Académique Documents
Professionnel Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/234881386
CITATIONS READS
101 173
2 authors, including:
M.J. Hurben
Seagate Technology
18 PUBLICATIONS 331 CITATIONS
SEE PROFILE
All content following this page was uploaded by M.J. Hurben on 03 June 2015.
The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
JOURNAL OF APPLIED PHYSICS VOLUME 83, NUMBER 8 15 APRIL 1998
tally for three types of inhomogeneities in bulk YIG, namely, modeled in terms of spherical voids or pores. The calculation
surface pits, volume pores or voids, and randomly distributed results in a relaxation rate expression which reflects the as-
single crystal grains in a polycrystal.4,7 The most detailed sumption of spherically symmetric inhomogeneities. While it
published theory is for scattering due to an isolated spherical is possible to extend the analysis to include nonspherical
pore imbedded in a large sample.3,4 Theoretical and experi- inhomogeneities,17,19,20 no attempt has been made to include
mental results have been reviewed in Ref. 2. such effects here. For a wide distribution of pore or void
Critical evidence for the importance of two magnon scat- shapes, it is reasonable to adopt an isotropic scattering
tering is contained in the data of Buffler10 on linewidth ver- limit in which the angular coupling term which derives from
sus frequency for YIG spheres which were polished to pro- the dipole field distribution around a spherical void is re-
duce pits of different sizes. Various studies of FMR placed by some average value. The spherical void scatter-
linewidths in ferrite materials have also demonstrated that ing and isotropic scattering limits give very different
detailed information on the two magnon scattering contribu- two magnon linewidth versus field angle profiles. These dif-
tion to the loss may be obtained if the number of spin waves ferences will provide a way to separate very different kinds
which are degenerate with the FMR frequency can be ad- of scattering processes.
justed experimentally through the use of different sample Previously two magnon analyses have been limited to
shapes, different field orientations, etc.1114 materials which were either isotropic or with relatively small
levels of magnetocrystalline anisotropy such that the behav-
B. Thin films and degenerate spin waves ior of the spin wave band was essentially the same as for
isotropic materials. Schlomann et al.,21 among others, has
A considerable amount of related work has been done on
shown that anisotropy can have a significant effect on the
thin films. Ferromagnetic resonance measurements in thin
spin wave dispersion and the corresponding spin wave band.
permalloy films,15,16 for example, demonstrated the potential
Explicit examples of such effects and further references may
importance of two magnon scattering processes through the
be found in Refs. 2124.
use of the thin film geometry and field orientation to change
The modifications in the spin wave band due to anisot-
the degeneracy condition. For FMR with an in-plane static
ropy have a large effect on the two magnon linewidth versus
magnetic field and an in-plane magnetized thin film, there
angle profiles. From a technological point of view, these ef-
are a large number of spin wave states which are degenerate
fects are extremely important. This is due, in part, to the
with the uniform mode and a significant contribution to the
continuing interest in hexagonal ferrite materials for
two magnon scattering linewidth is possible. If the static
millimeter-wave device applications,25 and the recent devel-
magnetic field is perpendicular to the film and sufficient to
opment of pulse laser deposited ~PLD! single crystal barium
saturate the magnetization perpendicular to the film, there are
essentially no spin wave states degenerate with the FMR ferrite ~BaM! films.26 These new BaM PLD films, as well
frequency and there should be almost no two magnon scat- as other PLD ferrite films,27,28 have rather large FMR line-
tering contribution to the linewidth. For an obliquely magne- widths. As these materials are developed and refined for
tized film, the FMR frequency lies between these two limits. millimeter-wave device applications, it will be important to
The number of degenerate spin waves to which the uniform consider the possibility of a two magnon scattering contribu-
mode can couple via the two magnon process, and hence the tion to the high frequency losses, the types of microstructure
relaxation rate and the linewidth, are strongly dependent on responsible for these losses, and ways to modify the film
the external field orientation. morphology to eliminate these effects and reduce the losses.
The angle dependence of the two magnon scattering con- The objective of this work, then, has been: ~i! to develop
tribution to the linewidth for films has been treated previ- a clear, practical, and operational formalism for the two mag-
ously by Sparks.17 This treatment was done for two limited non scattering relaxation rate in anisotropic ferrite films; ~ii!
cases, and only for isotropic materials. The first part of to make explicit the role of different types and sizes of in-
Sparks treatment developed an approximate expression for homogeneities on the two magnon losses; ~iii! to make ex-
the linewidth due to scattering from such spherical voids plicit the role of anisotropy modifications to the spin wave
within the film, but did not explicitly consider the role of the band on these processes; and ~iv! to provide example calcu-
void size. The second part considered scattering from large lations of profiles of linewidth versus external field angle
etch pits which extended through the film thickness. This which demonstrate these effects.
work shows the wealth of two magnon effects which are Section II defines basic parameters and establishes oper-
important for thin films. ating equations for static equilibrium and the uniform mode
FMR response when the static magnetic field is applied at
some angle u relative to the film normal. Section II also
C. Objective of this work
establishes operational equations for the FMR frequency,
The objective of this work is to provide a specific, ana- various effective field parameters, and a phenomenological
lytical, and operational theoretical formalism for two mag- relaxation rate which will be used to characterize intrinsic
non scattering relaxation and the FMR linewidth in thin losses. Section III provides a brief review of spin wave prop-
films. A summary of the theory is given in Ref. 18. erties for isotropic and uniaxially anisotropic materials. In
The analysis is based on the transition probability calcu- Sec. IV, the relaxation rate or inverse relaxation time for
lation of Sparks, Loudon, and Kittel4 ~SLK! for isotropic scattering between the uniform mode and the degenerate spin
spherical samples. In this approach, the microstructure is waves is developed along the same lines as used in the origi-
4346 J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton
S D
absorbed and a corresponding frequency linewidth Dv. This In the X-Y -Z frame, the tensor A is given by
linewidth is the interval, in frequency units, between the 24 p N XY 0 0
half-power points of the microwave absorption profile.
In the second approach, and the one which is favored 0 24 p N XY 0
A5 . ~5!
experimentally, the microwave frequency is fixed at v while HA
the strength of the external field Hext is varied. The two basic
0 0 24 p N Z 1
Ms
parameters of interest are then the FMR field H FMR and field
linewidth DH. The field H FMR is defined to be the value of For purposes of the usual uniform mode analysis, the
H ext at which the maximum microwave power is absorbed. components of the dynamic magnetization are assumed to
The linewidth DH is the interval, in field units, between the have an e i v t dependence. The analysis leads to an expression
half power points. for the FMR frequency v FMR which is valid to first order in
In order to determine the FMR parameters v FMR , the loss term 1/T
H FMR , Dv, and DH, a modified version of the torque equa- v FMR5 g ~ H x H y ! 1/2, ~6!
tion is often used. The modification involves the addition of
phenomenological damping terms which account for the re- H x 5H extcos~ u 2 f ! 1 21 @ 2H A 24 p M s ~ 3N Z 21 !#
laxation of the magnetization. The analysis leads to relatively
3cos2 f , ~7!
simple working equations for the FMR frequency v FMR and
linewidth Dv. Expressions for the corresponding field pa- H y 5H ext cos~ u 2 f ! 1 21 @ 2H A 24 p M s ~ 3N Z 21 !#
rameters H FMR and DH are much more involved. Determi-
nation of these parameters is discussed shortly. 3cos~ 2 f ! . ~8!
The two most commonly used phenomenological modi- The H x and H y parameters may be viewed as effective stiff-
fications to the torque equation which account for loss are ness fields which characterize the instantaneous torque ex-
the BlochBloembergen ~BB!30 and LandauLifshitz ~LL!32 erted on M(t) when it is tipped parallel toward the h(t)
approaches. In this work, the BB formalism will be used direction or perpendicular to the h(t) direction, respectively.
exclusively, because it is physically consistent with the two Although the above results give v FMR as a function of
magnon process. A thorough discussion of the various relax- H ext , the explicit dependence of the value of the external
ation mechanisms and the phenomenological damping for- magnetic field, H FMR , on the field angle u for resonance at a
malisms is given by Sparks.3 Various phenomenological given frequency v, can also be determined. Once the sample
damping approaches to ferromagnetic resonance are also dis- parameters g, 4 p M s , H A , and N Z are specified, Eqs. ~1!,
cussed by Lax and Button33 and Patton.2 ~6!, ~7!, and ~8!, with H ext and v FMR replaced by H FMR and
The modified torque equation of motion for the trans- v, respectively, can be solved for H FMR as a function of u.
verse dynamic magnetization m(t) with BB loss included
may be written as C. FMR field versus angle at fixed frequency
Figure 2 shows the calculated FMR field H FMR as a
dm~ t ! m~ t ! function of the external field angle u for the three represen-
52 g M~ t ! 3H~ t ! 2 . ~3!
