Giant lasing effect in magnetic nanoconductors
Extracted source text · arXiv cond-mat/0306522
%%% Mats final (minor) corrections June 10, 2003: as submitted to PRL
%%%% Anatoli's modification June 05, 2003
%Mats modifications made May 13, 2003
\documentstyle[prl,aps,epsf,two column]{revtex}
%\documentstyle[prl,aps,epsf,preprint]{revtex}
%%\oddsidemargin 15mm
%%\evensidemargin 15mm
%%\textwidth 14.5cm
%%\topmargin -1cm
%%\textheight 23 cm
\include{psfig}
\begin{document}
\title{Giant lasing effect in magnetic nanoconductors}
\author{A. Kadigrobov$^{1,2}$,
\fbox{Z. Ivanov} and T. Claeson$^{1}$}
\address{
$^{1}$Department of Microelectronics and Nanoscience, Chalmers
University of Technology,
SE-412 96 G\"oteborg, Sweden\\
}
\author{R. I. Shekhter$^{2}$, and M.~Jonson$^{2}$
}
\address{
$^{2}$Department of Applied Physics,
Chalmers University of Technology and G\"oteborg University,
SE-412 96 G\"oteborg, Sweden
}
% \begin{document}
\date{\today}
\maketitle
\begin{abstract}
We propose a new principle for a compact solid-state laser in the 1-100
THz regime.
This is a frequency range where attempts to
fabricate small size lasers up till now have met severe technical
problems.
The proposed laser is based on a new mechanism for creating spin-flip
processes
in ferromagnetic conductors. The mechanism is due to the interaction of
light
with conduction electrons; the interaction strength, being proportional
to
the large exchange energy, exceeds the Zeeman interaction by orders of
magnitude.
On the basis of this interaction, a giant lasing effect is predicted
in a system where a population inversion has been created by tunneling
injection of
spin-polarized electrons from one ferromagnetic conductor to another ---
the
magnetization of the two ferromagnets having different orientations.
Using experimental data for ferromagnetic manganese perovskites with
nearly
100\% spin polarization we show the laser frequency to be in the range
%$10^{12} \div10^{14}$ s$^{-1}$. We also show
1-100 THz.
The optical gain is estimated to be
%$g_{opt} \sim
of order $10^7$ cm~$^{-1}$, which exceeds
the gain of conventional semiconductor lasers by 3 or 4 orders of
magnitude.
A relevant experimental study is proposed and discussed. \\
\end{abstract}
The ability to synthesize magnetically ordered, layered
conductors with nearly 100$\%$ spin-polarization
of the conducting electrons \cite{Tokura}, has opened up a new field
in solid state
physics, the field of "spintronics" \cite{Prinz,Wolf}.
Spin-dependent tunneling of electrons is one phenomenon that has already
found commercial applications based on the resulting "giant"
magnetoresistance of certain layered structures \cite{Prinz,Wolf}.
Other applications are bound to follow.
The possibility to control, by means of a bias voltage, not only the
energy but also the spin of electrons injected into a magnetic conductor
makes it feasible to investigate the properties of highly excited
spin-polarized electrons. An example of such a system is presented in
Fig.\ref{hot}, where the hatched region corresponds to an equilibrium
distribution of (spin-up) electrons in a spin-polarized conductor.
The dotted area marks a non-equilibrium distribution of "hot"
(spin-down) electrons.
Relaxation of the spin-down electrons to an equilibrium configuration
requires
spin-flip processes and is therefore completely blocked if such
processes
are not allowed. In the presence of such a "spin lock" against
relaxation,
highly excited states in the material may have a long lifetime, which
may in turn
determine novel "spintronics" effects in spin-polarized conductors.
The objective of this Letter is to demonstrate how electromagnetic
radiation
may remove the spin-lock effect and to demonstrate some important
consequences
of this effect for spintronics. We will show that the radiation makes
the blockade
of relaxation weaker through
its coupling to the exchange interaction in magnetically ordered
conductors. This comes about via the dependence of the exchange constant
on
the momenta of the conduction electrons.
As a result a lasing effect is shown to occur in systems where an
inverted
electron population has been created by the tunneling injection of
spin-polarized
electrons from one ferromagnetic conductor to another (the orientation
of the
magnetization being different in the two ferromagnets).
An example of such a system is presented in Fig.(\ref{population}).
