Dilute Magnetic

Semiconductor
Introduction
Ellirassi et al. SAA 186 (2017) 120-131

ZnO

Hexagonal wurtzite
Zn

Zinc blende

Dilute Magnetic Semiconductor (DMS)

• a small fraction of the host semiconductor
sites are substituted by transition metal Rocksalt
dopants
• a new generation devices, i.e. spintronic
 Spintronics
Spintronics – material
– material aspects
aspects
Why to do not combine complementary properties and
functionalities of semiconductor and magnetic material
systems?
• hybrid structures
-- overlayers or inclusions of ferromagnetic metals =>
source of stray fields and spin-polarized carriers
-- soft ferromagnets => local field amplifiers
-- hard ferromagnets => local field generators

• ferromagnetic semiconductors
Families of magnetic
semiconductors
 Magnetic
Magnetic semiconductors
semiconductors
• magnetic semiconductors
short-range ferromagnetic super- or double
exchange
EuS, ZnCr2Se4, La1-xSrxMnO3, ...

short-range antiferromagnetic superexchange

EuTe, ...
 DMS:
DMS:standard
standard semiconductor
semiconductor ++
magnetic ions
magnetic ions

• Various magnetic ions:

- mostly 3d transition metals: Sc, ..., Cu
- rare earth (4f): Ce, ..., Tm
- also actinides (5f), 4d TM, ...
• Various hosts:
- II-VI: Cd1-xMnxTe, Hg1-xFexSe,...
- IV-VI: Sn1-xMnxTe, Pb1-xEuxS
- III-V: In1-xMnxSb, Ga1-xErxN, ...
- IV: Ge1-xMnx, Si1-xCex
- ....
 Most
MostofofDMS:
DMS: random
random antiferromagnet
antiferromagnet

short range antiferromagnetic superexchange

Example 1: (Zn,Mn)O
 Wang et al., Appl. Phys. Lett. 102 (2013) 102112

Introduction Shao et al., J. App. Phys. 115 (2014) 153902

Mukherjee et al., Phys. Rev. B 81 (2010) 205202
Lawes et al., Phys. Rev. B 71 (2005) 045201

ZnO r Zn2+ = 0.74 Å Mn-substituted ZnO (MZO):

r Mn2+ = 0.83 Å
▪ The optical properties can be tuned by changing
the magnetic moment
▪ exhibits a giant negative magnetoresistance
Zn ▪ Mn ion is isovalent to a Zn ion → control the
concentration of charge carriers independently
O
of that of spins
n-type semiconductor: MZO thin film
ZnO
• Wide direct band gap (3.37 eV) Mn substitution Ferromagnetic
(diamagnetic)
• Large exciton binding energy (60 Paramagnetic
meV) Diamagnetic

Effective carrier densities are the key in the magnetism of MZO MZO Nanoparticles?

ICMIA, SOLO, 21 NOV. 2019 9

Sample preparation
Coprecipitation Method H. Harsono et al. J. Nano Research 35 (2016) 67-76

• Low calcination-temperature (350°C)

• Single phase ZnO (hexagonal wurtzite)
• Impurity phase of ZnMn2O4 presences
for x  0.10
• The lattice parameters of a=b and c
slightly increase by the Mn-substitution
TEM image
XRD of MZO1
patterns of MZOnanoparticles
samples
• Nano sized (Mn,Zn)O particles

ICMIA, SOLO, 21 NOV. 2019 10

 Measurements- X-ray Absorption Spectroscopy (XAS)
Beamline specification
Radiation source
Focusing optics
Monochromator
Multipole wiggler
Toroidal focusing mirror
Double crystal monochromator using
Si(111)

Available photon energy range 4 -18 keV

Photon flux 107- 1010 ph/s @electron current 100 mA
Energy resolution 10-4
Beam size @ sample position 5 mm x 25 mm

BL1.1W To examine the oxidation state, symmetry, local coordination environment

https://www.slri.or.th/en/bl1-1w.html
ICMIA, SOLO, 21 NOV. 2019 11
Measurements- X-ray Absorption Spectroscopy (XAS)
Two regions:
1. The x-ray absorption near-edge structure (XANES)
2. The extended x-ray absorption fine structure (EXAFS)

Standard samples Analysis:

