U.S. patent application number 14/842038 was filed with the patent office on 2015-12-24 for hybridized oxide capping layer for perpendicular magnetic anisotropy.
The applicant listed for this patent is Headway Technologies, Inc.. Invention is credited to Keyu Pi, Ru-Ying Tong, Yu-Jen Wang.
Application Number | 20150372224 14/842038 |
Document ID | / |
Family ID | 51063881 |
Filed Date | 2015-12-24 |
United States Patent
Application |
20150372224 |
Kind Code |
A1 |
Pi; Keyu ; et al. |
December 24, 2015 |
HYBRIDIZED OXIDE CAPPING LAYER FOR PERPENDICULAR MAGNETIC
ANISOTROPY
Abstract
A hybrid oxide capping layer (HOCL) is disclosed and used in a
magnetic tunnel junction to enhance thermal stability and
perpendicular magnetic anisotropy in an adjoining reference layer.
The HOCL has an interface oxide layer adjoining the reference layer
and one or more transition metal oxide layers wherein each of the
metal layers selected to form a transition metal oxide has an
absolute value of free energy of oxide formation less than that of
the metal used to make the interface oxide layer. One or more of
the HOCL layers is under oxidized. Oxygen from one or more
transition metal oxide layers preferably migrates into the
interface oxide layer during an anneal to further oxidize the
interface oxide. As a result, a less strenuous oxidation step is
required to initially oxidize the lower HOCL layer and minimizes
oxidative damage to the reference layer.
Inventors: |
Pi; Keyu; (San Jose, CA)
; Wang; Yu-Jen; (San Jose, CA) ; Tong;
Ru-Ying; (Los Gatos, CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Headway Technologies, Inc. |
Milpitas |
CA |
US |
|
|
Family ID: |
51063881 |
Appl. No.: |
14/842038 |
Filed: |
September 1, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
13935826 |
Jul 5, 2013 |
9147833 |
|
|
14842038 |
|
|
|
|
Current U.S.
Class: |
257/421 |
Current CPC
Class: |
H01L 43/08 20130101;
H01L 43/10 20130101; H01L 27/222 20130101; H01L 43/04 20130101;
H01L 43/06 20130101; H01L 43/02 20130101; G11C 11/161 20130101;
H01L 43/12 20130101; H01L 43/14 20130101 |
International
Class: |
H01L 43/02 20060101
H01L043/02; H01L 43/10 20060101 H01L043/10 |
Claims
1. A magnetic tunnel junction (MTJ), comprising: (a) a reference
layer that forms a first interface with a surface of the tunnel
barrier layer, and that forms a second interface with an adjoining
interface oxide layer in a hybrid oxide capping layer structure;
(b) a tunnel barrier layer formed between the reference layer and a
free layer; (c) the free layer that contacts a tunnel barrier
surface opposite the first interface; and (d) the hybrid oxide
capping layer structure comprised of: (1) the adjoining interface
oxide layer which is made of a first metal or first alloy; (2) a
first transition metal oxide layer formed from a second metal or
second alloy that has an absolute value of free energy of oxide
formation less than that of the first metal or first alloy; and (3)
a second transition metal oxide layer formed from a third metal or
third alloy that has an absolute value of free energy of oxide
formation less than that of the first metal or first alloy.
2. The MTJ of claim 1 wherein the MTJ has a bottom spin valve
configuration wherein a seed layer, the hybrid oxide capping layer,
the reference layer, the tunnel barrier layer, and the free layer
are sequentially formed on a substrate, and each of the free layer
and reference layer exhibit perpendicular magnetic anisotropy.
3. The MTJ of claim 1 wherein the adjoining interface oxide capping
layer is comprised of one or more of MgTaOx, SrTiOx, BaTiOx,
CaTiOx, LaAlOx, MgO, TaOx, MnOx, VOx, and BOx.
4. The MTJ of claim 1 wherein each of the first transition metal
oxide layer and second transition metal oxide layer is comprised of
RuOx, PtOx, RhOx, MoOx, WOx, SnOx, or InSnOx.
5. The MTJ of claim 1 wherein the free layer is comprised of one or
more of CoFeB, CoFe, Co, Fe, CoB, FeB, and CoFeNiB.
6. The MTJ of claim 1 wherein the hybrid oxide capping layer has a
resistance x area (RA) value that is reduced by decreasing a
thickness of the hybrid oxide capping layer or reducing an
oxidation state in one or more of the adjoining interface oxide
layer, first transition metal oxide layer, and second transition
metal oxide layer.
7. The MTJ of claim 1 wherein one or more of the adjoining
interface oxide layer, first transition metal oxide layer, and
second transition metal oxide layer comprise less than a
stoichiometric amount of oxygen such that there is unoxidized metal
or unoxidized alloy remaining in the hybrid oxide capping
layer.
8. The MTJ of claim 1 wherein the hybrid oxide capping layer is
doped with one or more of Fe, Co, Ni, Ru, Cr, Au, Ag, and Cu to
increase conductivity therein.
Description
[0001] This is a continuation of U.S. patent application Ser. No.
13/935,826, filed on Jul. 5, 2013, which is herein incorporated by
reference in its entirety, and assigned to a common assignee.
RELATED PATENT APPLICATION
[0002] This application is related U.S. Pat. No. 9,006,704, which
is assigned to a common assignee and is herein incorporated by
reference in its entirety.
TECHNICAL FIELD
[0003] The present disclosure relates to a magnetic tunnel junction
(MTJ) in which a hybrid oxide capping layer is used to enhance
perpendicular magnetic anisotropy (PMA) in an adjoining free layer
to increase free layer coercivity and thermal stability while
maintaining other properties including magnetoresistance (MR) ratio
and resistance x area (RA) value.
