This work relates to the integration of the two-layer stack of the proposed multiferroic structure onto silicon substrates. Ba1-xSrxTiO3 is an excellent material for room-temperature voltage-tunable dielectric applications due to its high (ε=6000) dielectric constant. In this study we choose a composition of Ba0.4Sr0.6TiO3 (BST), which is cubic and paraelectric at 300K, and transforms to a ferroelectric tetragonal phase upon cooling through the Curie temperature (TC) at 200K. The main focus of the present work is to study what happens when BST is placed in contact with a room temperature ferromagnetic layer such as La0.7Sr0.3MnO3 (LSMO). In this study, the magnetic properties of a BST (200nm)/LSMO (63nm) heterostructure was compared to that of a single LSMO layer (63nm). Both films were deposited onto MgO/TiN buffered Si (100) using pulsed laser deposition (PLD) and a domain matching epitaxy (DME) paradigm. X-ray diffraction (XRD) measurements showed that these films were of single phase and epitaxial in nature, with an unrelaxed lattice strain of ∼0.2% that was predominately composed of thermal and defect-induced strain. The magnetic measurements showed that the Curie temperature (TC) of LSMO remained unchanged at 350K when the BST was in contact with the LSMO layer. Interestingly, at 4K both the coercive field (Hc) and the exchange bias (HEB) of the BST/LSMO heterostructure as compared to the lone LSMO film increased significantly from 400 to 800 Oe and from 155 to 305 Oe, respectively. These differences were found to disappear above 200 K, the ferroelectric TC of the BST over-layer. This strongly suggests that the observed changes in the magnetic behavior of the heterostructure was the result of stress and/or charge redistributions that resulted when the BST layer transformed from the cubic (paraelectric) to tetragonal (ferroelectric) phase at low temperature.

Ba1-xSrxTiO31–3 is an important material for applications such as in DRAM devices, new classes of tunable RF components, voltage-tunable dielectrics, and phase shifters, because of its large dielectric constant of more than 6000 and low microwave loss at room temperature. For this study a concentration of Ba0.4Sr0.6TiO3 (BST) is chosen, which is cubic and paraelectric at the (delete) ambient temperature, and then transforms to a ferroelectric tetragonal phase upon cooling through the Curie temperature (TC) at 200K.4 

Previous workers1–7 have deposited BST thin films on oxide substrates such as MgO, LaAlO3, SrTiO3, and Al2O3. However, high single crystal cost and small size are their (delete) major drawbacks of such substrates. Furthermore, the brittleness associated with those oxide substrates can cause difficulties in wafer-dicing, thereby reducing the device yield and reliability, further compromising affordability. In contrast, Si wafers are cheap, available in large dimensions, and are standard substrates in microelectronics. Hence, it is of great technological as well as scientific interest to integrate BST active thin films with Si substrates. In the past decade, there has been a great deal of published research8–11 reporting on the physical properties of two-phase multiferroic materials, where a ferroelectric material (such as Pb0.8Zr0.2TiO3, BaTiO3, BiFeO3) is placed in contact with ferromagnetic materials such as La0.7Sr0.3MnO3 (LSMO), CoFe, and Ni. However, to our knowledge, there have been no reports published on the magnetic properties of two-phase epitaxial multiferroic BST/LSMO epitaxially integrated with Si (100). With that in mind, we report on the magnetic properties of an epitaxial BST/LSMO thin-film heterostructure deposited onto Si (100) using a combination of TiN/MgO buffer layers. The TiN/MgO combination has excellent diffusion barrier properties. The temperature- and magnetic field-dependent measurements performed show that the Curie temperature (TC) for the LSMO and BST/LSMO samples remain constant at 350K, while the coercive field (Hc) and the exchange bias (HEB) measured at low temperature (4K) increased significantly from 400 to 800 Oe, and from 155 to 305 Oe, respectively, when BST is placed in contact with the (insert) LSMO layer. These differences were found to disappear above 200 K, the ferroelectric TC of the BST over-layer.