Our estimations show that a laser with an optical gain that
exceeds the gain of conventional semiconductor lasers by three or
four orders of magnitude can be built and argue that laser action can be
achieved provided care is taken to design the system so that the lasing
region is not too much heated.
The frequency of these lasers can be in a wide range that
includes the interval 1 - 100 THz.
%
%$\omega \sim 10^{12} - 10^{14}$ s~$^{-1}$.
%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}
\centerline{\psfig{figure=zlaser1.eps,width=3cm}}
\vspace*{2mm}
\caption{Schematic representation of the band structure of a
magnetically
ordered conductor. The hatched region corresponds to an equilibrium
distribution of spin-up electrons in the lower band, while the dotted
region
indicates a non-equilibrium distribution of "hot" spin-down electrons
in the upper band. Arrows show electron spin directions in the bands}
\label{hot}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%
The Hamiltonian for the electrons
in a magnetically ordered conductor can be written as
%
\begin{equation}
\hat{H}_0 = \hat{\sigma_0} \frac{\hat{p}^2}{2m*}
-\hat{\mbox{\boldmath$ \sigma$}}{\bf I}\,,
\label{ham0}
\end{equation}
%
where $m*$ is the effective mass, $ \hat{{\bf p}}$ is the momentum
operator,
$\hat{\mbox{\boldmath $\sigma$}}$ are Pauli matrices, $\sigma_0$ is
the $2\times 2$ unity matrix, and $I$ is the exchange energy.
According to Eq.(\ref{ham0}) the dispersion law for electrons with
spins pointing up/down is
%
\begin{equation}
E_{\uparrow \downarrow}\left(p \right) =\frac{p^2}{2m*} \mp I
\label{dispersion}
\end{equation}
%
We will deal with a system, schematically presented in
Fig.\ref{population}, in which two potential barriers divide the
magnetic conductor into three parts;
the magnetization of two adjacent magnetic conductors (regions A and B)
are
pointing in opposite directions.
A bias voltage $V$ is applied between regions A and C. We assume that
the spin relaxation time $\nu_s^{-1}$ \cite{absorpion}, the time of
energy relaxation without
changing the electron spin direction $t_E$ and the time of electron
tunneling $t_{tun}$
obey the inequality $t_E \ll t_{tun} \ll \nu_s^{-1}$.
In the absence of spin-flip processes, the energy relaxation of injected
spin polarized electrons
creates a non-equilibrium state in which equilibrium is established
only within each group of electrons with a fixed spin polarization.
Therefore, in region B electrons in the spin-up and spin-down
energy bands are in local equilibria corresponding to the different
chemical potentials $\mu_\uparrow $ and $\mu_\downarrow$,
respectively, while the system as a whole is far from equilibrium.
According to Eq.(\ref{dispersion}) the energy conservation law for
vertical transition
of electrons with emission of photons of frequency $\omega$
does not depend on the electron momentum:
%
\begin{equation}
E_{\uparrow }\left(p \right) - E_{\downarrow}\left(p \right) - \hbar
\omega = 2 I - \hbar \omega
\label{conserv}
\end{equation}
%
It follows that for $\omega = 2I/\hbar$ all "hot" electrons are in
resonance
with the electromagnetic field, and hence stimulated emission of light
due to transitions of
electrons from filled states of the upper band to the empty states of
the lower band is
possible for all electrons in the energy range
$\mu_\uparrow - \mu_\downarrow $.
As seen in Fig.\ref{population}, under the neutrality condition
$N_\uparrow +N_\downarrow =N^{(0)}$ ($N_\uparrow$ and $N_\downarrow$
are the densities of spin-up and spin-down electrons,
$N^{(0)}$ is the equilibrium electron density)
the population inversion
needed for lasing requires a large enough bias voltage,
$V > (2 I - \mu_0)/e$.
The conventional Zeeman term $\hat{H}_Z= g\mu_B {\bf
H}\hat{\mbox{\boldmath$ \sigma$}}$
describing interaction between the (hot) electrons and an
electromagnetic field does
provide a mechanism for stimulated radiative transitions between the
energy bands
containing electrons with opposite spin
directions. However, it is relatively small in magnitude and it is not
the most important
mechanism.
For ferromagnets, we would like to suggest a much more effective
mechanism of
interaction between light and conduction electron spins.