Sample Oxidation state
Mn foil 0 (MZO samples)
MnO 2+
Mn2O3 3+
MnO2 4+ https://bruceravel.github.io/demeter/
https://www.slri.or.th/en/bl1-1w.html
ICMIA, SOLO, 21 NOV. 2019 12
Measurements- Magnetization
SQUID Magnetometer (Quantum Design MPMS XL)
▪ The measurements were performed at 300 K
▪ Under applied magnetic fields in the range of -10 kOe to 10 kOe

The samples we study:

• (Mn0.03Zn0.97)O → MZO3
• (Mn0.05Zn0.95)O → MZO5
• (Mn0.07Zn0.93)O → MZO7

ICMIA, SOLO, 21 NOV. 2019 13

 XAS Data – Mn K-edge XANES spectra
Sample Pre-edge (eV) Absorption
edge/E0 (eV)
Mn-foil 6529.12 6539.00
MnO 6539.09 6543.61
Mn2O3 6539.11 6548.35
MnO2 6540.24 6552.20
MZO3 6539.07 6549.37
MZO5 6539.09 6549.45
MZO7 6539.47 6551.77

• The E0 is larger than that in Mn foil, MnO and Mn2O3, but it is

• The pre-edge energy is 6539 eV in all MZO
smaller than the E0 in MnO2
samples
• The oxidation states of MZO samples is between 2+ and 4+
ICMIA, SOLO, 21 NOV. 2019 14
XANES – Oxidation state
• The obtained oxidation state:
MZO3 : +3.12
MZO5 : +3.15
MZO7 : +3.85

• The Mn ion in MZO samples has a mixed

valence state of Mn3+/Mn4+

• The oxidation state of MZO tend to increase

with increasing Mn concentration.

(Interpolation of E0 to the oxidation state)

ICMIA, SOLO, 21 NOV. 2019 15

 XAS Data - EXAFS
Sample Interatomic Coordination Debye-Waller R-factor
distance Mn-O number factor (Å)
(Å)
MZO3 1.90 3.9 0.005 0.022
MZO5 1.88 2.6 0.003 0.019
MZO7 1.98 2.6 0.003 0.008

• The uncertainty between all fitting parameters is quite small

• the crystalline irregularity due to the Mn substitution is (The fit curve of Mn K-edge EXAFS
spectrum of MZO7)
relatively small
• The interatomic distance (R) of Mn-O is approximately 1.9
Å without phase correction → the enhancement in the amount
• The CNs are smaller than the ideal CN of 4 for Mn of oxygen vacancy (VO)

ICMIA, SOLO, 21 NOV. 2019 16

Magnetization
• The parent compound (ZnO) shows a diamagnetic
feature
• MZO exhibits paramagnetic nature
• M(H) of MZO5 and MZO7 slightly bend at the
vicinity of H= 0 kOe
A small superparamagnetic contribution coexists
within a strong paramagnetic background
• At a constant H, the M(H) increases with the increase
in the Mn concentration

Magnetism and oxygen vacancies?

(The isothermal magnetization/M(H) of MZO)

ICMIA, SOLO, 21 NOV. 2019 17

 Magnetization
Zn
O

Indirect double-exchange model in • Nanoparticles provide a unique platform to

Cu-doped ZnO thin films generate VO
• Mn ion is the source of magnetic moment
• The amount of VO also enhances by the
doping, thus affects magnetic properties of
MZO
• The VO plays a role to mediate the hopping
of doped electrons between different Mn
states
Double-exchange effect mediated by Mn magnetic Oxygen
moment vacancy (VO)
the large VO orbitals
Herng et al., Phys. Rev. B 105 (2010) 207201
ICMIA, SOLO, 21 NOV. 2019 18
Summary
❖ The oxidation state of MZO nanoparticle increases by the Mn-substitution
❖ MZO exhibits a mixed valence state of Mn3+/Mn4+
❖ The coordination number (CN) of Mn changes from its ideal CN of 4
❖ The CN number of MZO reduces by increasing Mn concentration, which indicates
the increase in the amount of oxygen vacancy (VO)
❖ The magnetization of MZO increases as the Mn concentration increases
❖ The magnetism of MZO is suggested to be related to the correlation between the
Mn magnetic moment and VO.