For thin films deposition four targets (TiN, MgO, LSMO and BST) were ablated sequentially in the same chamber using a rotating target assembly. At first, the deposition of a (insert) TiN layer of 40nm thickness (insert) was performed at 600°C with a chamber base pressure of 2 × 10-6 Torr. This was immediately followed by the deposition of an initial few mono layers of MgO (5nm) under a base vacuum of 2 × 10-6 Torr at 600°C. The remaining MgO (up to 25nm) was then deposited at the same temperature but under an oxygen pressure of 6 × 10-4 Torr. Finally, the LSMO and BST layers were deposited at a higher O2 partial pressure of 2 × 10-1 Torr and a substrate temperature of 750°C. During the entire deposition process, a laser energy density of 2.3 J/cm2 and pulse frequency of 10Hz was utilized. Upon completion of the deposition, the heterostructures were cooled slowly down to room temperature under a O2 partial pressure of 2×10-1 Torr. No other post-growth oxygen anneal was performed. To compare the magnetic properties, data of an LSMO sample without BST was also included. The thickness of the LSMO layer was kept constant at 63nm in both samples.

The structural properties of the BST/LSMO heterostructure were investigated using θ-2θ XRD scans measured with a Rigaku x-ray diffractometer using Cu Ka radiation (λ = 1.5418A°). To probe the oxidation states of the respective elements and to identify (if present) any surface metal contamination, XPS was performed using a SPECS FlexMod system equipped with an Al Kα monochromatic x-ray source (1486.7eV). The magnetic properties of the heterostructure were measured in a superconducting quantum interference device (SQUID) magnetometer (Quantum Design, MPMS-XL). All the magnetization measurements were performed in the plane of the film. The buffer layers MgO/TiN and silicon substrates were found to be diamagnetic.

Figure 1(a) presents the θ-2θ XRD pattern of the BST (200nm)/LSMO (63nm)/MgO(30nm)/TiN(40nm)/Si (100) heterostructure. This pattern shows that all the layers grew with a preferential (00l) orientation, suggesting either the textured or epitaxial growth of the multilayered structure. From the 2θ XRD data for the (002) peak, the out-of-plane (OOP) lattice parameters of BST and LSMO layer are determined as 3.956A° and 3.843A°, respectively. These are very close to the bulk lattice parameters of BST (3.965A°)1–7 and LSMO (3.870A°)12,13 layers. The estimated strain for BST and LSMO layers along the c-direction is ∼0.2% and 0.7%, respectively, which is attributed to unrelaxed thermal and defect-induced strain, and some additional unrelaxed lattice mismatch strain for the BST (delete LSMO and insert BST) layer grown on the LSMO (delete BST and insert LSMO). Φ-scan XRD was used to study the epitaxial growth and the in-plane (IP) orientation of BST layer with respect to Si (100) substrate. As displayed in Fig. 1(b), the φ-scan patterns of (111) reflection for BST and Si were measured. This XRD pattern shows 4 distinct peaks (though broad) separated by ∼90° which indicates its cubic symmetry, and establishing the cube-on-cube relationship of the BST with the underlying substrate Si (100). This confirms that all the layers are grown epitaxially as cube-on-cube i.e. (001)Epilayer//(001)Buffer and [110]Epilayer//[110]Buffer, without any rotation. It should be noted that the epitaxial growth of BST/LSMO/MgO/TiN on Si (100) is possible due to the epitaxial growth of large mismatched system based on the domain matching epitaxy (DME) paradigm,14,15 e.g., TiN (a = 4.24 A°) on Si (100) (a = 5.43 A°) where four lattice constants of TiN match with three of Si (100). According to DME paradigm, in large mismatch systems most of the lattice mismatch strain is relieved almost immediately upon initiation of growth, i.e. within the first couple of monolayers of growth. Thus, the lattice misfit strain accommodation is confined to the interfaces making it possible for the rest of the film to be grown relatively free of defects and lattice strain. In the current study, TiN (a = 4.24 A°) forms a good lattice match with MgO (a = 4.22 A°), which in turn has a 6-8 % lattice mismatch with BST (3.965A°) and LSMO (3.870A°). As reported in our previous work,16 the results obtained from the x-ray photo absorption spectroscopy (XPS) analyses confirm that the valences of Ba, Sr, Ti, and O elements of BST thin films are +2, +2, +4, and -2, respectively.