dt 2T tative materials described above and in the infinite thin film
limit where N Z 51 is satisfied. Figure 2~a! is for YIG at
Here H(t) represents the total effective magnetic field in the v /2p 510 GHz, Figure 2~b! is for BaM at 50 GHz, and
sample, g is the absolute value of the electron gyromagnetic Figure 2~c! is for ZnY at 10 GHz. The frequencies were
ratio, and 1/T is the transverse and longitudinal BB relax- chosen to be representative of typical operating frequencies
ation rates. A typical value for g is 1.763107 rad/s Oe or 2.8 for these materials. These same materialfrequency combi-
GHz/kOe in practical units. For a thin disk or film magne- nations will be used throughout this work for purposes of
tized out-of-plane, a 1/T value of 1.763107 rad/s would cor- numerical evaluations.
respond to a field swept linewidth of 1 Oe. Note that 1/T is in Consider the YIG results in ~a! first. For an external field
units of rad/s. The calculations presented below will yield applied perpendicular to the film plane at u 5 f 50, the
expressions for 1/T. When numerical results are presented, FMR field is a maximum, while the minimum occurs for the
these will be given in terms of 1/g T, in Oe, corresponding to in-plane configuration at u 5 f 590. The variation in H FMR
field linewidths. The BB formulation also includes a longi- with the field angle is due to the change in the sample de-
tudinal relaxation term for the z component of the total dy- magnetization and stiffness fields. For the BaM case in ~b!,
namic magnetization. This term is not considered in a linear the situation is reversed because of the very strong easy-axis
theory. anisotropy perpendicular to the film plane. Here, the FMR
The net magnetic field H(t) consists of the applied fields field is a minimum for u 5 f 50, which corresponds to the
Hext and h(t) along with the static and dynamic fields which easy direction, and a maximum when the field is in-plane,
result from sample demagnetization and non-Maxwellian ef- along a hard direction. For the ZnY result in ~c!, the anisot-
fective fields related to magnetocrystalline anisotropy. The ropy acts together with the demagnetization fields to pull the
net effective field H(t) can be expressed in terms of a modi- magnetization vector in-plane. For the out-of-plane orienta-
fied demagnetizing tensor A according to tion, the field required for ferromagnetic resonance is very
4348 J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton
FIG. 2. Calculated static external field for ferromagnetic resonance, H FMR , as a function of the external field angle u for the three representative materials in
the infinite thin film limit. ~a! is for YIG at 10 GHz. ~b! is for BaM at 50 GHz. ~c! is for Zn-Y at 10 GHz.
large, but this field then rapidly decreases as the field is It is important from the outset, therefore, to separate and
tipped away. The rapid change in H FMR reflects the severe understand the geometric linewidth angle dependences
misalignment between the static field and magnetization vec- which result in field swept experiments.
tors. For field angles above about 30 or so, the static mag- For a given 1/T value and a particular field angle u, the
netization is essentially in-plane. field linewidth DH can be determined directly from the ex-
The results in Fig. 2 demonstrate an important effect of plicit microwave susceptibility expressions obtained from
the external field orientation on the FMR response for a con- the uniform mode analysis. Although it is not possible to
stant microwave frequency, namely, that the strength of the obtain simple closed form expressions for DH as a function
applied static field and hence the net field within the material of the field angle u, it is possible to evaluate the susceptibil-
vary strongly with field angle. This effect will play a crucial ity response numerically and determine DH as the separation
role in the two magnon scattering and linewidth versus angle in field values which correspond to the half-power points of
theory developed shortly. the absorption. A simpler approach, and one which can be
used to demonstrate the intuitive connections between DH
and D( v / g ), is to invoke a simple connection between the
D. Linewidth considerations and linewidth versus two linewidths which is strictly valid only in the limit of
angle small linewidths. This connection may be written as
The FMR analysis also yields an expression for the BB ] H FMR
frequency swept linewidth Dv. This linewidth is conve- DH' g D~ v/g !. ~10!
]v FMR
niently expressed in magnetic field units as D( v / g ). The BB
linewidth in field units is The derivative can be determined from Eqs. ~1!, ~6!, ~7!, and
1 ~8!. In the limit that D v ! v FMR is satisfied, the linewidths
D~ v/g !5 . ~9! predicted by the two approaches agree to within several Oe.
gT
As will be discussed below, the linewidth conversion factor
In contrast with the FMR frequency result of Eq. ~6!, for g ] H FMR / ]v FMR is equal to one for u 50, greater than one
which v FMR varied as the geometric mean of two stiffness when magnetization rotation effects dominate the field swept
fields, the BB linewidth D( v / g ) is constant and proportional line broadening at angles between u 50 and u 590, and is
to the transverse relaxation rate. less than one for parallel resonance at u 590.
Unlike the simple D( v / g )51/g T result of Eq. ~9!, a The effect of the field orientation on the FMR linewidth
corresponding expression for the field swept linewidth DH in for a constant relaxation rate is illustrated in Fig. 3. The
the field swept FMR experiment at constant frequency is figure follows the form of Fig. 2, with ~a! for YIG, ~b! for
much more complicated. This can be understood from an BaM, and ~c! for ZnY. The calculations are based on the
examination of the field and magnetization vectors during infinite thin film limit and the same material parameters and
the course of an FMR experiment. For the frequency swept frequencies used in Fig. 2, along with 1/g T values of 0.5, 20,
case, the orientation and strength of Hext are held fixed, so and 15 Oe, for the YIG, BaM, and ZnY cases, respec-
that the equilibrium direction of static magnetization Ms also tively. These values correspond to reasonable out-of-plane
remains constant. For the field swept case, the direction of resonance linewidths due to intrinsic losses in these materials
Hext is fixed while its strength is varied. As can be seen from at the chosen frequencies.
Eq. ~1!, the orientation angle f for Ms will then vary as H ext Consider the results for the YIG film in Fig. 3~a!. Here,
is swept, unless u is either 0 or 90. As a result, DH can be the field swept linewidth DH is equal to 1/g T for the out-of-
a complicated function of u even for constant 1/T. It will be plane orientation and shows only a small increase at inter-
seen from the two magnon scattering theory that the relax- mediate angles. There is a small but indiscernible drop in
ation rate 1/T itself shows strong angle dependences as well. DH from 1/g T at u 590. In the case of YIG at 10 GHz, the
J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton 4349
FIG. 3. Calculated ferromagnetic resonance field swept linewidth DH as a function of the external field angle u for the three representative materials in the
infinite thin film limit and for constant values of the BB relaxation rate 1/T. ~a! is for YIG at 10 GHz with 1/g T50.5 Oe. ~b! is for BaM at 50 GHz with
1/g T520 Oe. ~c! is for ZnY at 10 GHz with 1/g T515 Oe.
conversion factor between the field linewidth and the relax- linewidth with the field angle roughly tracks the change in
ation rate given by Eq. ~10! is very close to one and essen- the slope of the H FMR vs u curve in each scenario.
tially independent of angle. The small peak around u 535 is
due to the small difference between the field angle u and the III. SPIN WAVE DISPERSION AND DEGENERATE SPIN
magnetization angle f as the field is rotated at intermediate WAVES
angles. For in-plane ferromagnetic resonance at u 5 f This section presents a brief review of spin wave con-
590, the YIG linewidth for BB damping is slightly smaller siderations which will be important for the two magnon pro-
than the perpendicular linewidth at 0.47 Oe. This relative cess. Dispersion relations of spin wave frequency v k versus
insensitivity of the YIG field swept linewidth to rotation is wave vector k are obtained. This section also provides an
due to the near alignment of Ms with Hext as the field is explicit examination of the change and frequency shift in the
rotated and H ext is maintained at the value needed for ferro- spin wave band for YIG, BaM, and ZnY films as one
magnetic resonance at 10 GHz. moves from the perpendicular FMR configuration with the
The situation is quite different for the BaM linewidth field at u 50 to the parallel FMR case with the field at u
results in ~b! of Fig. 3. Here, the field swept linewidth shows 590. These considerations lead to an intuitive understand-
ing of the change in the degenerate magnon situation and
a very significant variation with angle. Note that the large
hence, the states available for two magnon scattering, as one
uniaxial anisotropy for the BaM film, with the easy axis
moves from the perpendicular to the parallel FMR configu-
perpendicular to the film, causes the FMR field to be small at
rations. The presentation is limited to so-called bulk spin
u 5 f 50 and large at u 5 f 590. For intermediate u val- waves for which the wave number k is much greater than
ues, the Ms vector tends to be somewhat closer to the film 2 p /S, where S is the film thickness. The magnetostatic
normal than does Hext and the magnetization angle f lags mode limit in which k may be on the order of or smaller than
behind the field angle u. For ferromagnetic resonance at 50 2 p /S, but still above the pure electromagnetic wave limit
GHz and the combination of BaM parameters used here, v k /c, where c is the speed of light, will not be considered
the field swept linewidth for a constant 1/g T values of 20 Oe here. Possible effects due to magnetostatic modes and film
increases to a maximum value of about 33 Oe near an exter- thickness effects other than demagnetizing factors will be
nal field angle of 60 and then drops to a value slightly briefly considered at the end of the article.