BACKGROUND
[0004] Perpendicular magnetic anisotropy (PMA) is widely used in
devices requiring out-of-plane magnetization including Spin Torque
Magnetic Random Access Memory (STT-MRAM) that has been described by
C. Slonczewski in "Current driven excitation of magnetic
multilayers", J. Magn. Magn. Mater. V 159, L1-L7 (1996). In
STT-MRAM, a magnetic layer with PMA can serve as a free layer,
pinned layer, reference layer, or dipole compensation layer. PMA
layers are found in various designs of PMA spin valves, magnetic
tunnel junctions (MTJs), in PMA media in magnetic sensors and
magnetic data storage, and within other spintronic devices.
[0005] Compared with conventional MRAM, STT-MRAM has an advantage
in avoiding the half select problem and writing disturbance between
adjacent cells. The spin-transfer effect arises from the spin
dependent electron transport properties of
ferromagnetic-spacer-ferromagnetic multilayers. When a
spin-polarized current transverses a magnetic multilayer in a
current perpendicular to plane (CPP) configuration, the spin
angular moment of electrons incident on a ferromagnetic layer
interacts with magnetic moments of the ferromagnetic layer near the
interface between the ferromagnetic layer and non-magnetic spacer.
Through this interaction, the electrons transfer a portion of their
angular momentum to the ferromagnetic free layer. As a result,
spin-polarized current can switch the magnetization direction of
the ferromagnetic free layer if the current density is sufficiently
high, and if the dimensions of the multilayer are small.
[0006] For STT-MRAM to be viable in the 90 nm technology node and
beyond, the ultra-small MTJs (also referred to as nanomagnets) must
exhibit a MR ratio that is much higher than in a conventional
MRAM-MTJ which uses a NiFe free layer and AlOx as the tunnel
barrier layer. The critical current density (Jc) must be lower than
about 10.sup.6 A/cm.sup.2 to be driven by a CMOS transistor that
can typically deliver 100 .mu.A per 100 nm gate width. Furthermore,
a ferromagnetic layer with a long retention time is important for
device application. To achieve this property that requires a high
thermal stability, a free layer made of PMA material is preferred
in order to provide a high energy barrier (Eb) and high coercivity.
Strong PMA character is induced along an interface of a CoFeB layer
or the like and a metal oxide such as MgO, for example. A
ferromagnetic free layer must be thin enough so that induced PMA
overcomes in-plane anisotropy. Intrinsic PMA is realized in
laminated stacks including (Co/Pt).sub.n, (Co/Pd).sub.n, and
(Co/Ni).sub.n where n is the number of laminations but the MTJ may
suffer from a lower magnetoresistive (MR) ratio than when CoFe or
CoFeB is used for the free layer and/or reference layer.
[0007] When a memory element uses a free layer with a magnetic
moment lying in the plane of the film, the current needed to change
the magnetic orientation of a magnetic region is proportional to
the net polarization of the current, the volume, magnetization,
Gilbert damping constant, and anisotropy field of the magnetic
region to be affected. The critical current (i.sub.C) required to
perform such a change in magnetization is given in equation
(1):
i c = .alpha. e VMS g [ H k eff , + 1 2 H k eff , .perp. ] ( 1 )
##EQU00001##
where e is the electron charge, .alpha. is a Gilbert damping
constant, Ms is the saturation magnetization of the free layer, h
is the reduced Plank's constant, g is the gyromagnetic ratio,
H.sub.k.sub.eff.sub.,.parallel. the in-plane anisotropy field, and
H.sub.k.sub.eff.sub.,.perp. is the out-of-plane anisotropy field of
the magnetic region to switch, and V is the volume of the free
layer. For most applications, spin polarized current must be as
small as possible.
[0008] The value .DELTA.=kV/k.sub.BT is a measure of the thermal
stability of the magnetic element. If the magnetization lies
in-plane, the value can be expressed as shown in equation (2):
.DELTA. = M s VH k eff , 2 k B T ( 2 ) ##EQU00002##
where k.sub.B is the Boltzmann constant and T is the
temperature.
[0009] Unfortunately, to attain thermal stability of the magnetic
region, a large net magnetization is required which in most cases
would increase the spin polarized current necessary to change the
orientation of the magnetic region.
[0010] When the free layer has a magnetization direction
perpendicular to the plane of the film, the critical current needed
to switch the magnetic element is directly proportional to the
perpendicular anisotropy field as indicated in equation (3):
i c = .alpha. e MsVH k eff , .perp. g ( 3 ) ##EQU00003##
[0011] The parameters in equation (3) were previously explained
with regard to equation (1).
[0012] Thermal stability is a function of the perpendicular
anisotropy field as shown in equation (4):
.DELTA. = M s VH k eff , .perp. 2 k B T ( 4 ) ##EQU00004##
[0013] In both of the in-plane and out-of-plane configurations, the
perpendicular anisotropy field of the magnetic element is expressed
in equation (5) as:
H k eff , .perp. = - 4 .pi. M s + 2 K U .perp. , s M s d + H k ,
.chi. , .perp. ( 5 ) ##EQU00005##
where M.sub.s is the saturation magnetization, d is the thickness
of the magnetic element, H.sub.k,.chi.,.perp. is the crystalline
anisotropy field in the perpendicular direction, and
K.sub.U.sup..perp.,s is the surface perpendicular anisotropy of the
top and bottom surfaces of the magnetic element. In the absence of
strong crystalline anisotropy, the perpendicular anisotropy field
of a magnetic layer is dominated by the shape anisotropy field
(-4.pi.M.sub.s) on which little control is available. However, by
enhancing the surface (interfacial) perpendicular anisotropy
component, the perpendicular anisotropy (PMA) field is increased.
Although MTJ structures with reference layer/tunnel barrier/free
layer configuration such as CoFeB/MgO/CoFeB deliver a high MR
ratio, there is still a need to enhance the PMA field component in
a MTJ for higher thermal stability while maintaining a high MR
ratio.