FIG. 1.

(a) Typical θ-2θ (out of plane) XRD patterns of BST(200nm)/LSMO(63nm)/MgO(30nm)/TiN(40nm)/Si (100) heterostructures showing high quality, single phase and only (00l) reflections; (b) φ-scan XRD patterns of BST and Si of (111) reflection collected at 2θ = 38.90°, ω = 19.45° and χ = 55.07° for BTO; 2θ = 40.20°, ω = 20.10° and χ = 55.07° for LSMO; and 2θ = 28.46°, ω = 14.23° and χ = 54.74° for Si (100). This pattern shows 4 peaks separated by ∼90° indicating its pseudo cubic/rhombohedral symmetry, establishing the cube-on-cube relationship with the underlying substrate Si (100).

FIG. 1.

(a) Typical θ-2θ (out of plane) XRD patterns of BST(200nm)/LSMO(63nm)/MgO(30nm)/TiN(40nm)/Si (100) heterostructures showing high quality, single phase and only (00l) reflections; (b) φ-scan XRD patterns of BST and Si of (111) reflection collected at 2θ = 38.90°, ω = 19.45° and χ = 55.07° for BTO; 2θ = 40.20°, ω = 20.10° and χ = 55.07° for LSMO; and 2θ = 28.46°, ω = 14.23° and χ = 54.74° for Si (100). This pattern shows 4 peaks separated by ∼90° indicating its pseudo cubic/rhombohedral symmetry, establishing the cube-on-cube relationship with the underlying substrate Si (100).

Close modal

Figure 2 plots the temperature dependence of magnetic moment, (M vs T) measured in the plane of the sample (<100> direction) under the magnetic field of 300 Oe after cooling the BTO/LSMO (shown in black) and LSMO (shown in red) samples from T=400 K to T=4 K. It is noted that the ferromagnetic Curie temperature (TC) of LSMO is found to be unchanged at ∼350K for both the samples. The magnetic response is not saturated under 300 Oe field cooling, which is reflected in the magnetic moment (emu). This behavior is similar to the one we reported11 in the case of BaTiO3/LSMO heterostructures. The reason for the cross-over in these two curves around 250K is believed to reflect the removal of interfacial spin pinning effects (strain and charge ordering induced effects) introduced when the BST layer in the heterostructure undergoes a ferroelectric transition at 200K.

FIG. 2.

Magnetic moment vs. temperature (M-T) curves of LSMO(63nm) (shown in red), BST(200nm)/LSMO(63nm) (shown in black). The data were collected during the warming cycle after cooling the samples under the magnetic field of 300 Oe. As it can be noticed, the Curie temperature (TC) for all of the samples is found to be the same at ∼ 350 K. The magnetic field is applied along <100> direction of the sample. The magnetic response is not saturated under a 300 Oe field, hence, it is expressed only in magnetic moment (emu), not in emu/cc.

FIG. 2.

Magnetic moment vs. temperature (M-T) curves of LSMO(63nm) (shown in red), BST(200nm)/LSMO(63nm) (shown in black). The data were collected during the warming cycle after cooling the samples under the magnetic field of 300 Oe. As it can be noticed, the Curie temperature (TC) for all of the samples is found to be the same at ∼ 350 K. The magnetic field is applied along <100> direction of the sample. The magnetic response is not saturated under a 300 Oe field, hence, it is expressed only in magnetic moment (emu), not in emu/cc.

