Photoemission study of in diluted magnetic semiconductor...

Photoemission study of

room-temperature

ferromagnetism

in diluted magnetic

semiconductor

ZnGeP2:Mn

Master Thesis

Yukiaki Ishida

Department of Physics, University of Tokyo

January, 2003

Contents

1 Introduction 5

2 Physical properties of ZnGeP2:Mn 13

2.1 Ternary chalcopyrite semiconductor ZnGeP2 . . . . . . . . . 13

2.2 Novel II-IV-V2 chalcopyrite-based DMS . . . . . . . . . . . . 15

2.3 Synthesis of II-IV-V2:Mn . . . . . . . . . . . . . . . . . . . . 17

2.4 Magnetization of II-IV-V2:Mn and bulk Zn1−xMnxGeP2 . . . 18

3 Results and discussions 23

3.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.2 Photoemission spectra of ZnGeP2 substrate . . . . . . . . . 25

3.3 Mn deposition on ZnGeP2 . . . . . . . . . . . . . . . . . . . 25

3.4 Depth profile of ZnGeP2:Mn interface . . . . . . . . . . . . . 30

3.5 ZnGeP2:Mn prepared at

low temperatures . . . . . . . . . . . . . . . . . . . . . . . . 33

3.6 Photoemission spectra of MnP . . . . . . . . . . . . . . . . . 37

3.7 Magnetization measurements of

ZnGeP2:Mn . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4 Summary 43

4 Contents

Chapter 1

Introduction

The remarkable development currently made in the field of semiconductor

physics and nanotechnology is the manipulation of spins. During the past

decade, various types of spin polarization and its control have been made:

carrier spins, spins in magnetic ions, spins in superlattices, and even nuclear

spins, which were either injected, transported, changed of their spin states,

or exploited of their unique quantum nature [1.1, 2, 3, 4, 5]. The rapid devel-

opment of these spin manipulation is strongly associated with the progress

of semiconductor crystal growth and superstructure fabrication techniques.

In the field of crystal growth, one of the major impact was the develop-

ment of molecular beam epitaxy (MBE). MBE was a powerful technique for

the fabrication of diluted magnetic semiconductors (DMSs). This method,

which is a non-equilibrium growth technique, has allowed the incorporation

of transition and rare-earth metals into the host semiconductors beyond its

solubility limit [1.6, 7], that is, synthesis of products which were formerly

forbidden in the bulk has become possible. This led to the discovery of

III-V-based In1−xMnxAs [1.8] and Ga1−xMnxAs [1.9] films, which were the

first DMSs showing ferromagnetism. Furthermore, MBE has made possible

to prepare heterostructures of DMSs and semiconductors, allowing one to

explore over quasi-two-dimensional quantum wells of DMSs, as well as to

introduce artificial strain in layered DMSs.

DMS itself is already attractive because of its unique physical prop-

erties. The presence of localized magnetic ions in the semiconductor alloys

leads to the exchange interaction between the electrons constituting the local

moments of magnetic ions and the sp band electrons. This results in a large

Zeeman splittings of electronic (band and impurity) levels, effects of which

are manifested in the giant Faraday rotation [see Fig. 1.1 showing the en-

hanced magnetic circular dichroism (MCD) spectrum of Ga1−xMnxAs com-

pared to nonmagnetic GaAs] and the magnetic-field-induced metal-insulator

6 Chapter 1. Introduction

Figure 1.1: Enhanced magneto-optical effect observed in typical DMS

Ga1−xMnxAs. (a) MCD spectra of nondoped semi-insulating GaAs sub-

strate. (b)(c) MCD spectra of epitaxial Ga1−xMnxAs films at T = 5 K and

H = 1 T. Note that spectrum of GaAs has been magnified by a factor of

ten because of the weak signal (almost two orders weak) compared to that

of Ga1−xMnxAs.

transition. Particularly, the giant Faraday rotation in Cd1−xMnxTe and

Cd1−x−yHgyMnxTe are already utilized in commercial optical isolators.