below the perpendicular field linewidth at u 5 f 590. A. Classical bulk spin waves
As shown in ~c!, this effect is even more pronounced for
ZnY film parameters and 10 GHz ferromagnetic resonance. For a bulk material, the sample boundaries can be taken
In this case, the anisotropy is in-plane and serves to enhance to lie at infinity and do not influence the dispersive properties
of the spin wave modes. For thin film materials, however, the
the demagnetizing effects. Here, DH increases very rapidly
boundary conditions of the magnetic field vectors can sig-
as the external FMR field is rotated away from the perpen-
nificantly modify these dispersion relations. Calculation of
dicular FMR orientation at u 50. The peak in DH occurs at the so-called dipole-exchange modes, which are the normal
u '8 and is more than a factor of 5 greater than the line- modes for thin film materials, is in general extremely com-
width at u 50. It is important to keep in mind that the 1/g T plicated and is beyond the scope of this work. References for
values in each of these calculations were taken to be con- dipole-exchange modes include Wolfram and De Wames,34
stant, independent of u. De Wames and Wolfram,35 and Kalinikos et al.36
It is instructive to compare the angle dependences of the In this work, the approach of Sparks18 is followed. The
linewidths in Fig. 3 with the angle dependences of the FMR bulk spin wave dispersion relations are used to approximate
fields shown in Fig. 2. It can be seen that the variation in the the normal modes of the film. This results in a considerable
4350 J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton
FIG. 4. ~a! Relative orientations of the sample (X-Y -Z) and precessional
m~ r,t ! 5mk e 2i ~ kr2 v k t ! . ~12!
(x-y-z) frames. The z axis is taken to be parallel to the static magnetization
The two components of the vector mk are the complex x and
vector Ms and the x axis is taken to be parallel to the X axis. ~b! Orientation
of the propagation wave vector k relative to the x, y, and z axes in the y amplitudes of the spin wave and r denotes a general space
precessional frame. The polar angle u k is measured relative to the z axis. coordinate. The magnetic field H(r,t) consists of a number
The azimuthal angle f k is measured relative to the x axis. of terms which may be written as
H~ r,t ! 5H i z1han~ r,t ! 1hdip~ r,t ! 1hexch~ r,t ! . ~13!
simplification of the two magnon scattering analysis. As long Here, H i is an effective internal static magnetic field and the
as the microstructure responsible for the two magnon inter- three additional fields denoted by lower case h are dynamic
action is much smaller than the film thickness, the bulk spin fields associated with the spin wave dynamic magnetization
wave dispersion relations provide a reasonable model for of Eq. ~12!. The first dynamic field, han(r,t), is associated
two magnon calculations. with the crystalline anisotropy, the field hdip(r,t) is the spin
In order to determine the bulk spin wave dispersion re- wave dipole field, and hexch(r,t) is an effective exchange
lations, Eq. ~3! can be re-written in the lossless form, with field for the spin wave.
the magnetization and field vectors taken to be functions of The various terms in the effective field expression of Eq.
space as well as time ~13! have been developed in the literature in various formats.
dm~ r,t ! The basic approach is given in Ref. 23, among others. It is
5 g M~ r,t ! 3H~ r,t ! . ~11! assumed that the static equilibrium condition of Eq. ~1! is
dt
satisfied. This means that the static effective field is parallel
Analogous to Eq. ~2!, the total magnetization is now of the to the static magnetization vector, and hence, along the z
form M(r,t)'Ms 1m(r,t). The spatial dependences in the axis. The static field term H i may be written as
H(r,t) enter through the spatial dependence of m(r,t) and
the associated dipoledipole interactions, exchange interac- H i 5H ext cos~ u 2 f ! 22 p M s @ sin2 f 1N Z ~ 3 cos2 f 21 !#
tions, and anisotropy considerations. These effects are con- 1H A cos2 f . ~14!
sidered shortly.
The inclusion of spin waves in the model requires an The first term on the right side of Eq. ~14! is simply the
extension of the basic X-Y -Z coordinate system defined in component of the static external field along the direction of
Fig. 1. The two important extensions are depicted by the the static magnetization vector. The second term is the con-
diagrams in Fig. 4. First, it is useful to introduce an addi- tribution from the static demagnetization field. The third
tional reference frame as defined in Fig. 4~a!. Here, in addi- term in Eq. ~14! is the contribution to the static effective field
tion to the sample or laboratory X-Y -Z frame, one considers due to the uniaxial anisotropy. It is important to emphasize
an x-y-z frame of reference in which the static component of that this term does not represent a real magnetic field in the
the magnetization Ms is directed along the z axis. For the Maxwellian sense. It is an effective field based on the spe-
present purposes, this z axis is taken to lie in the Y -Z plane cific form taken for the anisotropy energy. If a different form
of the sample reference frame and is at an angle f relative to for E K ( f ) was used, E K 51K U sin2 f, for example, this last
the Z axis. Recall that f is the angle between the Ms and the term in Eq. ~14! would change, as would Eq. ~15! for the
film normal. With Ms along the z axis, the precession of the dynamic anisotropy field han .
full magnetization vector M yields a dynamic magnetization The next term in Eq. ~13! is the effective dynamic an-
m which, to first order, only has x and y components. Hence, isotropy field han(r,t). This term is needed to account for the
the x-y-z frame may be termed the precessional frame. dynamic effects of the uniaxial anisotropy associated with
This frame, as depicted in Fig. 4~a!, has the x axis along X the dynamic magnetization. For anisotropy as defined above,
and the y axis in the Y -Z plane and rotated away from the Y the dynamic effective anisotropy field is given by
axis by the angle f. Note that for a static magnetization
H A mk y
angle f 50, the x-y-z frame is equivalent to the X-Y -Z han~ r,t ! 5 ~ y sin2 f 2z sin f cos f !
frame. For other orientations, the x-y-z frame is obtained by MS
a clockwise rotation of the X-Y -Z frame about the X axis by 3e 2i ~ kr2 v k t ! . ~15!
an angle f.
One may now establish parameters to define the spin This effective anisotropy field is specified in the x-y-z
wave wave vector k. A suitable definition for these param- frame. Procedures to obtain such effective fields are dis-
eters may be based on ~b! of Fig. 4. The wave vector is cussed in Ref. 23.
J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton 4351
The next field, hdip(r,t), accounts for the dipoledipole For the case of a uniaxial material with the external field in
interaction due to the misalignment of nearby spins associ- the c plane, such that f 590 is satisfied, Eq. ~18! reduces to
ated with the plane wave excitation of Eq. ~12!. For the case the hexagonal ferrite dispersion relation of Joseph, Schlo-
of the uniform precession, where k50, there is no such di- mann, and Bady.21
pole field because all of the spins are parallel. The dipole Two other special cases of the spin wave dispersion re-
field can be conveniently expressed in terms of the wave lation will be useful for the two magnon calculations. The
vector and the dynamic magnetization for the spin wave ac- first is the dispersion relation corresponding to propagation
cording to along the internal field and static magnetization directions,
such that u k 50 is satisfied. This represents the lowest pos-
4p
hdip~ r,t ! 52 k@ kmk # e 2i ~ kr2 v k t ! . ~16! sible spin wave frequency. This frequency is denoted as v min
k2 and is given by
The dipole field given in Eq. ~16! is valid in the so-called v 2min5 g 2 ~ H i 1Dk 2 !~ H i 1Dk 2 2H A sin2 f ! . ~20!
magnetostatic approximation in which the wave number k
for a given frequency is much greater than the corresponding The second special dispersion relation is for propagation per-
pure electromagnetic wave wave number. These consider- pendicular to the static magnetization direction at u k 590.
ations are discussed in Ref. 37. This frequency, denoted as v max , corresponds to the top of
The last field term in Eq. ~13! is the effective exchange the spin wave band
field hexch(r,t). This field, like the static and dynamic anisot- v 2max5 g 2 ~ H i 1Dk 2 !~ H i 1Dk 2 2H A sin2 f !
ropy fields, is not a true Maxwellian field, but is needed in
order to account for the quantum mechanical exchange inter- 1 g 2 4 p M s ~ H i 1Dk 2 2H A sin2 f cos2 f k ! .
action. For a ferromagnetic material, this field can be written ~21!
as
Note that v max depends on the azimuthal spin wave propa-
D 2 gation angle f k , while v min does not depend on this angle.
hexch~ r,t ! 52 k mk e 2i ~ kr2 v k t ! . ~17! It is clear from Eqs. ~14! and ~18! that the spin wave
Ms
dispersion will depend strongly on the orientation of the
The form of this exchange field is discussed in Ref. 3, among static field and magnetization vectors. As shown in Fig. 2,
others. The D parameter characterizes the strength of the this field value can vary over a wide range as the orientation
exchange. For ferrite materials, D has a nominal value in the angle u of the external field is varied over the entire range of
531029 Oe cm2/rad2 range. This value will be used for the values from 0 to 90. In the case of an isotropic material,
explicit numerical results presented below. It is assumed that the effect of this field variation is to shift the spin wave
D is independent of k. Note that for the hexagonal ferrite dispersion frequency band, relative to the FMR frequency, as
materials, D may actually show a strong dependence on u is changed. For anisotropic materials, the effect is much
propagation direction.11,21 Such effects have not been in- more involved. The spin wave dispersion curves are not only
cluded in the present analysis. shifted, but change in shape as well.