This mechanism is based on the dependence of the exchange energy
(\ref{ham0}) on
the momentum ${\bf p}$ of the conduction electron. The momentum
dependence has to
do with the
overlap of the wavefunctions of the conduction electron and the magnetic
sub-system
(see, e.g., \cite{Zeiger}). It is determined by the value of
$ p a/\hbar$, where $a$ is the characteristic size of the orbital (that
is $a$ is an
atomic-scale length). This is why it varies with the momentum of the
conduction electron.
In the absence of an electromagnetic field the Hamiltonian which
describes this situation
can be written as
\begin{equation}
\epsilon(\hat{{\bf p}}) = \hat{\sigma_0} \frac{\hat{p}^2}{2m*} -
\hat{\mbox{\boldmath$ \sigma$}}{\bf I}(\hat{{\bf p}}) \,.
\label{implicit}
\end{equation}
In the presence of an electromagnetic field with vector potential ${\bf
A}
$ the momentum operator $\hat{{\bf p}}$ in Eq.(\ref{implicit}) must be
changed
to $\hat{{\bf p}} -(e/c) {\bf A}$. Expanding in powers of $(e/c) {\bf
A}$ one gets
an effective Hamiltonian $\hat{H}_{eff} =\epsilon(\hat{{\bf p}}) +
\hat{H}_{eff}^{(1)}$
where the perturbation Hamiltonian \cite{spin-orbit} is
\begin{equation}
\hat{H}_{eff}^{(1)} =-\frac{e}{2c}\hat{\mbox{\boldmath$ \sigma$}} (A_i
\frac{\partial{\bf I}}{\partial p_i} + \frac{\partial{\bf I}}{\partial
p_i} A_i)_{{\bf p} =\hat{{\bf p}}}
\label{effham}
\end{equation}
In Eq.(\ref{effham}) we have omitted terms that do not flip spins;
summation over double indices is implied: $a_ib_i \equiv {\bf a}{\bf
b}$.
%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}
\centerline{\psfig{figure=zlaser3.eps,width=8cm}}
\vspace*{2mm}
\caption{Schematic illustration of how electrons are injected into the
active region
for the case that the adjacent ferromagnets have different magnetization
directions (shown with long arrows; electron spins, $\vec{\sigma}$, are
shown
with short arrows with black dots at the end).
An electron from the left-side ferromagnet with its spin parallel to
the
magnetization direction passes through a sharp boundary (shown as a
vertical
thick line) between the ferromagnets without changing spin direction.
In the right-side
ferromagnet it emerges with its spin
in a superposition of the spin-up and spin-down states. The classical
precession of the spin is indicated by a dotted ellipse.}
\label{injection}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%
If the injected electrons are prepared in such a way that their spins
are not
parallel to the magnetization in the active region B (see Fig.\ref{population}),
the Hamiltonian
(\ref{effham})
produces spin-flips and hence stimulates the needed radiative
transitions of hot
electrons in the upper band to the lower energy band. This process is
illustrated in Fig.\ref{injection}, where an electron
(with its spin parallel to the magnetization) is impinging on the
boundary
from the left, passes through the boundary and is scattered into a
quantum superposition of spin-up and spin-down states in the active
region to the right of the boundary. The wave
function of the electron in the active region B is
%
\begin{equation}
\Psi ({\bf r}) =
e^{i{\bf p}_\perp{\bf r}_\perp}\sum_{k=1}^2 a_k \left( \begin{array}{c}
1 \\
I_+/(I_z +(-1)^k I) \end{array}\right)e^{i p_k x}
\label{wave}
\end{equation}
%
where the $x$-axis is perpendicular and the $y$- and $z$-axes are parallel to the
boundary; the projections of the electron momentum and the coordinate on the
boundary are ${\bf p}_\perp = (0, p_y,p_z)$ and ${\bf r}_\perp
=(0,y,z)$, while
$p_{1,2} = \sqrt{2m*(E\mp I) -p_\perp^2 }$ and $I_+ = I_x + i I_y$;
coefficients
$a_{1,2}(E)$ are found by matching the wave functions of the electron in
the active region (the right-hand side of Fig.~\ref{injection}) and in
the
injecting region (left-hand side of Fig.~\ref{injection}) at the
boundary
(we do not present their explicit form here).