The advent of the ferromagnetic DMSs In1−xMnxAs and Ga1−xMnxAs

added new possibilities to DMSs. Some newly fabricated devices are shown

in the following figures. Figure 1.2 shows the photoinduced ferromagnetism

realized in (In,Mn)As [1.1]. The photogenerated electron-hole pairs in the

GaSb substrate are split by the internal electrical field, so that the holes ac-

cumulate in the (In,Mn)As film on the surface [Fig. 1.2(a)]. Since (In,Mn)As

exhibits hole-induced ferromagnetism, the InMnAs film, which was formerly

a paramagnet, became a ferromagnet after light exposure [Fig. 1.2(b)(c)].

Another tactic device controlling the magnetic phase by an electrical

field is a field-effect transistor (FET) structure using (In,Mn)As [Fig. 1.3(a)]

[1.4]. The hole concentration in the (In,Mn)As layer is controlled by the

gate voltage. Through the hole-induced mechanism, the ferromagnetic phase

of InMnAs in the vicinity of TC was switched on and off by changing the

gate voltage. Figure 1.4 shows the p-n juction using magnetic semicon-

7

Figure 1.2: Photoinduced ferromagnetism in In1−xMnxAs [1.1]. (b) Magne-

tization curves at 5 K before (open circles) and after (soild circles) the light

irradiation. (c) Hall resistivity before (dashed line) and after (solid line) the

light irradiation. (b) and (c) show hysteritic curves after exposure to the

light.

Figure 1.3: Electric-field control of ferromagnetism in (In,Mn)As [1.4]. (a)

FET device using In1−xMnxAs. Negative bias (VG < 0) increases holes in

(In,Mn)As. (b) Hall resistance (RHall) showing hysterisis under negative

bias.


Figure 1.4: Electrical spin injection in ferromagnetic semiconductor het-

erostructure [1.3]. (a) Spin-polarized holes from p-(Ga,Mn)As (top) and

unpolarized electrons from n-GaAs substrate (bottom) are injected into

the quantum well (hatched region) under forward bias. (b) Electrolumi-

nescence intensity (solid curve) from the quantum well and its polarization

(red curve). The temperature is 6 K.

ductors, which succeeded in the injection of polarized spins into semicon-

ductors. Since the ferromagnetic (Ga,Mn)As is already highly resistive, it

was expected to render effective spin injection into semiconductors when

used as a spin polarizer. In fact, the p-i-n photodiode shown in Fig. 1.4 us-

ing (Ga,Mn)As as a p-type ferromagnetic semiconductor emitted a polarized

light through the recombination of polarized holes and unpolarized electrons

[1.3].

However, there were hampers for the use of the III-V-based ferromag-

netic DMSs, namely, the low Curie temperature (TC) and the difficulty in

controlling the carrier type and its density. As for Ga1−xMnxAs, the max-

imum TC has been ∼ 110 K and materials that show ferromagnetism is

restricted only to p-type doping. Achiving higher TC hopefully above room

temperature with controllable carrier types and densities have been one of

the strong driving forces for the fabrication of new ferromagnetic DMSs.

Particularly for the search of high-TC ferromagnetic DMSs, there has

been an interesting interplay between the theoretical calculations and ex-

periments. Dietl et al. [1.10] predicted, based on the mean field calculation

9

Figure 1.5: Prediction of TC for various DMSs. 2.5 % of Mn atoms in divalent

charge state and 3.5 × 1020 holes per cm3 had been assumed [1.10].

(fig. 1.5), that high TC can be achieved in wide-gap semiconductors such

as ZnO and GaN, Katayama-Yoshida et al. [1.11] have also predicted the

same tendancy from ab initio calculations based on local spin-density ap-

proximation. Motivated by such material designing, there came issues of

the fabrication of new ferromagnetic DMSs, such as ZnO:Co [1.12], ZnO:V

[1.13], and GaN:Mn [1.14], whose TC exceeded room temperature.