SUMMARY
[0014] One objective of the present disclosure is to provide a
composite capping layer that enhances PMA in an adjoining free
layer within a MTJ element thereby improving thermal stability
without degrading MR ratio or other magnetic properties in the
magnetic device.
[0015] A second objective of the present disclosure is to provide a
method of forming the composite capping layer according to the
first objective that avoids the diffusion of oxygen into the free
layer.
[0016] According to one embodiment, these objectives are achieved
with a magnetic tunnel junction (MTJ) comprised of a reference
layer, tunnel barrier layer, free layer, and a hybrid oxide capping
layer that includes at least an interface oxide layer contacting
the free layer at a first interface, and an upper oxide layer that
has an absolute value of free energy of oxide formation less than
that of the interface oxide layer. In an alternative embodiment,
the upper oxide layer may be a laminate of oxides each having an
absolute value of free energy of oxide formation less than that of
the interface oxide layer. In other words, the metal oxide selected
for the upper oxide layer should not attract oxygen from the
interface oxide layer during an anneal step which would undesirably
decrease the induced perpendicular anisotropy at the free
layer/HOCL interface. The one or more metals selected to form the
upper oxide layer should be more difficult to oxidize than the
metal or alloy chosen for the interface oxide layer. Preferably,
the interface oxide layer is in an under oxidized state following
an initial oxidation step and becomes further oxidized by
attracting oxygen from the upper oxide layer during a subsequent
anneal step. As a result, the initial oxidation step is
accomplished with weak oxidation conditions such as low oxygen
pressure and/or short oxidation time that reduce the risk of
oxidative damage to the free layer.
[0017] The upper oxide layer may be oxidized with less than a
stoichiometric amount of oxygen to give an underoxidized state in
order to promote good conductivity. High conductivity may also be
realized in the HOCL through doping of one or more oxide layers
with a metal component, or employing a phase transition composition
such as .alpha.-Ta in a TaOx layer rather than .beta.-Ta. In one
aspect, the interface oxide layer induces or enhances PMA in the
free layer, and the HOCL preferably has a thickness from 3 to 15
Angstroms to minimize the RA contribution to the MTJ. Moreover, the
tunnel barrier layer is preferably an oxide of a metal or alloy and
induces or enhances PMA in the free layer along a second
interface.
[0018] The free layer may be a single layer or a composite that is
an alloy of Fe with one or more of Co, Ni, and B and with a
thickness between 5 and 20 Angstroms. Preferably, the free layer is
thin enough so that the perpendicular surface anisotropy field at
the two oxide interfaces dominates the shape anisotropy field in
the plane of the free layer. The MTJ may have a bottom spin valve
structure represented by seed layer/reference layer/tunnel
barrier/free layer/HOCL or by seed layer/HOCL/reference
layer/tunnel barrier/free layer/capping layer. In another
embodiment with a top spin valve scheme, the MTJ stack has a seed
layer/HOCL/free layer/tunnel barrier/reference layer/capping layer
configuration.
[0019] In yet another embodiment, the free layer (FL) and HOCL may
form a laminate (FL/HOCL).sub.m where m.gtoreq.2 and is the number
of laminations. The sum of the thicknesses from the plurality of
HOCL layers must be controlled to maintain an acceptable RA value
for the MTJ stack.
[0020] According to a preferred method of forming a HOCL, a first
metal or alloy layer is deposited on the free layer and is
partially oxidized to form an interface oxide such as MgO, MgTaOx,
SrTiOx, BaTiOx, CaTiOx, LaAlOx, MnOx, or VOx. Then, a second metal
layer or metal laminate is deposited on the interface oxide layer.
Subsequently, a second oxidation process is performed to at least
partially oxidize the second metal layer or multiple metal layers
in a metal laminate to form one or more transition metal oxides
including RuOx, PtOx, RhOx, MoOx, WOx, SnOx, or InSnOx. During an
anneal step, oxygen from the transition metal oxide layer migrates
downward to further oxidize the interface oxide layer. In an
alternative embodiment, the second metal is deposited on an
unoxidized first metal or alloy layer. The second metal layer is at
least partially oxidized. During a subsequent anneal process,
oxygen from the upper oxide layer diffuses downward to at least
partially oxidize the first metal or alloy layer. As a result,
oxygen diffusion into the free layer is effectively avoided during
metal oxidation steps and during a post-anneal process to yield a
high MR ratio and enhanced PMA.
BRIEF DESCRIPTION OF THE DRAWINGS
[0021] FIG. 1 is a cross-sectional view showing a magnetic tunnel
junction (MTJ) with a hybrid oxide capping layer (HOCL) and a
bottom spin valve configuration according to one embodiment of the
present disclosure.
[0022] FIG. 2 is a cross-sectional view of a MTJ with a HOCL
according to a second embodiment of the present disclosure wherein
the MTJ has a SAF reference layer and a non-magnetic insertion
layer in the free layer (FL).
[0023] FIG. 3 is a cross-sectional view of a MTJ that features a
laminated stack with a (FL/HOCL).sub.m configuration where m is at
least 2.
[0024] FIG. 4 is a cross-sectional view of a bottom spin valve
structure where a HOCL is formed between a seed layer and a
reference layer.
[0025] FIG. 5 is a cross-sectional view of a top spin valve
structure wherein a HOCL is used to enhance magnetic properties in
an overlying free layer.
[0026] FIG. 6 is a plot of Kerr signal vs. applied field showing
PMA in a MTJ with a CoFeB free layer and a MgTaOx capping
layer.
[0027] FIG. 7 is a plot of Kerr signal vs. applied field that
illustrates enhanced PMA in a MTJ with a CoFeB free layer and a
MgTaOx/RuOx HOCL formed according to a method of the present
disclosure.
[0028] FIG. 8 is a plot of Kerr signal vs. applied field for a MTJ
stack in which a HOCL is formed with an upper oxide layer having a
higher entropy of oxide formation than a MgO layer.