ICMIA, SOLO, 21 NOV. 2019 19

Example 2:
(Ga,Mn)As
 Hole concentration vs. Mn fraction
in a diluted (Ga,Mn)As ferromagnetic
semiconductor
Raimundo R dos Santos (IF/UFRJ),
Luiz E Oliveira (IF/UNICAMP) e
J d’Albuquerque e Castro (IF/UFRJ)

Apoio:

Esta apresentação pode ser obtida do site

http://www.if.ufrj.br/~rrds/rrds.html
seguindo o link em “Seminários, Mini-cursos, etc.”
 Layout
• Motivation
• Some properties of (Ga,Mn)As
• The model: hole-mediated mechanism
• New Directions
 Motivation
Spin-polarized electronic transport
 manipulation of quantum states at
a nanoscopic level
 spin information in semiconductors

Metallic Ferromagnetism:
Interaction causes a relative
shift of  and  spin
channels
Spin-polarized device principles (metallic layers):

Parallel magnetic layers

  spins can flow

Antiparallel magnetic layers

  spins cannot flow

[Prinz, Science 282, 1660 (1998)]

 Impact of spin-polarized devices:
• Giant MagnetoResistance heads ( ! )  US$ 1 billion
• Non-volatile memories ( ? )  US$ 100 billion

GMR RAM’s

Magnetic Tunnel Junction

 Injection of spin-polarized carriers plays important role in
device applications

 combination of semiconductor technology with

magnetism should give rise to new devices;
 long spin-coherence times (~ 100 ns) have been
observed in semiconductors
Magnetic semiconductors:
• Early 60’s: EuO and CdCr2S4
 very hard to grow
• Mid-80’s: Diluted Magnetic Semiconductors
II-VI (e.g., CdTe and ZnS) II → Mn
 difficult to dope
 direct Mn-Mn AFM exchange interaction
PM, AFM, or SG (spin glass) behaviour
 present-day techniques: doping has led to FM for T < 2K
IV-VI (e.g., PbSnTe) IV → Mn
 hard to prepare (bulk and heterostructures)
 but helped understand the mechanism of carrier-mediated FM
• Late 80’s: MBE → uniform (In,Mn)As films on GaAs substrates:
FM on p-type.
• Late 90’s: MBE → uniform (Ga,Mn)As films on GaAs substrates:
FM; heterostructures
Spin injection into a FM semiconductor heterostructure

polarization of
emitted
electrolumiscence
determines spin
polarization of
injected holes

[Ohno et al., Nature 402, 790 (1999)]

 Some properties of (Ga,Mn)As

Ga: [Ar] 3d10 4s2 4p1

Mn: [Ar] 3d5 4s2

Photoemission
 Mn-induced hole states have 4p
character associated with host
semiconductor valence bands

EPR and optical expt’s

 Mn2+ has local moment S = 5/2

[For reviews on experimental data see, e.g., Ohno and Matsukura, SSC 117,
179 (2001); Ohno, JMMM 200, 110 (1999)]
Phase diagram of MBE growth

[Ohno, JMMM 200, 110(1999)]

Regions of Metallic or Insulating behaviours depend on

sample preparation (see later)
 x = 0.035

Open symbols: B in-plane

• hysteresis  FM with easy axis
in plane;
• remanence vs. T  Tc ~ 60 K

x = 0.053

Tc ~ 110 K

[Ohno, JMMM 200, 110(1999)]

Resistance measurements on
samples with different Mn
concentrations:
Metal → R  as T 
Insulator → R  as T 
 Reentrant MIT

[Ohno, JMMM 200, 110(1999)]

 Question: what is the hole concentration, p?

Difficult to measure since RHall dominated by the magnetic

contribution; negative magnetoresistance (R  as B )
Typically, one has p ~ 0.15 – 0.30 c , where c = 4 x/ a03,
with a0 being the AsGa lattice parameter
• only one reliable measurement: x = 0.053  3.5 x
1020 cm-3

Defects are likely candidates to explain difference between p and c:

• Antisite defects: As occupying Ga sites
• Mn complexes with As
Our purpose here: to obtain a phenomenological
relation p(x) from the magnetic properties
The model: hole-mediated mechanism
Interaction
between hole spin
and Mn local
moment is AFM,
giving rise to an
effective FM
coupling between
Mn spins
[Dietl et al., PRB 55, R3347 (1997)]
= Mn, S =5/2

= hole, S =1/2 (itinerant)