An upheaval from the experimental side was the discovery of room-

temperature ferromagnetic DMSs based on ternary chalcopyrite semiconduc-

tors [1.15]. They were a group of interfacial copounds where Mn ions were

densely incorporated in the surface region of the chalcopyrite semiconductor

substrates (cf. Section 2.3). It includes ZnGeP2:Mn, which is the subject of

this thesis. After this discovery of high-TC chalcopyrite-based semiconductor

alloys, there was another report of bulk Zn1−xMnxGeP2, which also showed

room-temperature ferromagnetism [1.16]. Consequently, chalcopyrite-based

DMSs became a hot issue in the field of DMS and attracted many researchers.

In fact, many theoretical calculations followed after these fascinating discov-

eries (see Section 2.2). The charm of the chalcopyrite-based DMSs seems

to root in the inherent structure and chemistry of the ternary chalcopyrite

semiconductors. The novel idea of using the chalcopyrites will be discussed


in Chapter 2, together with the physical properties of chalcopyrite-based

DMSs.

There seems to be little doubt that further advances will be made in

the near future for the high-TC ferromagnetic DMSs, yet, the newly fabri-

cated semiconductor alloys have to undergo the survays whether it can be

strictly categorized as DMSs. One of the properties of DMSs that must be

reminded is the strong sp-d exchange interaction that results in the large

Zeeman splittings, as aready discussed (cf. fig. 1.1) [1.17]. The confirmation

of this exchange interaction in each newly discovered ferromagnetic semi-

conductor alloys is necessary. One also has to be careful of the percipitation

of ferromagnetic clusters in the magnetic-ion-doped semiconductors, which

could possibly be the origin of ferromagnetism. As for ZnGeP2:Mn, the

percipitation of MnP, whose TC is 290K [1.18], is suspected.

This thesis is dedicated to the study of the interface of ZnGeP2:Mn

showing room-temperature ferromagnetism. We adopted photoemission spec-

troscopy and magnetization measurements bu SQUID magnetometer for this

study. PES is a powerful tool to investigate chemical reactions near the sur-

face region. Particulary for studying the interfacial compound of Mn-doped

chalcopyrites, where the Mn density varies as a function of the depth (cf. fig.

2.3(b), PES combined with sputter-etching provides us with the depth pro-

file of the interfacial compound. Chemical analysis of ZnGeP2:Mn is made

for both the surface region and the deep region using PES and Ar+-ion sput-

tering. At the same time, the use of Mn 3p-3d resonant PES, one can have

information of the degree of localization of Mn 3d electrons in ZnGeP2:Mn.

The confirmation of the localized nature of Mn 3d electrons will be the start-

ing point to discuss exchange interactions of these localized electrons and the

sp-band electrons, which is thought to be the important property of DMS.

Special care will be taken if there are any traces of MnP in the photoemission

spectra of ZnGeP2. After the study of ZnGeP2:Mn by PES, magnetization

measurements combined with sputter-etching will be performed to investi-

gate where the ferromagnetism comes from: the densely Mn incorporated

surface region of the interface, or the dilutely Mn scattered deep region of

the interface. Finally, I would like to discuss the features seen in the pho-

toemission spectra which could possibly be related to the mechanism that

strongly stabilizes the ferromagnetism above the room temperature.

References

[1.1] S. Koshihara, A. Oiwa, M. Hirasawa, S. Katsumoto, Y. Iye, C. Urano,

H. Takagi, and H. Munekata, Appl. Phys. Lett. 78, 4617 (1997).

[1.2] R. Fiederling, M. Keim, G. Reuscher, W. Ossau, G. Schmidt, A.

Waag, and L.W. Molenkamp, Nature 402, 787 (1999).

[1.3] Y. Ohno, D.K. Young, B. Beschoten, F. Matsukura, H. Ohno, and

D.D. Awschalom, Nature 402, 790 (1999).

[1.4] H. Ohno, D. Chiba, F. Matsukura, T. Ohmiya, E. Abe, T. Dietl, Y.