[0029] FIG. 9 is a table that lists the free energy of oxide
formation for different elements.
[0030] FIG. 10 is a cross-sectional view of a bottom spin valve
structure wherein a HOCL is used to enhance magnetic properties in
an overlying reference layer according to an embodiment of the
present disclosure.
DETAILED DESCRIPTION
[0031] The present disclosure is a magnetic element wherein a
hybrid oxide capping layer is used to enhance PMA in an adjoining
free layer or another ferromagnetic layer thereby improving thermal
stability of the magnetic element. Although the exemplary
embodiments depict a bottom spin valve or top spin valve
configuration, a dual spin valve design is also within the scope of
this disclosure as appreciated by those skilled in the art. The
magnetic element may be employed as a MTJ in a STT-MRAM or in other
spintronic devices, or as a propagation medium for a domain wall in
a domain wall motion device. The terms interfacial perpendicular
anisotropy and surface perpendicular anisotropy may be used
interchangeably. Absolute value of free energy of oxide formation
as used herein is a non-negative value. Since most free energy of
oxide formation values are negative, elements near the top of the
table in FIG. 9 have smaller absolute values of free energy of
oxide formation than elements near the bottom of the table.
[0032] Referring to FIG. 1, the most general embodiment of the
present disclosure is illustrated wherein a MTJ 1 is comprised of a
buffer layer 11, a reference layer 12, tunnel barrier 16, free
layer 17, and a hybrid oxide capping layer (HOCL) 21 that are
sequentially formed on a substrate 10. A top electrode 30 also
known as a bit line may be formed on a top surface of the HOCL. The
substrate may be a bottom electrode in a STT-MRAM, for example. The
buffer layer preferably includes one or more seed layers for
enhancing uniform crystal growth in overlying layers, and may
enhance or induce PMA in the overlying reference layer along a
first interface 12b. The buffer layer may be comprised of NiCr,
TiN, NiFe, NiFeCr, Mg/Ta, or other suitable seed layer
materials.
[0033] Preferably, both of the reference layer and free layer
exhibit PMA for optimum thermal stability in the MTJ element. In
one aspect, the reference layer is made of one or more
ferromagnetic layers (RL) including CoFe, CoFeB, Co, CoFeB/Co, and
the like. Although a reference layer with a RL composition
typically exhibits in-plane anisotropy, a thin reference layer from
about 5 to 20 Angstroms thick with one of the aforementioned
compositions may have PMA that is induced along top and bottom
interfaces 12b, 12t, respectively, wherein the PMA field in
reference layer 12 overcomes in-plane anisotropy. In an alternative
embodiment, the reference layer has intrinsic PMA derived from a
(A1/A2).sub.n laminated structure where A1 is a first metal or
alloy selected from one or more of Co, Ni, and Fe, A2 is a second
metal or alloy selected from one or more of Co, Fe, Ni, Pt, and Pd,
n is the number of laminates in the (A1/A2).sub.n stack, and A1 is
unequal to A2. Thus, the reference layer 12 may be one of
(Co/Pt).sub.n, (Co/Pd).sub.n, (Fe/Pt).sub.n, (Fe/Pd).sub.n,
(Co/Ni).sub.n, (CoFe/Ni).sub.n, Co/NiFe).sub.n, (Co/NiCo).sub.n,
(CoFe/NiFe).sub.n, or (CoFe/NiCo).sub.n, for example. The present
disclosure also anticipates that the reference layer may be a
composite such as (A1/A2).sub.n/CoFeB/Co with a lower (A1/A2).sub.n
laminated structure on the buffer layer, and an upper RL layer that
contacts the tunnel barrier layer 16. Furthermore, there may be a
non-magnetic spacer such as Ta formed between the (A1/A2).sub.n
laminate and the upper layer to give a (A1/A2).sub.n/Ta/RL
configuration.
[0034] Reference layer 12 contacts the tunnel barrier layer 16
along a second interface 12t. Preferably, the tunnel barrier layer
is a metal oxide including but not limited to MgO, Al.sub.2O.sub.3,
TiOx, ZnOx, and HfOx or laminates thereof formed between the
reference layer and an overlying free layer 17. As a result, the
tunnel barrier layer induces a spin dependent tunneling effect
between the reference layer and free layer, induces PMA in a
portion of the reference layer proximate to the second interface,
and induces PMA in a portion of the free layer along a third
interface 17b. When the free layer has a magnetic moment aligned in
a z-axis direction that is parallel to the magnetic moment of the
reference layer, a "0" memory state is realized. When the magnetic
moments of the free layer and reference layer are aligned
anti-parallel to one another along the z-axis, then a "1" memory
state exists.
[0035] According to one embodiment, a MgO tunnel barrier 16 is
formed by a natural oxidation (NOX) process whereby a first Mg
layer is sputter deposited on the reference layer and is
subsequently oxidized by a NOX method. Typically, the NOX process
comprises an oxygen flow in an oxidation chamber within a sputter
deposition main frame system. A subsequent annealing step after the
remaining MTJ layers are laid down essentially forms a uniform MgO
tunnel barrier layer wherein the second Mg layer in the tunnel
barrier stack is oxidized by oxygen diffusion from the underlying
MgO layer and by gettering oxygen from the free layer 17.
Optionally, the Mg deposition and NOX oxidation sequence may be
repeated one or more times before the uppermost Mg layer is
deposited on the tunnel barrier stack and subsequently oxidized
during an anneal step.
[0036] According to one embodiment, the free layer (FL) is a
ferromagnetic layer comprised of one or more of CoFeB, CoFe, Co,
Fe, CoB, FeB, and CoFeNiB, and preferably has a thickness between 5
and 20 Angstroms so that PMA induced along an interface 17b with
the tunnel barrier and PMA induced along an interface 17t with an
interface oxide layer 22 exceeds the shape anisotropy field in the
free layer to generate out-of-plane anisotropy in FL. When the free
layer has a Co.sub.WFe.sub.YB.sub.Z composition, preferably y is
greater than w, and z is <35 atomic %.
