m)
* Corresponding author. Fax: +31-40-274-4282; lenssen @natlab.research.philips.com.
email:
Fig. 1. Magnetoresistance curves of MBE-grown spin valves with a Cu-spacer thickness equal to 2.0, 2.5, 3.0 and 3.5 nm (from top to bottom), respectively.
0304-8853/96/$15.00 ¢) 1996 Elsevier Science B.V. All rights reserved SSDI0304-8853(95)00787-3
64
K.-M.H. Lenssen et al. / Journal of Magnetism and Magnetic Materials 156 (1996) 63-64
a:
•
= / I I
I 2
I 3
l 4
fro (rim) Fig. 2. Height of the (111) NisoFe2o peak in the X-ray diffractogram as a function of the thickness of the Ta buffer layer ((0) deposited at 10 to mbar; (©) deposited in 10 7 mbar N2).
magnetoresistance was confirmed by measurements of the anisotropic magneto-resistance (AMR) effect, where exchange biasing is absent: for 50nm NisoFe20 films on SiOx/4.6 nm Ta (showing (111) texture) and on Si without Ta (without clear texture), the A R / R ratio was respectively 2.6% and 1.9%. Since the Ta buffer layer is known to be important to initiate a good (111) texture, its properties were studied in more detail. For this purpose, a Ta wedge (thickness tTa from 1 to 5 nm) covered with 50nm Ni8oFe2o was grown on a SiO x substrate. This wedge was split in five parts and for each part the height of the (111) X-ray reflection of NisoFe2o was determined. As a reference also a NisoFe2o film without a Ta buffer layer was measured. The results are presented in Fig. 2. Although the presence of a Ta buffer layer clearly improves the texture of the NisoFe2o, the peak intensity decreases rapidly with increasing tTa above tTa = 1.5 rim. This is in contrast to the situation for sputtered spin valves where after an initial sharp increase saturation occurs at tTa = 3.0 nm [4]. The reason could be that the Ta films grow in different crystalline phases for MBE and sputter deposition. In order to investigate whether the phase required to initiate the (111) texture in the spin valve could be stabilized by impurities like N, a second, identical Ta wedge was deposited by MBE at an intentionally increased background pressure of 10 - 7 mbar N2, simulating the typical pressure during sputter deposition. However, as can be seen in Fig. 2, the (111) peak heights only decreased. In order to obtain more information about the relation between the structure of the Ta buffer layers and (111) texture in spin valves, an extended transmission electron microscopy (TEM) study was performed. For plan-view
TEM experiments, bilayers of tTa T a / 2 . 0 n m NisoFezo were deposited by MBE on TEM windows (Si3N 4 membranes) with tya = 0, 0.5, 1.0, 1.5, 2.0, 3.0 rim. To minimize oxidation effects the samples were analyzed by TEM immediately after the deposition. Only for the 0.5 and 1.0 nm Ta buffer was a relatively small fraction of (l 1 l) textured Nis0Fe2o grains observed. Electron diffraction showed that the occurrence of texture was correlated with a Ta layer consisting of randomly oriented, nm size grains, the structure of which was different from earlier reported /3-Ta, bcc-Ta and fcc-Ta phases [6]. Thicker Ta buffer layers were found to have the stable bcc-Ta phase with a = 0.3306 nm. For sputtered Ta films the desired structure for the initiation of texture was observed even in the thickest film studied (10 nm). In addition, cross-sectional (high-resolution) TEM was carried out on an MBE-grown and a sputtered spin valve with a 2.5 nm Cu spacer. (The latter was sputtered at an Ar pressure of 5 mTorr, near the optimum for the GMR ratio.) The Ta grains did not show a clear grain-to-grain epitaxy with the grains of the spin-valve stacking. The MBE spin valve showed a more undulating surface than the sputtered one, which can be expected because of the lower kinetic energy of the atoms involved in the deposition [2]. In the electron-diffraction patterns a result of the plan-view TEM studies was confirmed: randomly oriented films in the case of MBE and (111) texture in the sputtered spin valve. In conclusion, spin valves made by MBE showed a slightly lower GMR effect and a halved exchange-bias field as compared to sputtered samples (deposited at the for GMR optimal Ar pressure). The different structure of the Ta buffer layer plays an important role in this, because of its effect on the texture of the spin valve. Besides this, sharper interfaces and increased waviness were observed. References [1] V.S. Speriosu, J.P. Nozieres, B.A. Gurney, B. Dieny, T.C. Huang and H. Lefakis, Phys. Rev. B 47 (1993) 11579. [2] J.C.S. Kools, J. Appl. Phys. 77 (1995) 2993. [3] Th.G.S.M. Rijks, R. Coehoorn, J.T.F. Daemen and W.J.M. de Jonge, J. Appl. Phys. 76 (1994) 1092. [4] J.C.S. Kools, R. Coehoorn, J.P.W.B. Duchateau, Th.G.S.M. Rijks and I. Gideonse, in: Thin Films, Eds. G. Hecht, F. Richter and J. Hahn (DGM Informationsgesellschaft, Oberursel, 1994) p. 279. [5] R. Jungblut, R. Coehoorn, M.T. Johnson, J. aan de Stegge and A. Reinders, J. Appl. Phys. 75 (1994) 6659; R. Nakatani, K. Hoshino, S. Noguchi and Y. Sugita, Jpn. J. Appl. Phys. 33 (1994) 133. [6] See, for example, F. Schrey, R.D. Mathis, R.T. Payne and L.E. Murr, Thin Solid Films 5 (1970) 29, and references therein.