Ohno, and K. Ohtani, Nature 408, 944 (2000).

[1.5] K. Ono, D.G. Austing, Y. Tokura, and S. Tarucha, Science 297, 1313

(2002).

[1.6] L.A. Kolodziejski, R.L. Gunshor, T.C. Bonsett, R. Venkatasubrama-

nian, S. Datta, R.B. Bylsma, W.M. Becker, and N. Otsuka, Appl.

Phys. Lett. 47, 169 (1985).

[1.7] H. Munekata, H. Ohno, S. von Molnar, A. Segmuller, L.L. Chang,

and L. Eisaki, Phys. Rev. Lett. 63, 1849 (1989).

[1.8] H. Ohno, H. Munekata, S. von Molnar, and L.L. Chang, J. Appl.

Phys. 69, 6103 (1991).

[1.9] H. Ohno, A. Shen, F. Matsukura, A. Oiwa, A. Endo, S. Katsumoto,

and Y. Iye, J. Appl. Phys. Lett. 69, 363 (1996).

[1.10] T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, Science

287, 1019 (2000); T. Dietl, H. Ohno, and F. Matsukura, Phys. Rev.

B 63, 195205 (2001).

[1.11] H. Katayama-Yoshida, K. Sato, and T. Yamamoto, JSAP Interna-

tional 6, 20 (2002) and references therein.

[1.12] K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett. 79, 988 (2001).

12 References

[1.13] H. Saeki, H. Tabata, and T. Kawai, Solid State Commun. 120, 439

(2001).

[1.14] S. Sonoda, S. Shimizu, T. Sasaki, Y. Yamamoto, and H. Hori, J.

Cryst. Growth 237-239, 1358 (2002).

[1.15] G.A. Medvedkin, T. Ishibashi, T. Nishi, K. Hayata, Y. Hasegawa, and

K. Sato, Jpn. J. Appl. Phys. 39 L949 (2000).

[1.16] S. Cho, S. Choi, G.-B. Cha, S.C. Hong, Y. Kim, Y.-J. Zhao, A.J.

Freeman, J.B. Ketterson, B.J. Kim, Y.C. Kim, and B.-C. Choi, Phys.

Rev. Lett. 88, 257203 (2002).

[1.17] K. Ando, cond-mat/0208010.

[1.18] E.E. Huber Jr. and D.H. Ridgley, Phys. Rev. 135 A 1033.

Chapter 2

Physical properties of

ZnGeP2:Mn

2.1 Ternary chalcopyrite semiconductor ZnGeP2

Zinc germanium diphosphide (ZnGeP2) has been attracting interest due to

its promising applications for non-linear optical parametric oscillator laser

systems in the mid-infrared. It can provide continuous laser output over the

range from 3 to 8 µm at conversion efficiency above 50 % when used as an

optical parametric oscillator pumped at 2 µm. It shows a large nonlinear op-

tical coefficient, an appropriate birefringence, a wide range of transparency,

and high thermal conductivity, which are suitable for its applications. How-

ever, there is a broad optical absorption around 1 µm limiting the perfor-

mance of optical parametric oscillator (Fig. 2.1). This absorption center is

attributed to the point defects present in all bulk-grown ZnGeP2. Consid-

erable efforts are being made to reduce the magnitude of this absorption by

controlling the synthesis and growth procedures, as well as identifying the

responsible defects by electron paramagnetic resonance (EPR) and electron-

nuclear double resonance (ENDOR) studies.