In order to obtain dispersion relations of frequency v k as
a function of k, the torque equation of Eq. ~11!, together with
the above effective field equations, are first linearized in the B. Spin wave dispersion and two magnon scattering
x and y components of the dynamic magnetization mk ,
taken as m kx and m ky , respectively. One then obtains a set of It is instructive to consider representative spin wave dis-
two homogeneous equations in m kx and m ky . The secular persion band diagrams for the three material cases introduced
determinant from these equations leads to an expression for above. Such diagrams for YIG, BaM, and ZnY materials
the spin wave frequency v k as a function of the wave vector are shown in Figs. 57, respectively. In each case, band
k. The general spin wave dispersion relation for a uniaxial or diagrams are shown for a range of external field orientations
planar anisotropy material is obtained as from out-of-plane at u 50 to in-plane at u 590. For each
diagram in a given figure, the value of the internal field H i
v 2k 5 g 2 ~ H i 1Dk 2 !~ H i 1Dk 2 14 p M s sin2 u k 2H A sin2 f ! was adjusted to produce the FMR peak response at a particu-
lar fixed FMR frequency and a range of illustrative external
2 g 2 4 p M S H A sin2 f sin2 u k cos2 f k . ~18! field orientations. Figure 5 for YIG shows the band diagram
The validity of this dispersion relation is contingent upon the in the usual two dimensional format of spin wave frequency
satisfaction of the stability condition of Eq. ~1!, as well as the v k versus wave number k. Figures 6 and 7 for BaM and
other conditions specified above. ZnY show both two dimensional and three dimensional
Equation ~18! gives the general form of the bulk spin plots.
wave dispersion relation for a uniaxially anisotropic material Consider the diagrams in Fig. 5 for an isotropic thin YIG
magnetized at any angle f relative to the crystalline c axis in film and an FMR frequency of 10 GHz. The values of H ext
the magnetostatic approximation. If the anisotropy field H A needed to obtain the diagrams in Fig. 5 are the same as the
is set to zero, this equation reduces to the well known dis- H FMR values in Fig. 2 ~a!. Figure 2~a! is for the field applied
persion relation for an isotropic ferrite material out of the plane at u 50, 2~b! is for u 545, and 2~c! is for
u 590. In each case, two specific dispersion branches of
v 2k 5 g 2 ~ H i 1Dk 2 !~ H i 1Dk 2 14 p M s sin2 u k ! . ~19! spin wave frequency v k versus wave number k are shown.
4352 J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton
FIG. 5. Diagrams showing bulk spin wave dispersion curves of spin wave
frequency v k vs wave number k for a thin YIG film and three orientations of
the external static field, with the sample biased at the static field needed for
FMR at 10 GHz. The lower curve in each diagram designates spin waves
propagating parallel to the static internal field direction and the upper curve
is for spin waves propagating perpendicular to the internal field direction.
The region between the two curves represents the range of spin waves
between these limiting case propagation directions. For ~a! the static field is
applied perpendicular to the film plane at u 50. For ~b! the static field is
applied at u 545. For ~c! the static field is applied in the film plane at u
590. The FMR frequency is indicated by the large solid circles. In ~b! and
FIG. 7. Bulk spin wave dispersion diagrams for a thin ZnY film and three
~c!, the narrow band of spin wave modes, which is degenerate with the FMR
orientations of the external static field, with the sample biased at the static
frequency, is indicated by the shaded region.
field needed for FMR at 10 GHz. The top part of each set of diagrams shows
calculated dispersion curves in the usual spin wave frequency v k vs wave
number k format used in Fig. 5. For the upper diagrams, the top curves in ~a!
and the top pair of curves in ~b! and ~c! correspond to a polar spin wave
angle u k of 90, and the bottom curves in ~a!~c! correspond to u k 50. The
horizontal dashed line in ~a! and the hybrid dashed-hatched line in ~b!~c!
indicate the FMR frequency. The bottom part of each set shows schematic
illustrations of the spin wave band in three dimensions with the azimuthal
spin wave propagation angle out of the page. Note that the orientation of the
f k axis is reversed from the f k axis in Fig. 6. For these lower diagrams, the
top shaded surface corresponds to u k 590 and the bottom shaded surface is
for u k 50. The middle horizontal plane which crosses all the diagrams
indicates the FMR frequency. For diagram set ~a!, the static field is applied
perpendicular to the film plane at u 50. For set ~b!, the static field is
applied at u 57.5. For set ~c!, the static field is applied at u 590.
states range from spin waves at k50 for u k 590 to rather surface is warped. The warp is most pronounced at u 590
large k values at u k 50. Note that the maximum value of k and u k 590.
for degenerate states is in the range 3 43105 rad/cm, or Turn now to the FMR response. The dashed line in the
wavelengths in the 0.2 mm range. upper diagrams and the horizontal plane in the lower dia-
The shift in the frequency of the spin wave band relative grams, both labeled as FMR, indicate the FMR frequency of
to the YIG film FMR frequency as H ext is rotated from u 50 GHz which is common to all the diagrams. This FMR
50 to u 590 is due to two effects. First, it is evident from frequency is the same as for Fig. 2~b!. The shaded strip in the
Fig. 2~a! that the external field required for ferromagnetic upper diagrams and the shaded part of the FMR plane in the
resonance at 10 GHz, H FMR , is much larger at u 50 than at lower diagrams indicate the range of spin wave modes with
u 590. From the results in Fig. 2~a!, H FMR drops from over v k the same as the FMR frequency.
5 kOe for u 50 to less than 1 kOe for u 590. The corre- The basic effect conveyed by the diagrams in Fig. 6 for
sponding internal field H i also drops as the static field is the c plane BaM film is the same as for the isotropic YIG
rotated from perpendicular to parallel. Second, the changing film. As one moves from the perpendicular FMR configura-
role of the static and dynamic demagnetizing fields works to tion at u50 to the parallel FMR case at u590, with the
position the thin film FMR frequency precisely at the bottom external field adjusted to keep the FMR frequency constant,
of the k50 spin wave band limit at u 50 and precisely at the spin wave band of frequencies shifts relative to the FMR
the top of the k50 band limit for u 590. frequency and there is a substantial change in the spin wave
The net effect of these FMR and spin wave band shifts is states which are degenerate at this frequency. The situation at
a variation in the availability of spin wave states at the FMR u50 shown in the left-most diagrams in Fig. 6, is the same
frequency with angle. Since these are the spin wave states as for the YIG case: no degenerate spin wave states except at
which can contribute to two magnon scattering relaxation, k50. Now, as u is increased from zero, the entire band
one is able to vary this contribution to the overall relaxation moves down in frequency and various spin wave states move
simply by rotating the direction of the static field used to into the FMR frequency strip in the upper diagrams or FMR
produce the FMR response. If two magnon processes are frequency cut in the lower diagrams of Fig. 6.
important, the profile of FMR linewidth DH versus the field The situation here, however, is more complicated than
angle u at fixed frequency will be different from the constant for YIG. In order to understand the evolution of the degen-
erate spin wave states with angle u, it is necessary to follow
1/T profiles shown in Fig. 3. These differences can provide a
in detail the change in the shaded part of the FMR frequency
signature profile for the two magnon scattering process.
cut in the lower diagrams in Fig. 6. Consider the lower part
Turn now to the dispersion relations for anisotropic
of ~b!, for example. As long as u is not increased too much,
BaM and ZnY films. As made explicit in Eqs. ~18!, ~20!,
the FMR frequency cut extends across the spin wave band in
and ~21!, the spin wave frequencies in these materials de-
such a way that there are degenerate states for all values of
pend, in general, on the azimuthal spin wave propagation
all f k from 0 to 90 and k values from zero out to some
angle f k as well as the polar angle u k and the wave number
mzximum cut-off defined by the u k 50 edge of the band.