Using Eq.(\ref{wave}) as the initial proper state $\Psi_i ({\bf r})$
belonging to the initial energy $E_i$ and the final state $\Psi_f ({\bf
r})$
belonging to the final energy $E_f$ one sees that matrix element
$\left<\Psi_f ({\bf r})\right|\hat{H}_{eff}^{(1)}\left|\Psi_i ({\bf
r})\right>$
(the probability amplitude for a radiative electronic transition between
the unperturbed
energy bands) is not zero if ${\ p}_1(E_i)={\ p}_2(E_f)$. From here it
follows
that the difference between the initial and the final energies should be
$E_i({\bf p}) -E_f({\bf p})= 2I({\bf p})$.
%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}
\centerline{\psfig{figure=zlaser2.eps,width=8cm}}
\vspace*{2mm}
\caption{Schematic illustration of a population inversion in a magnetic
conductor
(region B).
A bias voltage $V$ is applied between magnetic conductors
($A$ and $C$)
which have opposite directions of their magnetizations (the thick
vertical lines represent
potential barriers). Arrows show the
electron spin
directions in the electron energy bands; $\mu_{0}$ is the equilibrium
chemical potential
(in the absence of bias). Population inversion requires the bias voltage
to be $V > (2 I - \mu_0)/e$. The insertion shows a possible realization
of a structure
in which two magnetic conductors with opposite magnetization directions
are coupled through a microbridge (B).}
\label{population}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%
Taking such wave-functions as the initial and the final states and
performing standard calculations \cite{Agrawal} using
Eq.(\ref{effham}) and Fermi's golden rule one gets the stimulated
transition rate per unit volume of active material as
%
\begin{eqnarray}
R_{st} = \frac{32\pi e^2 \mu}{ n^2}\frac{|{\bf I}^\prime|^2}{\omega
}\frac{ \hbar \nu_s \left( b_1 N_\uparrow - b_2 N_\downarrow\right)
}{\left(\hbar \omega -2I \right)^2 + \left(\hbar \nu_s\right)^2}
\label{ratesf1}N_{p}\\
\times \left\{
\left(
\left[\mbox{\boldmath$ \epsilon$}_a [\mbox{\boldmath$ \epsilon$}_{dr}
\cdot \mbox{\boldmath$ \epsilon$}_a]\right]\cdot \mbox{\boldmath$
\epsilon$}_{i}
\right)^2 +
\left(
\left[\mbox{\boldmath$ \epsilon$}_{dr} \cdot \mbox{\boldmath$
\epsilon$}_a\right]\cdot \mbox{\boldmath$ \epsilon$}_{i}
\right)^2
\right\} \nonumber
\end{eqnarray}
%
%
Here $\mu$ and $n$ are the magnetic permeability and the refractive
index of
the medium, respectively, $N_p$ is the photon density, $\omega$ is the
photon
frequency, the constants $b_1 $ and $b_2$ are of order unity,
%(they are
%determined by the energy dependence of $a_{1,2}$ in Eq.(\ref{wave})),
the unit vectors $\mbox{\boldmath$ \epsilon$}_a$
and
$\mbox{\boldmath$ \epsilon$}_{i}$ are directed along the
magnetizations
in the active (right-hand side of Fig.\ref{injection}) and injection
(left-hand side of Fig.\ref{injection}) regions, respectively, while
$\mbox{\boldmath$ \epsilon$}_{dr}$ is parallel to the vector
${\bf I}^\prime \equiv e_i \partial {\bf I}/\partial p_i$, where
${\bf e}$ is the unit polarization vector in the direction of
the vector potential ${\bf A}$.
Eq.(\ref{ratesf1}) was derived under the assumption that
$|I^\prime| p_{F\uparrow} < \hbar \nu_s$ ($p_{F\uparrow}$ is the Fermi
momentum of electrons in the upper band), that is the additional
dispersion
caused by the dependence of $I$ on the electron momentum ${\bf }$ (see
Eq.(\ref{implicit}))
is smaller than the broadening of the electron energy due to spin-flip
processes.