[0037] In an alternative embodiment wherein the free layer is made
of CoFeB, a non-magnetic spacer (S) such as Ta, Mg, Zr, Hf, Mo, W,
or Nb may be included to give a CoFeB/S/CoFeB or FL1/S/FL2
configuration where bottom and top ferromagnetic layers in the free
layer are designated as FL1 and FL2, respectively, as illustrated
in FIG. 2. In this case, the non-magnetic spacer has a thickness
from 0.5 to 10 Angstroms, and preferably 1 to 5 Angstroms. The
non-magnetic spacer serves as a moment diluting layer to decrease
magnetization in the free layer and thereby reduces the
demagnetizing field of the magnetic element. The thickness of each
of the FL1 and FL2 layers is preferably from 7 to 15 Angstroms.
Thus, the combined thickness of the FL1 and FL2 layers may be
greater than that of a single FL layer described above in order to
increase free layer volume and thereby compensate for a loss of Ms
in equation (2) presented previously. As a result, thermal
stability in a composite free layer with a FL1/S/FL2 configuration
is not adversely affected compared with a free layer FL.
[0038] Returning to FIG. 1, the free layer 17 may be a composite
with a lower FL portion adjoining the tunnel barrier layer 16, and
an upper (A1/A2).sub.n stack as described previously with respect
to reference layer composition to give a FL/(A1/A2).sub.n
configuration. Moreover, a non-magnetic spacer such as Ta may be
inserted in the aforementioned composite to provide a
FL/Ta/(A1/A2).sub.n free layer configuration.
[0039] In the aforementioned free layer configurations, there is
strong perpendicular surface anisotropy (K.sub.U1.sup..perp.,s in
equation 5) at interface 17b. Similarly, interface oxide layer 22
is responsible for strong perpendicular surface anisotropy
(K.sub.U2.sup..perp.,s) along interface 17t. The total thickness d
of the free layer 17 is thin enough so that the interfacial
perpendicular anisotropy
(K.sub.U1.sup..perp.,s+K.sub.U2.sup..perp.,s)/M.sub.sd is
significant compared with the shape anisotropy field. For example,
the shape anisotropy field 4.pi.Ms for a Co.sub.20Fe.sub.60B.sub.20
free layer is approximately 12000 Oe. We have found interfacial
perpendicular anisotropy may be greater than 12000 Oe by
maintaining thickness d in a range of 5 to 25 Angstroms, and
preferably 5 to 20 Angstroms. Under such conditions, an
out-of-plane magnetization is established in the free layer.
[0040] A key feature of the present disclosure is a hybrid oxide
capping layer (HOCL) 21 formed on the free layer 17. The HOCL has
at least a bottom interface oxide layer 22 and a transition metal
oxide 23 contacting a top surface of the interface oxide. In the
exemplary embodiment, a second transition metal oxide layer 24 is
formed as the uppermost layer in the HOCL wherein each transition
metal or alloy included in oxide layers 23, 24 has an absolute
value of free energy of oxide formation less than that of the metal
in the interface oxide layer. FIG. 9 lists a table of free energy
of oxide formation values for various elements. Preferably, each of
the first and second transition metals has a higher position in the
table and is more difficult to oxidize than the metal selected for
the interface oxide layer. In other words, each of the first and
second transition metal oxides has free energy of oxide formation
that is a smaller negative number and smaller in absolute value
than the interface oxide. The present disclosure encompasses a HOCL
scheme wherein a plurality of transition metal oxide layers may be
formed on the interface oxide as a laminated stack. Moreover, each
of the plurality of metals selected for the transition metal oxide
layers in the laminated stack has an absolute value of free energy
of oxide formation less than that of the metal chosen for the
interface oxide layer.
[0041] HOCL 21 contributes a RA value such that the overall RA for
the MTJ is a combination of RA from the tunnel barrier (RA.sub.TB)
and RA from the HOCL (RA.sub.HOCL) where total
RA=(RA.sub.TB)+(RA.sub.HOCL). RA value is dependent on the
thickness of an oxide layer whether it is the tunnel barrier layer
16 or the HOCL. For example, reducing the thickness of one or both
of tunnel barrier layer and HOCL decreases the total RA value.
Furthermore, total RA value may lowered by reducing the oxidation
state in one or more oxide layers within the tunnel barrier layer
and HOCL. When a NOX or ROX process is employed, a lower oxidation
state (under oxidized state) may be achieved in the tunnel barrier
and HOCL by using a shorter oxidation time or a lower O.sub.2 flow
rate to oxidize one or more metal layers.
[0042] According to one embodiment, the interface oxide layer 22 is
one or more of MgTaOx, SrTiOx, BaTiOx, CaTiOx, LaAlOx, MgO, TaOx,
MnOx, VOx, and BOx. Preferably, the interface oxide is MgTaOx or
the like that has a RA value less than that of an equivalent
thickness of MgO. The transition metal oxide layers 23, 24 are
preferably comprised of one or more of RuOx, PtOx, RhOx, MoOx, WOx,
SnOx, or InSnOx wherein the transition metal or alloy used to form
the first transition metal oxide (TM1Ox) 23 is different from the
transition metal or alloy selected to form the second transition
metal oxide (TM2Ox) layer 24. Thus, the HOCL stack may have a
MgO/TaOx/TM1O/TM2Ox or MgTaOx/TM1Ox/TM2Ox configuration, for
example, depending on whether Mg and Ta are deposited as individual
metal layers or as an alloy. As mentioned earlier, the present
disclosure anticipates that transition metal oxide layer 24 may be
omitted to give a HOCL bilayer design comprising oxide layers 22,
23.