k. It is now necessary to include f k as a parameter when
This range of degenerate spin wave states is indicated by the
dispersion curves are computed. Examples of the anisotropic
shaded part of the FMR frequency cut in the lower part of
spin wave band for the case of a BaM film and an FMR
Fig. 6~b! for u560 and the shaded strip in the upper part.
frequency of 50 GHz are shown in Fig. 6. The upper dia- The situation changes as one moves to u575 and Fig. 6~c!.
grams show calculated curves in the two dimensional v k vs Now, as shown by the shaded part of the FMR frequency cut
k format of Fig. 5. The lower diagrams show three dimen- in the lower diagram of the set, the available degenerate spin
sional plots of v k as a function of k and f k . Figure 5~a! is waves become more limited. Here, for small values of k, the
for u 50. Figures 5~b! and 5~c! are for u 560 and u allowed f k values for degenerate spin waves range from
575, respectively. Figure 5~d! is for u 590. Note the f k some lower f k limit up to f k 590. This lower f k limit
axis scale in the lower set of diagrams, which was chosen to increases and approaches 90 as one moves up to an external
provide the best perspective view of the spin wave band. field angle u590 and a parallel FMR configuration. This
The curves in the upper diagrams and the two shaded degenerate spin wave situation is indicated by the very small
curved surfaces in the lower diagrams correspond to different shaded section of the FMR frequency cut for the lower part
( u k , f k ) combinations. The bottom curved surfaces in the in Fig. 6~d!. The details of these changes in the available
lower diagrams are for u k 50 and the upper curved surfaces degenerate spin wave states as u is varied will have impor-
are for u k 590. Corresponding projections of the edge tant consequences for two magnon scattering.
curves for f k 50 and f k 590 from the bottom diagrams Figure 7 shows the evolution in the spin wave band with
yield the curves shown in the upper diagrams. The bottom the external field angle u for ZnY and an FMR frequency
curves in the upper diagrams, for example, are all for u k of 10 GHz. The format is the same as for the BaM diagram
590 and do not depend on f k . The other projections are in Fig. 6 and will not be described in detail. There are several
clear from the diagrams. The main effect of the anisotropy is important differences between the ZnY and BaM dia-
to produce spin wave frequency surfaces as a function of k grams. The ZnY film represents an easy-plane rather than
and f k for fixed values of u k . When the applied field is such an easy-axis situation. The first effect of this change in an-
that the rotational symmetry is broken, e.g., when u is any isotropy is that the warped spin wave band surfaces are re-
angle except zero and u k is also any angle except zero, the u k versed. Although the warped surfaces look the same in Fig. 7
4354 J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton
as in Fig. 6, note that the f k axes are reversed. The second through the addition of an ad hoc constant term to the k-
effect is in the movement in the band relative to the FMR dependent scattering factor in the coupling function.19 Fur-
frequency as the field angle u is changed. For the perpen- ther comments on this modification will be given below. The
dicular configuration and u 50, the FMR frequency point is SLK calculation has also been applied to the case of an ob-
located at the very bottom of the band, as before. For the liquely magnetized thin film,18 but the analysis was restricted
parallel configuration and u 590, however, the FMR fre- to isotropic materials and spherical void scattering.
quency point is at the bottom of the warped u k 590 spin It is important to establish a clear connection between
wave surface at k50, rather than at the top of this surface. the relaxation rate obtained from the SLK approach and the
There are never any excluded f k modes from the two mag- linewidths which are obtained in the experiment. For a given
non scattering process. As the field angle u is rotated from 0 two magnon relaxation rate 1/T TM , the FMR frequency line-
to 90, the spin wave band moves only part way down and width is simply D( v / g )51/g T TM . Once the frequency line-
there is no drastic change in the degenerate state situation as width is determined, the field linewidth DH for two magnon
for BaM. One would expect that the two magnon linewidth scattering can be found from Eq. ~10!. It should be noted that
profile for ZnY would be much simpler than for either many treatments specify the relaxation time t m associated
BaM or for YIG films. with the decay of the dynamic magnetization amplitude
rather than the decay of the energy associated with the dy-
IV. TWO MAGNON SCATTERING THEORY namic magnetization. In that case, the frequency linewidth is
The two magnon relaxation rate corresponds to the rate given by D( v / g )52/gt m . 38 The presentation here will use
at which energy is transferred from the uniform precession to energy decay as the basic decay rate.
degenerate spin waves due to the presence of magnetic inho- Several modifications have been made to the SLK ap-
mogeneities. It is assumed that these excited spin waves then proach. First, in order to deal with anisotropic films, the SLK
relax to the lattice extremely rapidly and do not affect the treatment has been extended to account for the effects of
two magnon decay. The basic formalism has been taken over anisotropy on both the FMR response and the spin wave
from the analysis developed by SLK.4 This section is divided dispersion properties. Second, the scattering calculation has
been modified to include scattering into a band of nearly
into three parts. The first part takes the SLK scattering relax-
degenerate spin waves with a frequency width which corre-
ation rate as a starting point, and introduces several modifi-
sponds to an average frequency linewidth for the scattered
cations which make the analysis more practical and appli-
spin waves. This allows the subsequent relaxation of these
cable. The second part provides a detailed and specific
scattered spin waves to be included in the analysis in a
development of the scattering integrals and limits which
simple and intuitive manner. Third, specific calculations of
must be invoked to perform practical two magnon scattering
angle dependent linewidths have been made for two limits:
calculations. The third part presents a qualitative and dia-
~i! a spherical void scattering limit in which the k-
grammatic discussion of these spin wave regimes which con-
dependent scattering factor from the SLK model is assumed
tribute to the scattering.
to apply and ~ii! an isotropic scattering limit in which this
A. A modified SLK two magnon scattering model scattering factor is assumed to be k-independent and equal to
For a scattering inhomogeneity of a given size, the SLK its average value over all angles. It is clear that purely spheri-
model predicts the rate at which energy is coupled from the cal voids represent a poor approximation to the inhomogene-
uniform precession of the magnetization vector over the en- ities in real materials. One would expect that these two limits
tire sample, which is excited through the FMR response, to a would bracket the real physical situation for ferrite films with
particular spin wave due to the dipolar interaction between various microstructure properties.
the scatterer and the spin wave. The net relaxation rate is The main result of the SLK theory is an expression for
then found from adding up the rates for all degenerate spin the relaxation rate due to a single scatterer. This rate is de-
wave states for all of the scatterers in the sample. In the termined from a Fermi golden rule calculation and can be
original SLK treatment, the scattering inhomogeneity was written as
modeled as a spherical void in an infinite sample. The as- 1 2p
sumption of spherical symmetry leads to a k-dependent cou-
T SLK
5
\ (
k0
@ F ~ k!# 2 d ~ \ v k 2\ v ! . ~22!
pling between the uniform precession and the degenerate
spin waves. That is, spin waves which propagate in certain Here, \ is Plancks constant, d (\ v k 2\ v ) is the Dirac delta
directions and at certain k values couple more strongly to the function, and the @ F(k) # 2 factor represents the coupling be-
uniform mode than others. This k dependence is due to the tween the spherical void and the degenerate spin waves. The
geometry of the dipole fields associated with the void. The sum spans the various k values in the spin wave band, but
SLK treatment then applied this result for a single spherical with the uniform mode excluded.
void to a sample with many identical scatterers by assuming The F(k) coupling factor is given by
that these scatterers operate independently.
One problem in the SLK formalism is that k-dependent m j i ~ kR !
scattering is associated directly with the spherical nature of F ~ k! 516p 2 M s R 3 ~ 3 cos2 u k 21 ! . ~23!
V kR
the void used for the analysis. Modifications were also intro-
duced to account for the expected occurrence of many non- Here, R is the scatterer radius, m is the Bohr magneton, V is
spherical scatterers. Nonspherical scatterers were included the sample volume, and j 1 (kR) is the first spherical Bessel
J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton 4355
function. In addition to the development in the original SLK case of nonspherical scatterers. In order to model the effect
paper, a detailed derivation of Eqs. ~22! and ~23! is given by of nonspherical inhomogeneities, it is reasonable to replace
Haas and Callen.39 It is useful to note that g, the magnitude (3 cos2 uk21)2 with its average value over all solid angles of
of the electron gyromagnetic ratio, is given by g 52 m /\. 0.8. This approach is designated isotropic scattering.
