POSSIBLE WAYS TO
CONTROL INTENSITY OF
STOCHASTIC RESONANCE IN
MAGNETIC NANOPARTICLES
A.G. Isavnin
Kazan State University, Russia
1. Introduction
Effect of stochastic resonance means previous
quite sharp increase and subsequent gradual decay of response of a multistable
system to a weak periodic modulation as internal noise of the system goes up
[1]. Such a phenomenon is well explored both theoretically and experimentally
within wide range of applications, from global climatic changes to nervous processes
in live organisms [2]. Generally external periodic signal is assumed to be
weak, so without noise any transitions between stable states of the system are
impossible. At certain noise intensity regular part of dynamics of the system
increases, that is the transformation of stochastic energy into coherent one
occurs. The two-state theory says that for a bistable system the response
reaches its maximum when mean time of transitions between stable states becomes
comparable to a half of period of modulation. At greater noise intensities the
coherency of signal and noise fails and the response goes down.
Single-domain magnetic particles with «easy axis»
anisotropy are bistable elements with steady states associated with two
opposite directions of easy axis. Such nanoparticles are technologically
important materials and their unique properties continuously attract attention
of scientists. In previous papers stochastic resonance was considered in fine
magnetic particles for thermal [3,4] and tunneling [5,6] modes of magnetization
reversal. Obtained results displayed specific non-monotonous dependence of the
response to radiofrequency signal on the noise level. Values of signal-to-noise
ratio and components of dynamic magnetic susceptibility derived within the
two-state approximation were verified by means of numerical calculations [7,8],
based on a Fokker-Planck equation with periodic drift term.
Present paper considers the influence of
additional external magnetic field applied transverse to the easy axis on dynamics
of the particle’s magnetic moment as stochastic resonance rises. As well, it
considers effects that emerge when source of periodic signal or the sample
moves towards each other.
2. Single-domain magnetic
particle under conditions of stochastic resonance
Hereafter it is assumed that stochastic resonance
is in thermal, overbarrier mode. Energy of the uniaxial monodomain particle
under such circumstances is
.
(1)
Here the first term describes interaction of the magnetic moment of
superparamagnetic particle with anisotropy field (K is anisotropy constant, v
is the particle’s volume, q
is the angle between magnetization vector and easy axis), the second term is
associated with permanent magnetic field Í1. Double well
potential (1) has maximum at q2=p/2 and minima at q1=arcsin(m0MH1/(2K)),
q3=p-arcsin(m0MH1/(2K)). The height of
potential barrier separating two minima is
DU = E(q2) - E(q1) = Kv -
m0MH1v
+ m02M2H12v/(4K)
.
(2)
As Í1
increases, the stable orientations of the magnetization vector shift to the
direction of the permanent transverse field. Bistability of the system vanishes
at H1=2K/(m0M)
and just one minimum remains at q=p/2.
Let us like in [4] associate input signal with
external radiofrequency field and input noise with temperature T of the sample.
Then output of the system is change of the magnetic moment. Under modulation of
the particle with weak external field Ícos(Wt),
applied along the easy axis, double-well potential
, (3)
begins to swing slightly - in one half of the modulation period 2p/W the right minimum becomes higher and the left one
becomes lower, in the next half of period - vice versa. The external periodic
signal is assumed to be weak enough, so it alone cannot lead to changes of
direction of the particle’s magnetic moment. This implies the condition m0MHv < DU holds. Thermal activation of the
system increases rate of switches of the particle’s magnetic moment and that
leads to possibility to surmount the potential barrier.
Further the discrete-orientations approximation
is used in the calculations. Therefore the magnetic moment of the particle is
allowed to be in just two states corresponding two minima of the double well.
It is convenient to define discrete variable x=Mcosq describing projection of the magnetization vector to
the easy axis. This variable can take only two values:
.
(4)
The two-state theory used trough this paper implies that instead of
continuous diffusion of the particle’s magnetic moment over a sphere we
consider its stochastic switches between two directions. Advantage of such
approximation is possibility of using the master equation for transition rates
that yields analytical solution. The master equation is [1,4]:
. (5)
Here n±
is the probability of discrete variable x=Ìcosq
to take value x± = ±Ì0.
W±(t) is escape
rate from ± state
corresponding to stable directions of the magnetic moment at angles q1, q3 to the easy
axis. Such rate is described with Kramers-type formula [9, 4]:
. (6)
The prefactor of Kramers formula a0 [3] is commonly near ferromagnetic
resonance frequency and for iron fine particles has value of order 108 -
1010 s-1.
Probability density in the discrete orientations
approximation reads:
. (7)
The
potential is symmetric with respect to x=0, so x+ = -x- =
M0. Solution of the master
equation (5) is:
. (8)
Here W=2a0exp(-DU/(kT)) is doubled Kramers
escape rate of the system from either minimum of the symmetric non-modulated
potential (1), A=m0M0Hv/(kT)
is dimensionless amplitude of external modulation, f=arctan(W/W)
is phase shift between response of the system and external periodic signal. Probability n+(t0)
is equal to 1 if initial orientation of the magnetization is +M0 ,
and 0 if at t = t0 x = -M0.
Power spectrum of the system as Fourier transform
of the autocorrelation function
, (9)
displays Lorentzian background associated with stochastic dynamics and d - spike describing regular motion of
the vector M at the frequency W
of external signal.
One of the main indexing values of stochastic
resonance is output signal-to-noise ratio (SNR). Using (9) and assuming the
amplitude of external alternating field to be small such ratio can be written
as:
. (10)
The temperature dependence of the SNR on T has
bell-shaped distinct maximum. The SNR goes up as permanent transverse field Í1
rises [10].