~ H Journalof magnetism and magnetic ~ i materials
N
ELSEVIER
MBE-grown spin valves K.-M.H. Lenssen *, J.C.S. Kools, A.E.M. De Veirman, J.J.T.M. Donkers, M.T. Johnson, A. Reinders, R. C o e h o o r n Philips Research lxlboratories. Prq~essor Holstlaan 4. NL-5656 AA Eindhot'en. The Netherland.~
Abstract Spin valves were grown by vapour deposition in an MBE system, instead of the usual sputter deposition. Compared to sputtered samples, the GMR ratio A R / R was slightly lower and the exchange-bias field was halved. Also, the (1 I1) texture was much less pronounced. The Ta buffer layer was studied by TEM analysis, providing new information about the required structure to initiate the (l I l) texture.
For the first time, we report on spin-valve multilayers that were grown by vapour deposition in a molecular beam epitaxy (MBE) system. It might be expected that MBEgrown films are superior, e.g. with respect to the interface quality, as compared to films made by means of sputter deposition. Both techniques have different characteristics, e.g. concerning the kinetic energy of the particles, possibly resulting in different film microstructures. Therefore, it is interesting to compare MBE-grown spin valves with sputtered samples. A series of spin valves with varying Cu thickness tcu was grown, consisting of S i ( 1 0 0 ) / 3 n m T a / S n m NisoF%o/tcu C u / 6 nm NisoFe2o/8 nm F%0Mn50/3 nm Ya (to, = 2.0, 2.5. 3.0 and 3.5 nm), using (co-)evaporation from elemental sources. During growth a magnetic field of 20 k A / m was applied along the long axis of the rectangular 4 × 12 mm 2 substrates; the base pressure was ~ 10-t0 mbar, and the substrate temperature was kept at 20°C. The magnetoresistance of the multilayers was measured in a four-terminal configuration (current and field both ahmg the long axis). The data are shown in Fig. 1. The largest effect of 3.9% was found for the spin valve with to, , = 2.5 rim. In order to determine the effective magnetic dead-layer thickness in MBE-grown Nis0Fe=o (Py), multilayers of the type Si(100)/3nm Ta/tpy NisoFez0/2nm C u / 3 n m Ya (tp>. = 2.0. 4.0, 6.0 and 8.0 nm) were deposited. The magnetic moment of these films was determined in a SQU1D magnetometer. A good linear relationship between magnetic moment and /py was found (slope of 625 k A / m ) , yielding a value of 0.5 nm as an estimation of the sum of
the effective dead-layer thicknesses at both interfaces. This value is smaller than the value obtained for comparable sputter-deposited systems ( ~ 0 . 8 n m [1]), which is in agreement with the expected sharper interfaces associated with a low-energetic deposition technique like MBE. The ratio A R / R of the giant magneto-resistance (GMR) effect was somewhat ( ~ 1%) less than that of sputtered spin valves [2,3]. In Fig. I it can be seen that this is not the result of insufficient exchange biasing, although the exchange-bias field is only half the value fi)und for sputtered samples. X-ray diffraction showed that the degree of (111) texture of the films was much lower, as indicated by the large width of rocking curves around the (111) peak of more than 15° (as compared to typically 5 ° for sputtered samples [4]). This can explain the reduced exchange-bias field [5], and possibly also the lower GMR ratio. That (l 1 l) texture is also directly correlated with
32
30 2~ 2~
24 2~
~
-5
I 0
i 5
1 10
I
I 15
20
tt (kA/m)
* Corresponding author. Fax: +31-40-274-4282; lenssen @natlab.research.philips.com.
email:
Fig. 1. Magnetoresistance curves of MBE-grown spin valves with a Cu-spacer thickness equal to 2.0, 2.5, 3.0 and 3.5 nm (from top to bottom), respectively.