In the opposite limiting case $|I^\prime| p_{F\uparrow} \gg \hbar
\nu_s$
one has
%
\begin{eqnarray}
R_{st}= \frac{32 \pi}{\hbar \omega}\frac{e^2 \mu}{n^2} |{\bf
I}^{'}|\left( b_3 N_\uparrow^{2/3} - b_4 N_\downarrow^{2/3}\right)
\label{ratesf2}N_{p}\\
\times \left\{
\left(
\left[\mbox{\boldmath$ \epsilon$}_a [\mbox{\boldmath$ \epsilon$}_{dr}
\cdot \mbox{\boldmath$ \epsilon$}_a]\right]\cdot \mbox{\boldmath$
\epsilon$}_{i}
\right)^2 +
\left(
\left[\mbox{\boldmath$ \epsilon$}_{dr} \cdot \mbox{\boldmath$
\epsilon$}_a\right]\cdot \mbox{\boldmath$ \epsilon$}_{i}
\right)^2
\right\} \nonumber
\end{eqnarray}
%
where the constants $b_{3,4} \sim 1$.
One of the necessary conditions for the lasing effect to be realized is
(see, e.g., \cite{Agrawal}):
\begin{eqnarray}
R_{st} =\nu_p N_p \,,
\label{lasing}
\end{eqnarray}
where $\nu_{p}$ is the photon relaxation rate.
We consider the case that damping of
electromagnetic waves is mainly due to
absorption by free charge carriers, the frequency of the photon
relaxation
being $\nu_{p}= 2 k \omega /n$ (see \cite{Blatt}).
For estimating the parameters of the problem we use standard formulae
for the refractive index $n$ and the absorption coefficient $k$ for
metals
subject to electromagnetic fields (see, e.g. \cite{Ziman}).
One finds that
\begin{eqnarray}
\nu_{p}= \frac{\mu}{n^2} \frac{4 \pi \sigma (0)}{ 1+(\omega t_0)^2} \,.
\label{2nk}
\end{eqnarray}
where $\sigma_0 $ is the static conductivity of the conductor, $t_0$ is
the transport electron relaxation time.
Using Eq.(\ref{ratesf1}) and Eq.(\ref{2nk}) one can
rewrite
Eq.(\ref{lasing}) as
\begin{eqnarray}
\frac{\left(N_\uparrow - N_\downarrow\right)}{N_\uparrow+N_\downarrow}
\approx\frac{\hbar \nu_s}{2 I}\frac{p_0^2/m*}{\hbar t_0^{-1} }
\frac{1}{1+(\omega t_0)^2} \,,
\label{lasing1}
\end{eqnarray}
where $p_0 =\hbar/a \sim 10^{-19}$ erg sec/cm.
It seems that for achieving the lasing effect the most favorable
materials are
ferromagnetic manganese perovskites with nearly 100\% spin polarization
of
the conduction electrons (\cite{Tokura,Park,Tokura2}). The high degree
of polarization
of the carriers permits the creation of a population inversion of the
energy bands
in the active region B (see Fig.\ref{population}).
%On the other hand, the
%width of
%the domain walls in these materials is $\delta \approx 4\times 10^{-6}$
%cm
%(see \cite{Mathur}), while the
%the "magnetic rotation length" $L_I = \sqrt{\hbar D/I } \sim 10^{-5}$
%cm.
%(for the experimental mean free path $l_0 = 1.4 \time 10^{-7}$ cm and
%the Fermi
%velocity $\approx 10^{8}$ cm/s, see \ref{Tokura2}).
%It means that electrons
%pass the domain wall nearly without changing the direction of its spin
%as the
%probability amplitude for the spin to flip is $\sim \delta/L_I \approx
%10^{-1}$.
%That is advantageous for injection of electrons in the active region
%(see Fig.\ref{injection}).
Here and below we use
experimental
values of the needed parameters:
the mean free path $l_0 = 1.4 \times 10^{-7}$ cm, the Fermi velocity
$v_F \approx 10^{8}$ cm/s, $t_0 \approx 10^{-15} $ s and $m* = 0.3 m_e$
where
$m_e$ is the free electron mass, the number of carriers $\approx 3.4
\times 10^{21}$,
the resistivity $\rho \sim 10^{-4}\div 10^{-3}$ (\cite{comment}).