3. DYNAMIC MAGNETIC
SUSCEPTIBILITY
The SNR determines fraction of regular part of
the system dynamics but contains no information about phase of output signal.
In the study of stochastic resonance there is another approach, namely to
consider the effect as amplification of a weak signal [11]. Following this way
one should also find phase lag between output and input signals. As in general
case the magnetic moment of particle does not change its direction coherently
with external alternating field, it is real part of dynamic magnetic susceptibility
that could be an appropriate index to describe amplitude amplification. Let us
introduce complex susceptibility c =
Rec - iImc and suppose for simplicity the initial phase of the
modulation to be zero. Then expression
for time-dependent magnetization is
.
(11)
Asymptotic probabilities associated with stationary mode can be derived
from (8) at t0 tending to minus infinity:
, n - (t) = 1 - n + (t)
. (12)
Taking into account (7), one can obtain asymptotic averaged value áx(t)ñ :
.
(13)
Obviously, real part of the susceptibility that describes co-phase
motion of M and H, and imaginary part that describes motion with lag = p/2
are given with following expressions:
, (14)
.
And absolute value of the dynamic susceptibility is
. (15)
The real part is presented in Fig.2 as a function
of temperature of the system. As in the case of zero permanent field [4,7], the
curves have the specific bell-shaped form, typical for stochastic resonance.
The auxiliary permanent field Í1 , applied transverse to the easy
axis results in changing the potential barrier (2) and increases the response
of system to external signal [12]. More precise results may be derived with
non-constant value of prefactor α0 in (6) as analyzed in [13].
4. CASE OF MOVING
SAMPLE
Now let us consider the effects of motion. Suppose
the sample with nanoparticles is moving towards the source of the external rf
modulation with velocity V<<c (here c is velocity of light in space).
Then, due to Doppler effect, we shall observe frequency of the external signal
is to go up:
.
(16)
This “new” frequency W’
can be now used in (10), (14) and (15) instead of W.
If the motion is in the opposite direction, then V in the denominator of (16)
changes its sign and we have the frequency of external signal going down. As it
was shown in [4,7], at lower frequencies one can observe higher values of SNR
and components of the dynamic susceptibility c.
Of course, the same effects will occur if the sample is fixed and the source of
periodic modulation is moving.
Experimental scheme of measurement of the dynamic
magnetic susceptibility can be found for example in [14]. Observation of
dynamics of magnetic moment for individual particle is presented in [15]. For
an ensemble of fine particles it is convenient to use frozen ferrofluids. By
diluting ferrofluid the desired concentration of magnetic clusters may be
produced. It means that the strength of interactions between nanoparticles is
possible to tune. By freezing ferrofluid while permanent magnetic field applied
one can get solid dispersion of fine particles with their easy axes aligned in
the same direction.
5. CONCLUSION
This paper considers effect of thermal stochastic
resonance in the system of single domain magnetic nanoparticles with uniaxial
anisotropy. In addition, the influence of a permanent magnetic field applied
transverse to the easy axis is examined. Such a field leads to a shift of
stable directions of the magnetic moment in space and diminish the height of
potential barrier separating two minima. As a result, the signal to noise ratio
and dynamic magnetic susceptibility reveals higher values. It is important for
observation of stochastic resonance and for amplification of weak periodic
fields.
Another interesting effect presented in this
paper is the Doppler frequency shift that also can vary the intensity of
stochastic resonance.
So, there are two mores mode to tune the bistable
magnetic system for amplification of external weak periodic signal. Such regimes
seem to be quite useful in cases with difficulties to vary thermal noise
intensity (e.g. near Curie temperature, etc.). The results of present
investigation allow not only to enhance weak signals, but choose and set
parameters of the system to avoid this enhancement when it is unwelcome.
REFERENCES
1.
McNamara B., Wiesenfeld K. // Phys.Rev.A (1989), V.39, N
9, P.4854-4869.
2.
Wiesenfeld K., Pierson D., Pantazelou
E., Dames C., Moss F. // Phys.Rev.Lett.
(1994), V.72, P.2125-2129.
3.
Sadykov E.K. // J.Physics: Cond.Matt.
(1992), V.4, P.3295-3298.
4.
Sadykov E.K., Isavnin A.G. // Physics
of the Solid State (1994), V.36, N 11, P.1843-1844.
5.
Sadykov E.K., Isavnin A.G., Boldenkov
A.B. // Physics of the Solid State (1998), V.40, N 3, P.474-476.
6.
Isavnin A.G. // Physics of the Solid
State (2001), V.43, N 7, P.1263-1266.
7.
Sadykov E.K., Isavnin A.G. //
Hyperfine Interactions (1996), V.99, P.415-419.
8.
Isavnin A.G. // Russian Physics
Journal (2002), V.45, N 11, P.1110-1114.
9.
Brown W.F.(jr.) // Phys.Rev. (1963),
V.130, N 5, P.1677-1686.
10.
Isavnin A.G. // Vestnik OGU (2006), N
10, P.390-393.
11.
Jung P., Hanggi P. // Phys.Rev.A (1991), V.44, N 12,
P.8032 -8042.
12.
Isavnin A.G. // Russian Physics
Journal (2005), V.48, N 5, P.511-515.
13.
Isavnin A.G. // Russian Physics
Journal (2007), V.50, N 5, P.54-58.
14.
Abu-Aljarayesh I., Bayrakdar A.,
Yusuf A., Abu-Satia H. //
J.Appl.Phys. (1993), V.73, N 10, P.6970-6972.
15.
Wernsdorfer W., Bonet Orozco E., Hasselbach
K., Benoit A., Barbara B., Demoncy N., Loiseau A., Pascard H., Mailly D. //
Phys.Rev.Lett.(1997), V.78, P.1791-1794.