Close modal

In Fig. 3(a), we present the isothermal M-H data collected on LSMO (63nm) layer (without BST layer) at 4K, when the sample was cooled under no magnetic field (zero field cooling, ZFC, shown in black) and under (insert) a 300 Oe (field cooling, FC, shown in red) magnetic field. We observed a noticeable exchange bias (HEB, a shift in M-H loop along the field axis) of 155 Oe and a coercive field (Hc) of 400 Oe. Most likely, this shift in M-H loop is due to the magnetic interaction between the unsaturated interfacial spins in the LSMO thin film and the spins in the bulk of LSMO. A similar low temperature intrinsic exchange bias of ∼ 200 Oe has been reported17 on LSMO thin films deposited on LaSrAlO4 (001) substrates. The authors have argued that the intrinsic exchange behavior could be originated from an antiferromagnetic (delete ‘the’, and insert ‘an antiferromagnetic’) exchange coupling between ferromagnetic LSMO layer and an interfacial (delete ‘unprecedented’, and insert ‘interfacial’) LaSrMnO4-based spin glass phase, formed under a large interfacial strain and resultant (delete ‘subsequent’, and insert ‘resultant) self-assembly. Figure 3(b) shows the variation of magnetic moment as a function of magnetic field (M-H) measured at 4K on BST(200nm)/LSMO(63nm) heterostructures. Interestingly, we observed an increase in Hc up to 800 Oe (two-fold) and HEB of 305 Oe as is reflected by the (delete ‘in’, and insert ‘by the’) M-H loop shift (represented by an arrow) between ZFC (shown in black) and FC (shown in red) M-H curves. In addition, the ZFC and FC M-H loops for BST (200nm)/LSMO (63nm) sample shifted significantly along the magnetic moment axis. It appears that one needs to employ more than 1000 Oe magnetic field to reach full saturation of the M-H loops.

FIG. 3.

Comparison of isothermal ZFC (shown in black) and FC (shown in red) (300 Oe) M-H curves measured on LSMO (63nm) sample (a), and BST(200nm)/LSMO (63nm) sample (b). The magnetic field is applied along the <100> direction of the sample. It can be noted that the coercive field (Hc) and the exchange bias (HEB) increased from 400 to 800 Oe and from 155 to 305 Oe, respectively when BST (200nm) is placed in contact with LSMO (63nm). As it can be noticed, the ZFC and FC M-H loops for BST(200nm)/LSMO(63nm) sample shifted significantly both along field and magnetic moment axis.

FIG. 3.

Comparison of isothermal ZFC (shown in black) and FC (shown in red) (300 Oe) M-H curves measured on LSMO (63nm) sample (a), and BST(200nm)/LSMO (63nm) sample (b). The magnetic field is applied along the <100> direction of the sample. It can be noted that the coercive field (Hc) and the exchange bias (HEB) increased from 400 to 800 Oe and from 155 to 305 Oe, respectively when BST (200nm) is placed in contact with LSMO (63nm). As it can be noticed, the ZFC and FC M-H loops for BST(200nm)/LSMO(63nm) sample shifted significantly both along field and magnetic moment axis.

Close modal

To better understand the origin of enhanced Hc and HEB in BST(200nm)/LSMO(63nm) sample, additional ZFC and FC m-H measurements have been performed at 200K, see Fig. 4. The shift along both the axes is absent (i.e, ZFC and FC m-H curves are overlapped) from the BST(200nm)/LSMO(63nm) sample measured at 200K, the ferroelectric Curie temperature of BST layer.1–7 The enhancements in Hc and HEB at low temperature appears to be directly related to the ferroelectric nature of BST layer, and the resulting strain (∼0.4%) and/or charge redistributions that take place as the BST layer transformed from the cubic (paraelectric) to tetragonal (ferroelectric) phase at low temperature (200K).18 This may also be consistent with the number of previous reports19–22 on the formation of an antiferromagnetic region or layer due to strong pinned spins and magnetic depletion layer observed at various ferromagnetic and ferroelectric interfaces. For instance, using in situ-magnetometry studies of magnetoelectric La1−xSrxMnO3/Pb(Zr,Ti)O3 heterostructures, Leufke and co-authors have shown the coexistence of antiferromagnetic and ferromagnetic regions at this La1−xSrxMnO3/Pb(Zr,Ti)O3 interface, argued to be (insert) due to the surface-charge dependent electronic phase separation. A similar antiferromagnetic region was (delete ‘s were’, and insert ‘was’) observed at another ferromagnetic and ferroelectric interface of La0.7Ca0.3MnO3/BaTiO3 by Alberca and co-workers. The authors argued that the evolving surface morphology and associated strain due to BaTiO3 could cause the magnetic phase separation into ferromagnetic and antiferromagnetic regions. In another interesting study reported by Zhao and co-authors, employing spin wave resonance on La0.7Sr0.3MnO3/BaTiO3 heterostructures, it has been shown that BTO layer modifies the in-plane magnetic anisotropy of an adjacent (insert) LSMO layer and also induces surface spin pinning with (insert) both in-plane and out of plane orientation (insert). In-depth interface-sensitive structural and magnetic measurements such as scanning transmission electron microscopy (STEM-Z) coupled with electron energy loss spectroscopy (EELS) and polarized neutron reflectivity (PNR) are planned to shed additional light on the current observations and identify (delete ‘pinpoint’, and insert ‘identify’) the source of the interesting magnetic properties observed in the present study. (delete: reported on BST/LSMO interface).