[0043] One or more of the oxide layers 22-24 in the HOCL may be in
an underoxidized state meaning less than a stoichiometric amount of
oxygen is used to partially oxidize a metal or alloy layer. As a
result, one or more oxide layers 22-24 in the HOCL may be comprised
of a certain number of unoxidized metal (or alloy) atoms. However,
as the oxygen content in the interface oxide layer decreases, the
strength of the induced PMA field in the adjoining free layer is
also reduced. Thus, a compromise must be reached between a
sufficiently high oxidation state in the interface oxide layer to
generate or enhance PMA in the free layer 17 while avoiding a
saturated or "over" oxidized state where unreacted oxygen is able
to diffuse into the free layer from the interface oxide layer 22
and cause oxidative damage to degrade magnetic performance
including PMA and MR ratio.
[0044] According to one preferred process of fabricating the HOCL,
a first metal or alloy is sputter deposited on a top surface of the
free layer 17. Thereafter, a first oxidation step such as a natural
oxidation (NOX) or radical oxidation (ROX) process is performed to
achieve at least partial oxidation in the resulting interface oxide
layer 22. An over oxidized state should be avoided and can be
detected in an experiment where first metal films with a constant
thickness are oxidized for various lengths of time under a NOX
process, for example. Resistivity measurements are able to indicate
a time t1 when an over oxidized state is reached. All oxidation
times less than t1 are said to generate an "under" oxidized or
partial oxidation state for the interface oxide which is desired
for the preferred embodiment. Next, a second metal layer referred
to as a first transition metal layer, or a stack of two different
transition metal layers is deposited on the interface oxide layer
followed by a second oxidation step. The second oxidation step is
relied on to at least partially oxidize the second metal layer or
the stack of two transition metal layers and thereby form
transition metal oxide layer 23, or transition metal layers 23, 24,
respectively. In an alternative embodiment, a first transition
metal layer is formed on the interface oxide layer and is oxidized
with a second oxidation step to form transition metal oxide layer
23. Thereafter, a second transition metal layer is deposited on
oxide layer 23 and is oxidized with a third oxidation step to form
transition metal oxide layer 24.
[0045] A key concept of the present disclosure is to utilize oxygen
from one or more of the transition metal oxide layers formed during
the second oxidation step, and possibly a third oxidation step, to
increase the oxidation state of the under oxidized interface oxide
layer formed during the first oxidation step. In other words, the
HOCL structure defined herein takes advantage of oxygen gettering
by the interface oxide from the one or more transition metal oxide
layers during the second oxidation step and/or during a subsequent
anneal process. In effect, oxygen from at least a second transition
metal oxide layer is purposely leaked into the interface oxide
thereby increasing the oxidation state of the interface oxide layer
and requiring a less strenuous first oxidation step of the first
metal or alloy. As a result, weak oxidation conditions (low oxygen
pressure or flow rate, and/or short oxidation time) during the
first oxidation step will cause less oxidative damage in the
adjoining free layer than in prior art processes which involve
oxidation of a capping layer. Furthermore, the enhanced oxidation
state in the interface oxide layer as a result of the second
oxidation step and anneal will promote higher PMA in the adjoining
free layer.
[0046] As indicated above, higher conductivity (lower RA values)
may be achieved by reducing the oxidation state in one or more of
the HOCL layers 22-24. In an alternative embodiment, a transition
metal oxide with relatively high conductivity is chosen to optimize
performance in STT-MRAM and other spintronic devices that benefit
from low resistance. A lower resistance in one or both of the
aforementioned transition metal oxide layers 23, 24 may be realized
by doping with one or more of Fe, Co, Ni, Ru, Cr, Au, Ag, and
Cu.
[0047] In another embodiment, RA values in one or more of the HOCL
layers may be minimized and conductivity improved by selection of a
different phase transition. For example, in interface layer 22,
lower resistivity is achieved by oxidation of an .alpha.-Ta layer
rather than oxidizing a .beta.-Ta layer in a Mg/Ta stack. The
resulting interface oxide layer is a MgO/TaOx composite. In another
embodiment, .alpha.-Ta is co-sputtered with Mg to give a MgTa alloy
layer with a Mg:Ta ratio of about 1:1 to 2:1 and the alloy is then
oxidized to yield a MgTaOx interface layer.
[0048] All layers in MTJ 1 may be formed in an Anelva C-7100 thin
film sputtering system or the like which typically includes three
physical vapor deposition (PVD) chambers each having five targets,
an oxidation chamber, and a sputter etching chamber. At least one
of the PVD chambers is capable of co-sputtering. Usually, the
sputter deposition process involves an argon sputter gas and the
targets are made of metal or alloys to be deposited on a substrate.
Once all of the layers in the MTJ are laid down on a substrate, a
high temperature anneal may be performed in an oven by applying a
temperature of about 250.degree. C. to 500.degree. C., and
preferably near 400.degree. C. for a period of 10 minutes to 2
hours. Thereafter, an array of MTJ elements with substantially
vertical sidewalls may be fabricated by a process involving a
conventional photolithography patterning and reactive ion etch
(RIE) sequence that is well known in the art and will not be
described herein. Subsequently, an insulation layer 40 may be
deposited to electrically isolate adjacent MTJ elements. Only one
MTJ element is shown to simplify the drawing. A chemical mechanical
polish (CMP) process is typically employed to form a smooth top
surface of the insulation layer which becomes coplanar with a top
surface of the MTJ array. Then an additional metal level that
includes a bit line 30 may be formed on the MTJ elements to
continue the fabrication of a magnetic device.