0304-8853/96/$15.00 ¢) 1996 Elsevier Science B.V. All rights reserved SSDI0304-8853(95)00787-3
64
K.-M.H. Lenssen et al. / Journal of Magnetism and Magnetic Materials 156 (1996) 63-64
a:
•
= / I I
I 2
I 3
l 4
fro (rim) Fig. 2. Height of the (111) NisoFe2o peak in the X-ray diffractogram as a function of the thickness of the Ta buffer layer ((0) deposited at 10 to mbar; (©) deposited in 10 7 mbar N2).
magnetoresistance was confirmed by measurements of the anisotropic magneto-resistance (AMR) effect, where exchange biasing is absent: for 50nm NisoFe20 films on SiOx/4.6 nm Ta (showing (111) texture) and on Si without Ta (without clear texture), the A R / R ratio was respectively 2.6% and 1.9%. Since the Ta buffer layer is known to be important to initiate a good (111) texture, its properties were studied in more detail. For this purpose, a Ta wedge (thickness tTa from 1 to 5 nm) covered with 50nm Ni8oFe2o was grown on a SiO x substrate. This wedge was split in five parts and for each part the height of the (111) X-ray reflection of NisoFe2o was determined. As a reference also a NisoFe2o film without a Ta buffer layer was measured. The results are presented in Fig. 2. Although the presence of a Ta buffer layer clearly improves the texture of the NisoFe2o, the peak intensity decreases rapidly with increasing tTa above tTa = 1.5 rim. This is in contrast to the situation for sputtered spin valves where after an initial sharp increase saturation occurs at tTa = 3.0 nm [4]. The reason could be that the Ta films grow in different crystalline phases for MBE and sputter deposition. In order to investigate whether the phase required to initiate the (111) texture in the spin valve could be stabilized by impurities like N, a second, identical Ta wedge was deposited by MBE at an intentionally increased background pressure of 10 - 7 mbar N2, simulating the typical pressure during sputter deposition. However, as can be seen in Fig. 2, the (111) peak heights only decreased. In order to obtain more information about the relation between the structure of the Ta buffer layers and (111) texture in spin valves, an extended transmission electron microscopy (TEM) study was performed. For plan-view
TEM experiments, bilayers of tTa T a / 2 . 0 n m NisoFezo were deposited by MBE on TEM windows (Si3N 4 membranes) with tya = 0, 0.5, 1.0, 1.5, 2.0, 3.0 rim. To minimize oxidation effects the samples were analyzed by TEM immediately after the deposition. Only for the 0.5 and 1.0 nm Ta buffer was a relatively small fraction of (l 1 l) textured Nis0Fe2o grains observed. Electron diffraction showed that the occurrence of texture was correlated with a Ta layer consisting of randomly oriented, nm size grains, the structure of which was different from earlier reported /3-Ta, bcc-Ta and fcc-Ta phases [6]. Thicker Ta buffer layers were found to have the stable bcc-Ta phase with a = 0.3306 nm. For sputtered Ta films the desired structure for the initiation of texture was observed even in the thickest film studied (10 nm). In addition, cross-sectional (high-resolution) TEM was carried out on an MBE-grown and a sputtered spin valve with a 2.5 nm Cu spacer. (The latter was sputtered at an Ar pressure of 5 mTorr, near the optimum for the GMR ratio.) The Ta grains did not show a clear grain-to-grain epitaxy with the grains of the spin-valve stacking. The MBE spin valve showed a more undulating surface than the sputtered one, which can be expected because of the lower kinetic energy of the atoms involved in the deposition [2]. In the electron-diffraction patterns a result of the plan-view TEM studies was confirmed: randomly oriented films in the case of MBE and (111) texture in the sputtered spin valve. In conclusion, spin valves made by MBE showed a slightly lower GMR effect and a halved exchange-bias field as compared to sputtered samples (deposited at the for GMR optimal Ar pressure). The different structure of the Ta buffer layer plays an important role in this, because of its effect on the texture of the spin valve. Besides this, sharper interfaces and increased waviness were observed. References [1] V.S. Speriosu, J.P. Nozieres, B.A. Gurney, B. Dieny, T.C. Huang and H. Lefakis, Phys. Rev. B 47 (1993) 11579. [2] J.C.S. Kools, J. Appl. Phys. 77 (1995) 2993. [3] Th.G.S.M. Rijks, R. Coehoorn, J.T.F. Daemen and W.J.M. de Jonge, J. Appl. Phys. 76 (1994) 1092. [4] J.C.S. Kools, R. Coehoorn, J.P.W.B. Duchateau, Th.G.S.M. Rijks and I. Gideonse, in: Thin Films, Eds. G. Hecht, F. Richter and J. Hahn (DGM Informationsgesellschaft, Oberursel, 1994) p. 279. [5] R. Jungblut, R. Coehoorn, M.T. Johnson, J. aan de Stegge and A. Reinders, J. Appl. Phys. 75 (1994) 6659; R. Nakatani, K. Hoshino, S. Noguchi and Y. Sugita, Jpn. J. Appl. Phys. 33 (1994) 133. [6] See, for example, F. Schrey, R.D. Mathis, R.T. Payne and L.E. Murr, Thin Solid Films 5 (1970) 29, and references therein.