FIG. 4.

The ZFC (shown in black) and FC (shown in red) (300 Oe) M-H loops measured on BST(200nm)/LSMO(63nm) sample at 200K (b). We observed no exchange bias (HEB) at temperatures above the ferroelectric Curie temperature (200K) of BST(200nm) layer.

FIG. 4.

The ZFC (shown in black) and FC (shown in red) (300 Oe) M-H loops measured on BST(200nm)/LSMO(63nm) sample at 200K (b). We observed no exchange bias (HEB) at temperatures above the ferroelectric Curie temperature (200K) of BST(200nm) layer.

Close modal

BST (200 nm)/LSMO (63 nm) heterostructures were deposited onto (delete ‘integrated’, and insert ‘deposited onto’) MgO(30nm)/TiN(40nm) buffered Si (100) using pulsed laser deposition. XRD measurements showed (insert) that these films were (delete ‘are’, and insert ‘were’) of single phase and epitaxial in nature. The estimated strain for BST and LSMO layers along the c-direction was (delete ‘are’, and insert ‘was’) ∼0.2% and 0.7%, respectively, which is attributed to unrelaxed thermal and defect-induced strain, and some additional unrelaxed lattice mismatch strain for the BST (delete ‘LSMO’, and insert ‘BST’) layer grown on the LSMO (delete ‘BST’, and insert ‘LSMO’). The temperature- and magnetic field-dependent measurements carried out on BST/LSMO layer showed that the Curie temperature (TC) of LSMO remained (delete ‘remains’, and insert ‘remained) constant 350 K. More interestingly, the coercive field and the exchange bias (delete ‘were’) increased from 400 to 800 Oe and 155 to 305 Oe measured at 4K, when BST was (delete ‘is’, and insert ‘was’) in contact with LSMO layer. Such HEB was (delete ‘is’, and insert ‘was’) found to disappear at 200 K, which is the ferroelectric TC of BST layer. From our experimental findings, the ferroelectric nature of BST was (delete ‘is’, and insert ‘was’) believed to produce these observed shifts in the magnetic properties of BST/LSMO heteterostructure by pinning the spins at the BST/LSMO interface via stress and charge induced effects that are induced by the ferroelectric transition. This work can advance the knowledge in integrating the multiferroic structures with CMOS compatible substrates such as Si (100).

Part of this work was funded by UTEP start-up grants. SRS acknowledges financial support from the National Academy of Science (NAS), USA. Also, the authors are pleased to acknowledge the support of the Army Research Office under Grant W911NF-16-2-0038. Also, the authors acknowledge the use of the Analytical Instrumentation Facility (AIF) at North Carolina State University, which is supported by the State of North Carolina and the National Science Foundation.