[0049] Referring to FIG. 2, a second embodiment is shown in which
the MTJ has a bottom spin valve configuration and the reference
layer 12 has a synthetic antiferromagnetic (SAF) structure with an
AP2/coupling/AP1 configuration to improve thermal stability of the
MTJ and also reduce the interlayer coupling Hin (offset) field
applied to the free layer 17. Each of the AP2 layer 13 and AP1
layer 15 may be comprised of one or more of Co Fe, CoFeB, and Co
and has a thickness from about 5 to 20 Angstroms. The coupling
layer 14 is typically made of Ru, Ir, or Rh. The thickness of the
AP2 layer and AP1 layer may be modified to adjust the Ho (offset
field) to approximately 0 Oe. In another embodiment, one or both of
the AP1 and AP2 layers may have an (A1/A2).sub.n configuration with
intrinsic PMA.
[0050] Free layer 17 may have a FL composition as described in the
first embodiment. However, an alternative free layer structure
having a FL1/spacer/FL2 configuration may be advantageously used.
Each of the FL1 layer 18 and FL2 layer 20 may be comprised of one
or more of CoFe, CoFeB, CoFeNiB, CoB, Co, Fe, or FeB. A
non-magnetic spacer 19 made of Ta, Mg, Zr, Hf, Mo, W, Nb, or the
like and preferably having a thickness between 1 and 5 Angstroms is
formed between the FL1 an FL2 layers and serves as a moment
diluting layer. Above the free layer is a HOCL 21 that has a
bilayer configuration with a lower interface oxide layer 22 and an
upper transition metal oxide layer 23 as described previously.
However, a second transition metal oxide layer 24 may be inserted
as the uppermost layer in the HOCL as described in the first
embodiment.
[0051] Referring to FIG. 3, the present disclosure also encompasses
an embodiment wherein a HOCL is employed to enhance PMA in an
adjoining reference layer within a MTJ that has a seed
layer/HOCL/reference layer/tunnel barrier/free layer/HOCL
configuration. As in previous embodiments, a seed layer 11 is
preferably formed on substrate 10. In one aspect, a first HOCL 21a
is formed on the seed layer and comprises a lower transition metal
oxide layer 23a and an upper interface oxide layer 22a. Reference
layer 12 may be a single layer or composite RL as described
previously or may have a SAF configuration as depicted in the
exemplary embodiment. Likewise, free layer 17 may be a FL as
described earlier in the first embodiment or may be a composite
with a FL1/spacer/FL2 configuration. There may be a second HOCL 21b
formed on the free layer with a lower interface oxide layer 22b and
an upper transition metal oxide layer 23b. The composition and
attributes of the interface oxide layers 22a, 22b are equivalent to
that of interface oxide layer 22 in previous embodiments. Moreover,
transition metal oxide layers 23a, 23b may have the same
composition as that of transition metal oxide layer 23 described
previously. The present disclosure also anticipates an embodiment
wherein a second transition metal oxide layer is included in one or
both of the HOCL 21a and HOCL 21b. In HOCL 21a, a second transition
metal oxide layer (not shown) may be inserted between seed layer 11
and transition metal oxide layer 23a. HOCL 21b may have a trilayer
structure equivalent to HOCL 21 in FIG. 1.
[0052] In an alternative embodiment, depicted in FIG. 10, the
second HOCL formed on the free layer in FIG. 3 is omitted and is
replaced by a capping layer 29 to provide a seed
layer/HOCL/reference layer/tunnel barrier/free layer/capping layer
configuration for the MTJ.
[0053] In FIG. 4, a fourth embodiment is depicted in which a
laminated (FL/HOCL).sub.m stack where m.gtoreq.2 is formed on a
tunnel barrier layer in a bottom spin valve configuration. Thus,
all of the layers are retained from the first embodiment except the
FL/HOCL structure is repeated one or more times to form a plurality
of FL and HOCL layers. In the exemplary drawing, m=2 and the
reference layer forms an interface 12s with seed layer 11 and
another interface 12t with tunnel barrier layer 16. PMA is enhanced
in free layer 17a through a first interface 17s1 with the tunnel
barrier and a second interface 17t1 with a first HOCL 21b. Note
that individual layers within HOCL 21b (and HOCL 21c) are not shown
but it is understood that each of the HOCL are comprised of two or
more oxide layers including a lower interface layer and at least
one transition metal oxide layer as the upper layer in each FL/HOCL
stack. A second free layer 17b is formed on the first HOCL and is
ferromagnetically coupled to the first free layer. PMA in free
layer 17b is induced or enhanced through an interface 17s2 with the
first HOCL and by interface 17t2 with an overlying second HOCL 21c.
It should be understood that the reference layer may have a RL
structure as described in the first embodiment or a SAF structure
found in the second embodiment. One or both of the interface oxide
layer and upper transition metal oxide layer in each HOCL may
comprise a certain number of unoxidized metal or alloy atoms to
achieve higher conductivity therein. Furthermore, one or more HOCL
may comprise a dopant that is one or more of Fe, o, Ni, Ru, Cr, Au,
Ag, and Cu to promote higher conductivity.
[0054] According to a fifth embodiment shown in FIG. 5, MTJ 1 may
have a top spin valve structure in which a HOCL, free layer, tunnel
barrier layer, reference layer, and capping layer are sequentially
formed on a buffer (seed) layer. Therefore, the top spin valve
structure retains the same layers from the second embodiment except
the positions of reference layer 12 and free layer 17 are switched,
and HOCL 21 is inserted between seed layer 11 and the free layer.
Furthermore, transition metal oxide layer 23 becomes the bottom
layer in the HOCL stack and the interface oxide layer 22 is the
upper oxide layer in order for the latter to maintain an interface
with the free layer. In this case, a top surface of the interface
oxide adjoins a bottom surface of FL1 layer 18 along interface 17s
while FL2 layer forms a second interface 17t with tunnel barrier
16. In an alternative embodiment, the reference layer may have a RL
structure and/or the free layer may have a FL structure as defined
in the first embodiment. Moreover, the HOCL may be comprised of two
transition metal oxide layers wherein a second transition metal
oxide layer 24 is inserted between transition metal oxide layer 23
and seed layer 11. Capping layer 29 may comprise one or more of Ru,
Ta, or other capping layer materials used in the art.