The matter of units will be considered shortly. It is important to note that Sparks,18 Seiden and
Equations ~22! and ~23! are the key working equations Sparks,19 and Sage20 have all proposed various modifications
for two magnon scattering calculations. The net relaxation to the SLK model to account for nonspherical scatterers. In
rate is in the form of a sum over all k states. The sum is each case, the resultant angular coupling function included
weighted by the Dirac delta function. This insures that only both spherical void and isotropic scattering terms. The
degenerate spin waves, that is, spin waves with v k 5 v , are present approach appears to be intuitively plausible and op-
included in the summation. Each relaxation rate sum term is erationally convenient.
proportional to @ F(k) # 2 , which is related to the interaction Consider now the role of the @ j 1 (kR)/kr # 2 term in the
energy between the uniform mode and the spin wave at wave relaxation rate expression. This function has a value of 1/9
vector k. for kR50 and is essentially zero for kR. p . This factor
In practice, the Dirac delta function in Eq. ~22! should be insures that the degenerate spin waves which participate in
replaced by a function of finite width in order to account for the two magnon process have wavelengths on the order of
the subsequent relaxation of the degenerate spin waves. One the scatterer size or larger. This means that for large inho-
simple approximation to the delta function which has finite mogeneities, only spin waves with low wave numbers can
width and unit area is contribute. For smaller inhomogeneities, higher wave num-
H
ber spin waves can participate. For very small R values,
1 Dvi Dvi
, v2 , v k, v 1 degenerate spin waves with all available k values can con-
G~ v,vk!5 \D v i 2 2 ~24! tribute appreciably to the scattering. It will be seen that the
0, otherwise. size of the scattering inhomogeneities will have a significant
Here, D v i is the intrinsic frequency linewidth which charac- effect on the shape of the linewidth versus field angle profile.
terizes the magnetic loss for the material in the absence of For simplicity, and following the SLK approach, it is
two magnon scattering. This linewidth is taken to be an order assumed that the material contains N independent, identical
of magnitude estimate to describe the intrinsic relaxation of scatterers of radius R. The net linewidth is then given by Eq.
the degenerate spin waves. ~22! multiplied by a factor of N.
Turn now to the @ F(k) # 2 term. Note that @ F(k) # 2 varies
as M 2s . This dependence on the square of the magnetization
reflects the dipoledipole origin of the two magnon scatter- B. Two magnon scattering integrals and limits
ing interaction. The specific form of the (3 cos2 uk21)2 term
is due to the assumption of spherically shaped scatterers. In order to evaluate the two magnon relaxation rates, it is
Because of the spatial dependence of the dipole field associ- useful to convert the sum over k states to an integration over
ated with the spherical void, spin waves which propagate in k space. The sum in Eq. ~22! is replaced by an integral.
(k 8 p 3 E E E d 3 k.
certain directions couple more strongly to the scatterer than V
others. For a spin wave propagating parallel to the static ~25!
magnetization direction, at u k 50, (3 cos2 uk21)2 is equal to
4. For a spin wave propagating perpendicular to the z axis at The factor V/8p 3 represents the number of states per unit
u k 590, (3 cos2 uk21)2 is equal to 1. For cos2 uk51/3, or volume in k space. The two magnon relaxation rate 1/T TM
u k 554.7, this angular coupling function is equal to zero. for N scatterers is given by
EE E
Spin waves at this propagation angle do not participate in the ` 2p p
1 NV
two magnon process at all. The use of the (3 cos2 uk21)2 5 @ F ~ k!# 2 G ~ v , v k !
term will be identified as spherical void scattering. T TM 4 p 2 \ 0 0 0
The vanishing of the two magnon coupling at u k
3k 2 sin u k d u k d f k dk. ~26!
554.7 for spherical void scattering is related to properties
of the dipole field around the void. If one considers a conical It is more useful, however, to cast Eq. ~26! in a form for
surface defined by all rays which extend away from the void which the integration limits directly reflect the spin wave
at an angle of 54.7 relative to the saturation magnetization band and range of degenerate states. To do so, it is necessary
direction, the static dipole field along this surface is always to eliminate the G( v , v k ) function in the integrand. Since
perpendicular to the static magnetization. The dynamic di- G( v , v k ) is equal to 1/\D v i for degenerate spin waves but
pole field associated with the uniform precession is, there- is zero otherwise, this function can be replaced by 1/\D v i if
fore, parallel to the direction of the static magnetization. The the integration limits are modified appropriately. The first
SLK analysis shows that this parallel dynamic field does not integration, over the cos uk variable, reflects the range of u k
couple to spin waves propagating in the coincident direction values for the degenerate spin waves for a given k and f k .
at u k 554.7. The second integration is over f k . For an isotropic material,
In addition to the assumption of spherical inhomogene- this integration is trivial because the spin wave dispersion
ities and the corresponding (3 cos2 uk21)2 coupling term, it relations show no dependence on the f k . For a uniaxial
will be useful to consider possible angle dependences for the material, the f k dependence can play a significant role and
4356 J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton
3 EE E
0
` p /2
f ~k!
b ~ k, f k !
a ~ k, f k !
S ~ 3 cos2 u k 21 !
j i ~ kR !
kR D 2
5
Dvi
1, v6 ,vmin C. Discussion of two magnon scattering regimes
2
A S Dvi
D The a(k, f k ), b(k, f k ), and f (k) limits can best be un-
2
Dvi
v 2max2 v 6 v min, v 6 , v max. derstood from graphical illustrations of the cos uk and f k
2 2
5 , integrations. The cos uk integration is placed in perspective
v 2max2 v 2min
in Fig. 8. The figure shows various spin wave dispersion
Dvi curves of frequency v k as a function of wave number k, all
0, v6 . v max
2 for a given value and orientation of the external magnetic
~28! field and some specific azimuthal spin wave propagation
angle f k . The format is the same as with the previous two
dimensional spin wave band diagrams. The top and bottom
The 6 signs apply to a(k, f k ) and a(k, f k ), respectively. dispersion branches are shown as solid curved lines labeled
The frequencies v min and v max are the same as given by Eqs. v min and v max , respectively. These curves correspond to the
~20! and ~21!. The limit values of zero and one for a(k, f k ) dispersion curves for u k 590 and u k 50, as in previous
and b(k, f k ) apply when the wave number k is greater than diagrams. The point labeled v FMR denotes the FMR fre-
the largest value of k allowed by the dispersion or when k is quency and the operating frequency of interest. Given the
less than the smallest allowed value, respectively. When both orientation angle f for the static external field, H ext , the
limits are zero or one, of course, the integral will vanish. value of the field has been set to yield this FMR frequency.
The function f (k) represents the minimum value of f k The solid horizontal lines in Fig. 8 delineate the range of
at v k 5 v for a given field orientation. This f (k) may be degenerate spin wave states which are within 1/2D v i of the
written in implicit form through the condition operating frequency. This width has been exaggerated for the
cos2 @ f ~ k !# sake of illustration. For FMR frequencies in the GHz range,
this degenerate band of states typically will have a width in
g 2 ~ H i 1Dk 2 !~ H i 1Dk 2 2H A sin2 f 14 p M s ! 2 v 2 the range of several tenths of a MHz to tens of MHz, de-
5 .
4 p M s g 2 H A sin2 f pending on the intrinsic loss level of the ferrite film under
consideration. The intersection of this horizontal band with
~29!
the overall spin wave band from v min to v max gives the re-
Equation ~29! is valid provided that the right hand side is gion of states in k and u k , for a given f k , which contribute
positive and less than or equal to 1. Otherwise, the function to the two magnon scattering. In order to illustrate the cos uk
f (k) is simply zero. By this definition, f (k) is zero for an integration limits, the diagram shows a particular k value as
isotropic material where H A 50. indicated by the vertical dashed line. The intersections of this
J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton 4357
selects the isotropic scattering limit in which this term is A. Linewidth versus angle profiles: Basic
replaced by its average value of 0.8 over all solid angles. dependences
One is now in a position to evaluate a two magnon re- It will be useful to develop first a qualitative picture for
laxation rate 1/T TM at these different values of f, and hence, the expected angle dependences and the effect of scatterer
to obtain information on the two magnon frequency line- size and type on the linewidth versus angle profiles. These
width 1/g T TM or the corresponding field linewidth DH as a issues are best addressed for the case of an isotropic material
function of f or u. For each f, this 1/T TM is obtained from where H A 50 because of the considerable simplifications to
Eq. ~27!. For a given f value, the first step is to determine the FMR field and spin wave dispersion expressions which
the effective internal field H i as given by Eq. ~14!. One can result in this limit.