Inserting these values into Eq.(\ref{lasing1}) one finds
the lasing condition to be
\begin{eqnarray}
\frac{N_\uparrow -N_\downarrow}{N_\uparrow+N_\downarrow} \approx 5
\frac{\hbar \nu_s}{I}
\nonumber
\end{eqnarray}
For the case $|I^\prime| p_{F\uparrow} \gg \hbar \nu_s$
(see Eq.(\ref{ratesf2})), the lasing
condition
Eq.(\ref{lasing}) is
\begin{eqnarray}
\frac{N_\uparrow^{2/3} -
N_\downarrow^{2/3}}{N_\uparrow+N_\downarrow}\approx 0.5 \times 10^{-7}
\,{\rm cm}
\nonumber
\end{eqnarray}
From these equations it follows that the lasing condition $R_{st}=\nu_s
N_p$ is easily
satisfied since one needs
$\hbar \nu_s/I $ to be less than $10^{-1}$ while the theoretical
estimate of the spin
relaxation rate $\nu_s$ gives the value $10^{-2}$ for this ratio.
Estimations based on the above experimental values of the
parameters
show the optical gain to be $g_{opt} =(n/c)R_{st} \sim 10^7$
cm$^{-1}$
and the threshold current density to be
$j_{th} = e l \nu_s N_\uparrow \sim 10^7 \div10^8 $ A/cm~$^{-2}$ if
the length of
the active region is $l = 10^{-5}$ cm. Estimations for $N_e \sim 10^{18}$
cm~$^{-3}$ shows
the optical gain and the threshold current to be $g_{opt} \sim 10^3
\div 10^4
$ cm~$^{-1}$
and $j_{th} \sim 10^5$ A/cm~$^{-2}$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We predict an extremely large optical gain in systems with a high
density of charge carriers. The price to be paid
for the gain exceeding what can be achieved in semiconductors by 3 or
4 orders
of magnitude is the high currents needed for an effective tunneling
pumping of
the system. The current value $j = 10^6 \div 10^8 $ A/cm$^2$ seems to be
very large for homogeneous bulk metals because of the accompanying Joule
heating.
Special measures are needed to avoid heating the active, lasing region.
One solution to that problem is to arrange for the current injection to
be
inhomogeneous in space. This can be achieved
if the magnetic conductors are electrically connected through a
point contact. The spreading out of the current
far from
the narrow point contact provides for an efficient dissipation of heat
\cite{Yanson}.
A current density $j \sim 10^8$ A/cm$^2$ can be reached without
significant heating of the
contact region. On the other hand,
the extremely large optical gain $g_{opt} \sim 10^7$ cm$^{-1}$ means
that
it is enough to have a small volume of active lasing region.
Such a structure can be prepared on the basis of the technique suggested
in Ref.~\cite{Zdravko} for fabrication of biepitaxial films of
$La_{0.7}Sr_{0.3}MnO_3$
with
$45^\circ$
in-plane rotated domains.
%This technique makes it possible to fabricate
%a chain of alternating
%square domains of $La_{0.7}Sr_{0.3}MnO_3$, all adjacent domains
%touching
%corner to corner square domains of the same material but with $45^\circ$
%in-plane rotated crystal axes.
% On the base of the technique
%suggested in paper \cite{Zdravko} for preparation of a biepitaxial film
%with $45^\circ$
% one can fabricate an array of domains of $La_{0.7}Sr_{0.3}MnO_3$
%as is schematically shown in Fig.\ref{chess}.
In summary we have proposed a new principle for a compact solid-state
laser working
in the 1-100 THz regime.
The proposed laser is based on a new mechanism for creating spin-flip
processes
in ferromagnetic conductors. The mechanism is due to the interaction of
light
with conduction electrons; the interaction strength, being proportional
to
the large exchange energy, exceeds the Zeeman interaction by orders of
magnitude.
On the basis of this interaction, a giant lasing effect was predicted
for systems where a population inversion can be created by tunneling
injection of
spin-polarized electrons from one ferromagnetic conductor to another ---
the
magnetization of the two ferromagnets having different orientations.
Using experimental data for ferromagnetic manganese perovskites with
nearly
100\% spin polarization we show the laser frequency to be in the range
%$10^{12} \div10^{14}$ s$^{-1}$. We also show
1-100 THz.
The optical gain is estimated to be
%$g_{opt} \sim
of order $10^7$ cm~$^{-1}$, which exceeds
the gain of conventional semiconductor lasers by 3 or 4 orders of
magnitude.
An experimental study based on a point contact geometry to avoid heating
by the necessarily large injection currents was proposed and discussed.
Acknowledgements.
We thank L.Y. Gorelik, V. Kozub, G.D. Mahan and R. Gunnarsson for
useful discussions.