1.
J. F.
Scott
,
Annu. Rev. Mater. Sci.
28
,
79
(
1998
).
2.
A.
Tombak
,
J.-P.
Maria
,
F.
Ayguavives
,
Z.
Jin
,
G. T.
Stauf
,
A. I.
Kingon
, and
A.
Mortazawi
,
IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS
12
,
3
(
2002
).
3.
C. J. G.
Meyers
,
C. R.
Freeze
,
S.
Stemmer
, and
R. A.
York
,
Appl. Phys. Lett.
109
,
112902
(
2016
).
4.
A.
Ahmed
,
I. A.
Goldthorpe
, and
A. K.
Khandani
,
App. Phys. Revs.
,
2
,
011302
(
2015
).
5.
W.
Chang
,
J. S.
Horwitz
,
A. C.
Carter
,
J. M.
Pond
,
S. W.
Kirchoefer
,
C. M.
Gilmore
, and
D. B.
Christy
,
Appl. Phys. Lett.
74
,
1033
(
1999
).
6.
S. B.
Qadri
,
J. S.
Horwitz
,
D. B.
Chrisey
,
R. C. Y.
Auyeung
, and
K. S.
Grabowski
,
Appl. Phys. Lett.
66
,
1605
(
1995
).
7.
Y.
Gim
,
T.
Hudson
,
Y.
Fan
,
C.
Kwon
,
A. T.
Findikoglu
,
B. J.
Gibbons
,
B. H.
Park
, and
Q. X.
Jia
,
Appl. Phys. Lett.
77
,
1200
(
2000
).
8.
J.
Chakhalian
,
J. W.
Freeland
,
A. J.
Millis
,
C.
Panagopoulos
, and
J. M.
Rondinelli
,
Rev. of. Mod. Phys.
86
,
1189
(
2014
).
9.
J. T.
Heron
,
D. G.
Schlom
, and
R.
Ramesh
,
Appl. Phys. Rev.
,
1
,
021303
(
2014
).
10.
S. H.
Baek
and
C. B.
Eom
,
Acta Materialia
61
,
2734
(
2013
).
11.
C.
Lu
,
W.
Hu
,
Y.
Tian
, and
T.
Wu
,
Appl. Phys. Revs.
2
,
021304
(
2015
).
12.
S. S.
Rao
,
J. T.
Prater
,
F.
Wu
,
C. T.
Shelton
,
J.-P.
Maria
, and
J.
Narayan
,
Nano Lett.
13
,
5814
(
2013
).
13.
S. R.
Singamaneni
,
F.
Wu
,
J. T.
Prater
, and
J.
Narayan
,
J. Appl. Phys.
116
,
224104
(
2014
).
14.
J.
Narayan
,
P.
Tiwari
,
X.
Chen
,
R.
Chowdhury
, and
T.
Zheleva
,
Appl. Phys. Lett.
61
,
1290
(
1992
).
15.
J.
Narayan
and
B. C.
Larson
,
J. Appl. Phys.
93
,
278
(
2003
).
16.
S. R.
Singamaneni
,
J. T.
Prater
,
S.
Punugupati
, and
J.
Narayan
,
Appl. Phys. Revs
108
,
142407
(
2016
).
17.
B.
Cui
,
C.
Song
,
G. Y.
Wang
,
H. J.
Mao
,
F.
Zeng
, and
F.
Pan
,
SCIENTIFIC REPORTS
3
,
2542
(
2013
).
18.
S.
Ríos
,
J. F.
Scott
,
A.
Lookman
,
J.
McAneney
,
R. M.
Bowman
, and
J. M.
Gregg
,
J. Appl. Phys.
99
,
024107
(
2006
).
19.
P. M.
Leufke
,
R.
Kruk
,
R. A.
Brand
, and
H.
Hahn
,
Phys. Rev. B
87
,
094416
(
2013
).
20.
A.
Alberca
,
C.
Munuera
,
J.
Tornos
,
F. J.
Mompean
,
N.
Biskup
,
A.
Ruiz
,
N. M.
Nemes
,
A.
de Andres
,
C.
León
,
J.
Santamaría
, and
M.
García-Hernández
,
Phys. Rev. B
86
,
144416
(
2012
).
21.
A.
Alberca
,
C.
Munuera
,
J.
Azpeitia
,
B.
Kirby
,
N. M.
Nemes
,
A. M.
Perez-Muñoz
,
J.
Tornos
,
F. J.
Mompean
,
C.
Leon
,
J.
Santamaria
, and
M.
Garcia-Hernandez
,
Scientific Reports
5
,
17926
(
2015
).
22.
Y.-L.
Zhao
,
Y.
Sun
,
L.-Q.
Pan
,
K.-S.
Li
, and
D.-B.
Yu
,
Appl. Phys. Lett.
102
,
042404
(
2013
).