Simplifying the model even further:
• neither multi-band description nor spin-orbit → parabolic
band for holes
• hole and Mn spins only interact locally (i.e., on-site) and
isotropically – i.e., Heisenberg-like – since Mn2+ has L = 0
• no direct Mn-Mn exchange interactions
• no Coulomb interaction between Mn2+ acceptor and holes
• no Coulomb repulsion among holes → no strong correlation
effects
• ... 0

hole
Mn
Mean-field approximation
Nearly free holes moving under a magnetic field, h, due to the Mn
moments:

 2  
−  − h  k =  k  k ,  = 1
2

 2m * 2 

Hole sub-system is polarized: m = m I = n (R I ) − n (R I )

Pauli paramagnetism:
1
m  p 3h
Now, the field h is related to the Mn magnetization, M :

h = J pd  (r − R I ) M (R I ) = J pd Mc
I

Assuming a uniform Mn concentration

Mn magnetization

We then have
1
m = A J pd M x p 3

A depends on m* and on several

constants
The Mn local moments also feel the polarization of the holes:

 J pd S  
M = nMn g BM = nMn g B SBS   m 
 2k BT  
Brillouin function 1
m = A J pd M x p 3

Linearizing for M → 0, provides the self-consistency

condition to obtain Tc:
Setting S = 5/2, we can write an expression for p(x):

Now, there are considerable uncertainties in the experimental

determination of m* and on Jpd [e.g., 55 10 to 15040 meV nm3].

But, within this MFA, these quantities appear in a specific

combination,
m * J 2pd

which can then be fitted by experimental data.

In most approaches x (c or n) and p are treated as independent
parameters

[Schliemann et al., PRB 64, 165201 (2001)]

Fitting procedure

• Only reliable estimate for p is 3.5  1020 cm-3, when x = 0.053.

• For this x, one has Tc = 110 K
• We get

−2 3 2
( m * me ) J pd = 2.4  10 eV nm
2


Results for p (x):

Note approximate linear behaviour for Tc(x) between x = 0.015-0.035

 p(x) constant in this range

We then get

Notice maximum of
1h/Mn
p(x) within the M phase
 correlate with MIT

Early predictions

log!

[Matsukura et al., PRB 57, R2037 (1999)]

Assume impurity band:

F  p1/3, increases to the right, towards VB

(a) Low density: unpolarized holes, F below mobility edge

(b) Slightly higher densities: holes polarized, but F is still below
the mobility edge
(c) Higher densities: F reaches maximum and starts decreasing,
but exchange splitting is larger  still metallic
(d) Much higher densities: F too low and exchange splitting too
small  F returns to localized region
Picture supported by recent photoemission studies

[Asklund et al., cond-mat/0112287]

Magnetiztion of the Mn ions

1. Maxima decrease as T
increases

2. Operational “window”
shrinks as T increases

Simple model is able to: predict p(x); discuss MIT; M(x)

[RRdS, LE Oliveira, and J d’Albuquerque e Castro, JPCM (2002)]
 New directions
I. New Materials/Geometries/Processes
1. Heterostructures
(Ga,Mn)As/(Al,Ga)As/(Ga,Mn)As  spin-
dependent scattering, interlayer coupling,
and tunnelling magnetoresistance
2. (InyGa1-y)1-x MnxAs has Tc ~ 120 K,
apparently without decrease as x increases
3. (Ga,Mn) N has Tc ~ 1000 K !!!!!
4. Effects of annealing time on (Ga,Mn)As
 250 oC annealing
 Tc grows with annealing time, up to
2hrs; for longer times, Tc decreases
 M as T → 0 only follows T 3/2
(usual spin wave excit’ns) for
annealing times longer than 30min

 All samples show metallic

behaviour below Tc
 xx decreases with annealing time,
up to 2 hrs, and then increases again

[Potashnik et al., APL (2001)]

 Two different regimes of
annealing times (~2 hrs):
• FM enhanced
• Metallicity enhanced
• lattice constant suppressed
 changes in defect
structure:
• As antisites and
correlation with Mn
positions?
• Mn-As complexes?

More work needed to ellucidate nature of defects and

their relation to magnetic properties
II. Improvements on the model/approximations
1. Give up uniform Mn approximation
 averaging over disorder configurations
(e.g., Monte Carlo simulations)
2. More realistic band structures
3. Incorporation of defect structures
4. Correlation effects in the hole sub-system
[for a review on theory see, e.g., Konig et al., cond-mat/0111314]