Financial support from the Royal Swedish Academy of
Sciences
(AK) is gratefully acknowledged.
%%%%%%%%%%%%%%%%%%%%%%%
%\begin{figure}
%\centerline{\psfig{figure=zlaser4.eps,width=8cm}}
%\vspace*{2mm}
%\caption{~Schematic~picture~ of a chain array of
%$La_{0.7}Sr_{0.3}MnO_3 $
%domains. The arrows show directions of the magnetization in the
%domains}
%\label{chess}
%\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%
\begin{thebibliography}{99}
\bibitem{Tokura} "Collosal Magnetoresistive Oxides" edited by
%%%Yoshinori
Y.~Tokura (Gordon and Breach Science Publishers, Amsterdam,
2000).
\bibitem{Prinz} G.A. Prinz, Science, {\bf 282}, 1660-1663 (1998).
\bibitem{Wolf} S.A. Wolf {\em et al.},
%%%, S.A. and Awschalom, D.D. and Buhrman, R.A. and Daughton, J.M.
%%% and von Molnár, S. and Roukes, M.L.
%%% and Chtchelkanova, A.Y. and Treger, D.M.,
Science {\bf 294}, 1488 (2001).
\bibitem{absorpion} The main mechanism for relaxation of hot electrons is
the spin-flip process caused by emission of spin-waves.
\bibitem{Zeiger} H.J. Zeiger and G.W. Pratt, "Magnetic Interactions In
Solids", Clarendon Press, Oxford, 1973.
\bibitem{spin-orbit} The Hamiltonian Eq.(\ref{effham}) gives an
effective
"spin-orbit" interaction in magnetic films which are usually under a
considerable strain
caused by the substrate and grain boundaries (see, e.g., \cite{Wu,Soh} and
references therein).
If $p a/\hbar \ll 1$, for bulk materials $\partial{\bf I}/\partial p_i
=0$, one should take the next term in the expansion,
that is change
$\partial{\bf I}/\partial p_i$ to $ p_k \partial^2{\bf I}/\partial p_k
\partial p_i$ in Eq.(\ref{effham}).
\bibitem{Wu} Yan Wu, Y. Suzuki, U. Rudiger, J Yu, and A.D. Kent,
Appl. Phys. Lett. {\bf 74}, 2295 (1999).
\bibitem{Soh} Yeong-Ah Soh {\em et al.},
%%% , P.G. Evans, Z. Cai, B. Lai, C.-Y. Kim, G. Aeppli,
%%% N.D. Mathur, M.G. Blamire, E.D. Isacs,
J. Appl. Phys. {\bf 91}, 7742 (2002).
\bibitem{Agrawal} G.P. Agrawal and N.K. Dutta "Long-wavelength
semiconductor lasers", Van Nostrand Reinhold Company, New York, 1986;
\bibitem{Blatt} F.J. Blatt "Physics of electronic conduction in
solids", McGraw-Hill Book Company, 1968.
\bibitem{Ziman} J.M. Ziman "Principles of the theory of solids",
Cambrige, 1964.
\bibitem{Park} J.-H. Park {\em et al.},
%%% E. Vescovo, H.-J. Kim, C. Kwon, R. Ramesh and T.Venkatesan,
Nature, {\bf 392}, 794 (1998).
\bibitem{Tokura2} Y.-D. Chuang {\em et al.},
%%%, A.D. Gromko, D.S. Dessau, T. Kimura, Y. Tokura,
Science, {\bf 292}, 150 (2001).
\bibitem{comment}
The values are taken for a compound similiar to the
perovskites manganites found to be nearly 100\% spin-polarized, see
\cite{Tokura2}.
%%%\bibitem{Mathur} S.J. Lloid, N.D. Mathur, J.C. Loudon, and P.A. Midgley,
%%%Phys. Rev. B {\bf 64}, 172407 (2001)
\bibitem{Yanson}
A.V. Khotkevich, I.K. Yanson "Atlas of point contact spectra of
electron-phonon interactions in metals", translation by R.C.
Reinertson, Kluwer Academic Publishers, Boston/Dordrecht/London,
1995.
\bibitem{Zdravko}
R. Mathieu, P. Svedlindh, R.A. Chakalov, and
Z.G.~Ivanov, Phys. Rev. B, {\bf 62}, 3333 (2000).
\end{thebibliography}
\end{document}