then write explicit expressions for a(k, f k ) and b(k, f k ) As a further simplification, the angle dependence of the
based on Eqs. ~28!, ~20!, and ~21!. In order to evaluate Eq. relaxation rate is easiest to understand for the case of isotro-
~27!, the integral over cos uk should be performed first. This pic scattering. The effect of scatterer size on the linewidth
integration can be done analytically. The result is a compli- versus u profile will be first examined for this limit. For
cated double integral over k and f k which can be handled by isotropic scattering, the coupling between the uniform mode
a number of numerical methods. One approach would be to and the degenerate spin waves does not depend on the spin
use a quadrature algorithm for the f k integration nested wave propagation direction. In this case, the relaxation rate is
within a second quadrature algorithm for the k integration. simply related to the number of degenerate spin waves which
lie below the cut-off wave number k5 p /R and the corre-
sponding weighting of these states according to
V. EXPLICIT TWO MAGNON LINEWIDTH VERSUS @ j 1 (kR)/kr # 2 coupling term in the scattering integral. For the
ANGLE RESULTS case of very small scatterers, such that p /R is much greater
than the largest available k value at u 590, all of the avail-
In this section, some representative two magnon results able degenerate spin waves at a given u participate equally in
are presented. These are shown in terms of the frequency the scattering. In this situation, the only consideration which
swept linewidth D~v/g! and field swept linewidth DH versus contributes to the angle dependence of the FMR linewidth at
the angle u between the static magnetic field and the film constant frequency is the variation in the number of degen-
normal. Various linewidth angle profiles for the YIG, BaM, erate states with u. All states out to the largest available k
and ZnY materials, different size voids, and different scat- value for the specified u value, taken as k cut , contribute.
tering limits are shown. These results will demonstrate the These contributions are weighted, of course, according to
different types of linewidth angle profiles which can result @ j 1 (kR)/kr # 2 .
from two magnon scattering. This effect was shown through the dispersion diagrams
Recall from Eq. ~9! that the frequency linewidth D~v/g! in Fig. 5. These same dispersion diagrams are repeated in
is simply proportional to the relaxation rate. This means that Fig. 10, along with an additional diagram which shows the
the angle dependences for D~v/g! are identical to those for resulting two magnon frequency linewidth 1/g T TM as a func-
1/T. The only difference is that D~v/g! conveniently ex- tion of u. Figures 10~a!, 10~b!, and 10~c! show the spin wave
presses the calculated losses in terms of magnetic field units dispersion diagrams for three different field orientations as in
which correspond to linewidths. Recall as well that the field Fig. 5. These diagrams include, however, additional vertical
swept resonance linewidth DH can have significant angle dashed lines which indicate the position of the p /R cut-off
dependences even for a constant 1/T or, therefore, a constant wave number for two magnon scattering from the j 1 (kR)
D~v/g!. For YIG films and frequencies in the 10 GHz range term in the F(k) coupling function of Eq. ~23!. For very
and above, the D~v/g! and DH profiles are almost the same. small scatterer sizes, R is also very small and these p /R lines
This is due to the relatively small saturation induction and are always outside the spin wave band at the FMR frequency
the nearly parallel alignment of Hext and Ms as the field is point. This means that the entire band of degenerate states
rotated from out-of-plane to in-plane and Hext is adjusted to indicated by the shaded horizontal strip of modes in ~b! and
satisfy the requisite FMR condition at a given angle. For ~c! contributes to the two magnon scattering losses.
uniaxial and planar anisotropy materials, however, the larger Figure 10~d! shows the frequency linewidth as a function
4 p M s values and large anisotropy fields cause the angle of the field angle u which results. For small angles, the spin
between Hext and Ms to vary considerably as one sweeps the wave band is high in frequency, few states are degenerate,
field through the FMR response curve. This leads to a sig- and the resulting relaxation rate is small. In the limit u 0,
nificant difference in the angle profiles for D~v/g! and DH in the number of degenerate states goes to zero as well for a
these materials. thin film of infinite extent, and the two magnon linewidth
The g, 4 p M s , and H A values used for the results shown also goes to zero. As the angle is increased, however, the
below are the same as those used to calculate the field versus spin wave band shifts down in frequency relative to the FMR
angle and linewidth versus angle results shown in Figs. 2 and frequency position, more and more states become degenerate
3. The exchange constant D in the spin wave dispersion re- with the FMR frequency, and the linewidth increases. As one
lations is taken to be independent of the propagation angle approaches the parallel field FMR condition at u 590, one
and to have a nominal value of 531029 Oe cm2/rad2, the reaches the configuration with the maximum number of de-
same value used for the spin wave dispersion relations of generate spin waves and the two magnon linewidth reaches a
Figs. 57. maximum. It should be emphasized that the linewidth versus
J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton 4359
FIG. 10. ~a!~c! show schematic spin wave dispersion curves of spin wave FIG. 11. ~a!~c! show schematic spin wave dispersion curves of spin wave
frequency v k vs wave number k. These diagrams are for a thin isotropic frequency v k vs wave number k in the same format and under the same
film and three orientations of the external static field with the sample biased conditions as for Fig. 10. Also as in Fig. 10, the vertical dashed line at k
at the static field needed for FMR at some constant frequency, as indicated. 5 p /R in each diagram shows the position of the effective wave number
These diagrams follow the same format as Fig. 5. For ~a! the static field is associated with an inhomogeneity radius R. For these diagrams, however, a
applied perpendicular to the film plane at u 50. For ~b! the static field is large size inhomogeneity is assumed, so that the k5 p /R point is always
applied at some intermediate angle. For ~c! the static field is applied in the inside the spin wave band at the FMR frequency point. The solid line in ~d!
film plane at u 590. The shaded horizontal regions in ~b! and ~c! show the shows the variation in the two magnon scattering frequency linewidth
band of available degenerate spin wave states for two magnon scattering in 1/g T TM with the external field angle. The dashed line repeats the small size
each case. The vertical dashed line at k5 p /R in each diagram shows the inhomogeneity result from Fig. 10.
position of the effective wave number associated with an inhomogeneity
radius R. For these diagrams, a small size inhomogeneity is assumed, so that
the k5 p /R point is always outside the spin wave band at the FMR fre-
quency. The solid line in ~d! shows the variation in the two magnon scat- linewidth versus angle profile. At small values of u, as long
tering frequency swept linewidth 1/g T TM with the external field angle. as k cut is less than p /R, the two magnon linewidth is the
same as in Fig. 10. This result is shown by the part of the
solid curve in Fig. 11~d! out to 10 or so. The kink in ~d!
angle profile result in Fig. 10~d! is for: ~i! isotropic scattering occurs at the u value for which p /R5k cut is satisfied. For
and ~ii! very small scatterer sizes. The profile which results is larger u values, k cut is now greater than p /R. Now, the
a reflection of the variation in the number of degenerate spin modes which can contribute to the two magnon scattering
waves with angle, and no other effects. truncate at k5 p /R rather than at k cut and the two magnon
For YIG parameters and an FMR frequency in the GHz linewidth is reduced from the dashed line result in ~d! to the
regime, the results in Fig. 10 would correspond to small solid line result. The space between the solid line and the
scatterers with sizes on the order of 0.05 mm or smaller. If dashed line in ~d! represents those modes between k5 p /R
one were to increase the size of the scatterers, one could and k5k cut which are now excluded from the scattering pro-
move the p /R cut-off wave number inside the spin wave cess. As a result, the linewidth continues to increase with
band for u values above some lower limit. This effect, the angle, but at a lower rate than for the case of small scatterers.
resulting change in the scattering, and the change in the line- The increase in linewidth with u is now due to the increase in
width versus field profile for a large scatterer size are shown the number of degenerate states for the low k modes only,
in Fig. 11. The format of Fig. 11 is the same as Fig. 10, with that is, those modes which satisfy the condition k, p /R.
spin wave band diagrams in ~a!, ~b!, and ~c!, and a linewidth The results in Figs. 10 and 11 show how the details of
versus u diagram in ~d!. Note that for ~b! and ~c!, the vertical the microstructure affect the two magnon linewidth versus
dashed lines at k5 p /R are now well inside the spin wave field angle profile for thin films. The truncation effect which
band at the FMR frequency. The solid line in ~d! shows the is evident from the solid and dashed curves in Fig. 11~d!
calculated two magnon linewidth versus angle profile and the shows that the size of the scatterer plays a crucial role in
dashed line repeats the result from Fig. 10~d!. shaping these profiles. For the example shown, one can see
As shown by the kink in the solid curve in Fig. 11~d! that the position of the kink in the profile, if such a kink can
near u 515 or so, and departure of the solid line from the be observed, ~i! indicates a relatively large scatterer size and
dashed line at larger u values, the change from a small scat- ~ii! provides a direct indication of the size of the scatterer. In
terer size to a large scatterer size has a drastic effect on the the case of YIG parameters, a kink should be observable for
4360 J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton
ity p parameter of 0.01. The dashed lines show the two mag- also been neglected. It is reasonable to expect that the ellip-
non frequency swept linewidth D( v / g )51/g T TM angle pro- ticity would influence the coupling between the inhomoge-
files. The solid lines show the two magnon field swept neities and the degenerate spin waves. Third, a scattering
linewidth DH TM profiles. The field swept linewidths were Hamiltonian based on fluctuations in the effective anisotropy
obtained from the frequency swept two magnon linewidths field H A rather than the saturation induction 4 p M s might
and Eq. ~10!, with the FMR derivative factor provide a better model of the microstructure in highly aniso-
g ] H FMR / ]v FMR evaluated numerically at each u. From the tropic ferrites.