[0055] To illustrate the benefits of a MTJ with a HOCL formed
according to an embodiment of the present disclosure, a series of
MTJ stacks labeled A-C were prepared with different oxide capping
layers. All MTJ stacks have the same sequence of layers formed
below the oxide capping layer and are unpatterned. In particular,
each MTJ has a NiCr/CoFeB/MgO/CoFeB/oxide capping layer
configuration where NiCr is the seed layer, the reference layer and
free layer each have a Co.sub.20Fe.sub.60B.sub.20 composition and a
12 Angstrom thickness, and the tunnel barrier layer is MgO. In a
first sample corresponding to MTJ stack A, a 6 Angstrom thick
MgO/TaOx capping layer previously employed by the inventors is
used. For MTJ stack B, the oxide capping layer is a MgO/TaOx/RuOx
HOCL of the present disclosure wherein the MgO/TaOx layer is
prepared by depositing a 3 Angstrom thick film of Mg on the CoFeB
free layer followed by depositing a 3 Angstrom thick Ta film on the
Mg layer and then oxidizing with a NOX process comprising a 2
standard cubic centimeters per minute (sccm) to 2 standard liters
per minute (slm) oxygen flow for a period of 5 to 600 seconds. A
MgTaOx interface oxide layer may be used in place of a MgO/TaOx
composite interface oxide layer if a MgTa alloy is deposited and
then oxidized by the aforementioned NOX process. Thereafter, a 10
Angstrom thick Ru film was deposited on the TaOx layer in the
composite interface oxide layer followed by a second NOX process.
The resulting RuOx layer in the HOCL is under oxidized to promote
higher conductivity. A third sample corresponding to MTJ stack C
has a MgO/TaOx/VOx capping layer. The upper oxide layer was formed
by depositing a 10 Angstrom thick V layer on the TaOx layer and
then performing an oxidation similar to the second NOX process
above to give an under oxidized VOx layer. All MTJ stacks were
deposited at room temperature and annealed at 400.degree. C. for 30
minutes.
[0056] Referring to FIGS. 6-8, magnetic properties of the MTJ
stacks A-C, respectively, were measured perpendicular to the film
plane. Sample A in FIG. 6 exhibits substantial PMA character.
However, stack B (FIG. 7) shows improved PMA compared with stack A
as evidenced by a greater horizontal distance between the two
vertical portions of the curves than in FIG. 6. Furthermore, there
is a slight enhancement in the saturation signal (maximum Kerr
signal value) for stack B which indicates the free layer was
damaged less during the oxidation process of the HOCL than during
formation of the MgO/TaOx capping layer in stack A. Stack C (FIG.
8) demonstrates that the selection of a metal for the upper oxide
layer must have a lower absolute value of free energy of oxide
formation than that of the MgO and TaOx interface oxide layers. In
the stack C example, the opposite is true. The free energy of oxide
formation for V.sub.2O.sub.4 is -1329 kjoules/mole (-318 kcal/mole)
listed in Table 1 of "Some Thermodynamic Relations Among the
Vanadium Oxides, and their Relation to the Oxidation State of the
Uranium ores of the Colorado Plateaus", R. Garrels, p. 1251-1265,
U.S. Geological Survey, Washington, D.C. Therefore, the absolute
value of 1329 kj/mole is greater than that of TaOx (788 kj/mole)
and for MgO (1100 kj/mole) in the MgO/TaOx composite interface
oxide. As a result, V is believed to getter oxygen from the
underlying MgO and TaOx layers thereby lowering the interface oxide
oxidation state which in turn reduces PMA formed by the interface
with the CoFeB free layer. Moreover, PMA is suppressed and is
essentially zero for stack C. On the other hand, VOx may be
employed as an interface oxide in a HOCL according to an embodiment
of the present disclosure since vanadium's oxygen gettering
property maintains a substantial oxidation state in a HOCL such as
VOx/TM1Ox or VOxJTM1Ox/TM2Ox where TM1 and TM2 are transition
metals or transition metal alloys as described previously.
[0057] In embodiments wherein a composite interface oxide layer
such as MgO/TaOx is employed in a HOCL, preferably the lower oxide
layer contacting the free layer (or ferromagnetic layer) has a
higher absolute value of free energy of oxide formation than the
upper oxide layer in the interface oxide stack to prevent oxygen
from diffusing out of the lower oxide into the upper oxide layer.
Therefore, MgO/TaOx is highly preferred over a TaOx/MgO interface
oxide configuration.
[0058] In some cases, WOx, MoOx, and SnOx are advantageously used
as the one or more transition metal oxide layers to impart higher
thermal stability to a HOCL.
[0059] All of the embodiments described herein may be incorporated
in a manufacturing scheme with standard tools and processes. In
summary, a substantial gain in thermal stability and PMA is
realized without sacrificing other magnetic properties such as MR
ratio which is an important advantage in enabling 64 Mb and 256 Mb
STT-MRAM technology, and other magnetic devices where low RA value,
high MR ratio, and high thermal stability are critical parameters.
A key feature is the use of an interface oxide/transition metal
oxide capping layer configuration where the transition metal has an
absolute value of free energy of oxide formation less than that of
the first metal or alloy chosen for the interface oxide. As a
result, the first metal or interface oxide with an under oxidized
state getters oxygen from the transition metal oxide during an
anneal process to allow a less strenuous oxidation process of the
first metal. The oxidation sequence of forming a hybrid oxide
capping layer as described herein minimizes oxidative damage in the
adjoining free layer and leads to an increase in PMA and thermal
stability.
[0060] While present disclosure has been particularly shown and
described with reference to, the preferred embodiment thereof, it
will be understood by those skilled in the art that various changes
in form and details may be made without departing from the spirit
and scope of this disclosure.
* * * * *