results in Fig. 3, it is evident that this factor is equal to one at
u 50, greater than one when Ms rotation effects are impor- VII. SUMMARY
tant at intermediate angles, and less than one for resonance
The two magnon scattering contribution to the ferromag-
when the static field is in plane at u 590.
netic resonance relaxation rate and linewidth has been evalu-
The results in Fig. 16 are self explanatory. In the BaM
ated as a function of field angle for isotropic and anisotropic
case, the roughly 50% effect at u ;60 from Fig. 3 serves
disks and films. The analysis was based on the original SLK
mainly to magnify the angle profile for DH TM . The basic
treatment but has been extended to include anisotropic ma-
profile structure found for 1/g T TM is retained. Although it is
terials, nonspherical scatterers, sample demagnetization fac-
not readily evident from the curves shown, the field swept
tors, and a nonzero linewidth for the scattered spin wave
linewidth at u 590 is slightly smaller than the frequency
modes. The analysis yields characteristic angle dependences
swept linewidth. This is a direct consequence of the fact that
in the two magnon scattering linewidth which reflect the
the rotation effects for u values different from 0 and 90
scatterer size, sample shape, scattering interaction, and crys-
cause g ] H FMR / ]v FMR to be greater than unity and to
talline anisotropy. Specific numerical evaluations were done
broaden the field swept FMR line, while dynamic demagne-
for three types of materials: YIG, BaM, and ZnY. These
tizing effects which affect the FMR frequency field equation
explicit calculations serve to demonstrate the various effects
at u 590 give a g ] H FMR / ]v FMR multiplier which is less
of the spin wave band shift relative to the FMR frequency,
than unity.
anisotropy, and microstructure on the two magnon relaxation
The situation is quite different for ZnY. Here, the
process. The results demonstrate that measurement of the
600% effect for the 1/g T TM to DH TM conversion, combined
FMR linewidth as a function of angle for film and disk ma-
with relatively sharp peaks for both 1/g T TM and
terials can provide useful information on the microstructure
g ] H FMR / ]v FMR over roughly the same range in u values,
and two magnon losses in the sample.
leads to a very significant change. One finds that ~i! the peak
in DH TM is accentuated to the extreme and shifted in u, and
ACKNOWLEDGMENTS
~ii! the DH TM for large values of u actually falls below
1/g T TM . For ZnY, this decrease in DH TM below 1/g T TM at The authors acknowledge Dr. P. Kabos, Dr. B. A.
large u is much more distinct than for BaM, as was also the Kalinikos, Dr. M. A. Wittenauer, and Dr. A. K. Srivastava
case in Fig. 3. for helpful discussions during the course of this study. Dr. A.
K. Srivastava provided a critical reading of the final manu-
script. This work was supported, in part, by the United States
VI. ADDITIONAL CONSIDERATIONS Office of Naval Research Grant Nos. N00014-94-1-0096 and
N00014-91-J-1324, the United States Army Research Office,
As discussed in Sec. I, an important consideration which
Grant No. DAAH04-95-1-0325, and the National Science
has been neglected in this work involves the effect of film
Foundation, Grant No. DMR-9400276.
thickness on the normal spin wave modes. For thin films, the
bulk spin wave dispersion relations are not the true normal 1
See, for example, F. Bertaut and F. Forrat, C. R. Acad. Sci. 242, 382
modes of the system, due to the film surfaces and the corre- ~1956!; G. Giello and M. A. Gilleo, Acta Crystallogr. 10, 239 ~1957!.
sponding boundary conditions on the dynamic magnetiza- 2
C. E. Patton, in Magnetic Oxides, edited by D. J. Craik ~Wiley, London,
tion. Instead of the continuous spectra of spin wave disper- 1975!, Ch. 10, pp. 575645.
sion curves within the band, the spin wave spectrum is
3
M. Sparks, Ferromagnetic Relaxation Theory ~McGraw-Hill, New York,
1959!.
discretized. One would then expect the angle dependence of 4
M. Sparks, R. Loudon, and C. Kittel, Phys. Rev. 122, 791 ~1961!.
the linewidth to reflect the discrete nature of the normal 5
H. B. Callen, J. Phys. Chem. Solids 4, 256 ~1958!.
6
modes. The dipole-exchange mode dispersion relations de- R. C. Fletcher, R. C. LeCraw, and E. G. Spencer, Phys. Rev. 117, 955
veloped by Kalinikos et al.36 for obliquely magnetized aniso- ~1960!.
7
E. Schlomann, A.I.E.E. Special Pub. T-91, 600 ~1956!.
tropic thin films, for example, might be used in place of the 8
M. Motizuki, M. Sparks, and P. E. Seiden, Phys. Rev. 140, A972 ~1965!.
bulk dispersion relations in order to determine the effect of 9
E. Schlomann, Phys. Rev. 182, 632 ~1969!.
film thickness. 10
C. R. Buffler, J. Appl. Phys. 30, 172S ~1959!; 31, 222S ~1960!.
There are a number of additional considerations which
11
M. Mita, J. Phys. Soc. Jpn. 22, 529 ~1967!.
12
J. Nemarich, Phys. Rev. 136, A1657 ~1964!.
may be important in the two magnon calculation but have 13
A. S. Risley and H. E. Bussey, J. Appl. Phys. 35, 896 ~1964!.
been neglected in this analysis. First, the exchange constant 14
A. S. Risley, E. G. Johnson, and H. E. Bussey, J. Appl. Phys. 37, 656
D has been taken to be independent of propagation angle. As ~1966!.
15
C. E. Patton, C. H. Wilts, and F. B. Humphrey, J. Appl. Phys. 38, 1358
mentioned above, this may not be the case for highly aniso-
~1967!.
tropic BaM hexagonal ferrite materials. Second, the effect 16
C. E. Patton, J. Appl. Phys. 39, 3060 ~1968!.
of the ellipticity of the uniform mode and the spin wave has 17
M. Sparks, Phys. Rev. 1, B3856 ~1969!.
J. Appl. Phys., Vol. 83, No. 8, 15 April 1998 M. J. Hurben and C. E. Patton 4365
18
M. J. Hurben, D. R. Franklin, and C. E. Patton, J. Appl. Phys. 81, 7458 30
N. Bloembergen, Phys. Rev. 78, 572 ~1950!.
~1997!. 31
W. H. von Aulock, Handbook of Microwave Ferrite Materials ~Academic,
19
P. E. Seiden and M. Sparks, Phys. Rev. 137, A1279 ~1965!. New York, 1965!.
20
J. Sage, Phys. Rev. 185, 859 ~1969!. 32
L. Landau and E. Lifshitz, Phys. Z. Sowjetunion 8, 153 ~1935!.
21
E. Schlomann, R. I. Joseph, and I. Bady, J. Appl. Phys. 34, 672 ~1963!. 33
B. Lax and K. J. Button, Microwave Ferrites and Ferrimagnetics
22
C. E. Patton and W. Jantz, IEEE Trans. Magn. 14, 698 ~1978!. ~McGraw-Hill, New York, 1962!.
23
C. E. Patton, Phys. Status Solidi B 92, 211 ~1979!. 34
T. Wolfram and R. E. De Wames, Phys. Rev. B 4, 3125 ~1971!.
24
C. E. Patton, Phys. Status Solidi B 93, 63 ~1979!. 35
R. E. De Wames and T. Wolfram, J. Appl. Phys. 41, 987 ~1970!.
25
C. Vittoria, J. Magn. Magn. Mater. 21, 109 ~1980!. 36
B. A. Kalinikos, M. P. Kostylev, N. V. Kozhus, and A. N. Slavin, J. Phys.:
26
P. C. Dorsey, R. Seed, C. Vittoria, D. B. Chrisey, C. Carosella, P. Lubitz, Condens. Matter 2, 9861 ~1990!.
and J. S. Horwitz, IEEE Trans. Magn. 28, 3216 ~1992!. 37
M. J. Hurben and C. E. Patton, J. Magn. Magn. Mater. 139, 263 ~1995!.
27 38
P. C. Dorsey, S. E. Bushnell, R. G. Seed, and C. Vittoria, IEEE Trans. See, for example, V. Kambersky and C. E. Patton, Phys. Rev. B 11, 2668
Magn. 29, 3069 ~1993!. ~1975!.
28 39
F. J. Cadieu, R. Rani, W. Mendoza, B. Peng, S. A. Shaheen, M. J. Hurben, C. W. Haas and H. B. Callen, in Magnetism, edited by G. T. Rado and H.
and C. E. Patton, J. Appl. Phys. 81, 4801 ~1997!. Suhl ~Academic, New York, 1963!.
29
C. Kittel, Phys. Rev. 73, 155 ~1948!. 40
J. A. Osborn, Phys. Rev. 67, 351 ~1945!.