The chalcopyrite structure of ZnGeP2 is shown in Fig. 2.2. It differs

from the zinc blend structure in doubling of the unit cell along a fourfold axis,

rendering the system body-centered tetragonal. The II-IV-V2 chalcopyrites

are easily seen as an extension of the III-V zinc-blende compounds. The

combination of group-II and group-IV atoms plays the role of two group-

III atoms and jointly supplies three valence electrons per atom, to combine

with five electrons provided by each group-V ion. The lattice parameters

for ZnGeP2 are a = 10.31791 A, η = 0.9800, and u = 0.25816, where η ≡c/a is the distortion parameter and u is the anion displacement parameter

[2.1]. ZnGeP2 shows slight tetragonal compression leading to a nonideal

14 Chapter 2. Physical properties of ZnGeP2:Mn

6

5

4

3

2

1

01086420

#1 #2 #3 #4

ZnGeP2A

bsor

ptio

n co

effic

ient

(cm

-1)

Wavelength (µm)

transparent region

defect related absorption

Figure 2.1: Optical absorption in ZnGeP2. Spectra are shown for single crys-

tals of ZnGeP2 treated under various conditions: (1) as-grown; (2) annealed

in the presence of ZnGeP2 powder; (3) annealed in vacuum; (4) annealing

under a P atmosphere of 0.5 atm [2.2]. The annealing temperature was

600◦C.

η = c / 2a= 0.9800

u = 0.25816

a = 10.31791 (a.u)

a

c

Zn Ge P

Figure 2.2: Crystal structure of ZnGeP2. The structure parameters are from

ref. [2.1].

2.2. Novel II-IV-V2 chalcopyrite-based DMS 15

tetragonal distortion parameter η < 1. Besides there is a distortion of

the anion sublattice involving a shift by each anion away from its nearest

neighbors of one cation and towards those of other kind of cation, leading to

the nonideal anion displacement parameter u �= 1/4. This anion sublattice

shift is primarily the consequence of the atomic size difference between the

group-II atom (Zn) and the group-IV atom (Ge) (Zn is larger than Ge).

2.2 Novel II-IV-V2 chalcopyrite-based DMS

Recently, a new function was proposed by Medvedkin et al. by incorporating

Mn into the chalcopyrite semiconductors including ZnGeP2, which starts to

show ferromagnetism above room temperature [2.3, 4, 5, 6]. The develop-

ment of this series of new materials were motivated by the novel idea that,

while the II-IV-V2 chalcopyrite semiconductors have similar electrical prop-

erty to III-V semiconductors, the large amaount of Mn may be incorporated

due to the presence of the group II cations at the same time. (cf. Section

2.1)

III-V-based DMSs, In1−xMnxAs and Ga1−xMnxAs, were the first DMSs

to show ferromagnetism [2.8, 9, 7]. It is generally thought that the divalent

Mn ions substituting for the group III sites generate holes (e.g., Ga3+ + Mn

→ Ga + Mn2+ + hole) which are responsible for the hole-meditated ferro-

magnetism. However, the hetero-valence substitution hampers the ability to

dope high concentration of Mn ions in III-V semiconductors (maximum of ∼7 % in GaAs), which results in the relatively low TC of ∼ 110 K. Overcoming

this limited ability of doping was the key in realizing III-V-based high-TC

DMSs (one can find an interseting work of Mn δ-doped GaAs which resulted

in the nominal Mn concentration of ∼ 15 % and TC as high as 172 K [2.8]).

On the other hand, a high concentrarion of Mn ions can be incorporated

in II-VI semiconductors. For example, CdTe can accomodate Mn ions up

to 77 % [2.9]. This is owing to the ease of isovalent substitution of Mn2+

for the group II site. However, without the presence of carriers meditating

the local magnetic moments, superexchange interaction between the mag-

netic moments becomes significant in the ground state, leading either to

paramagnetic, antiferromagnetic, or spinglass for these II-VI-based DMSs.

Theoretical calculations have predicted that ferromagnetism can be achieved

by sufficient hole doping in the II-VI-based DMSs [2.10]. Realization of such

ferromagnetic DMSs were achieved in modulation-doped CdMnTe quantum

well [2.11] and in p-Zn1−xMnxTe:N [2.12], although the TC has been limited

to ∼ 3 K.

The advantage of using II-IV-V2 chalcopyrites as the host semiconduc-


tors of DMSs is therefore the following. First, Mn could easily substitute

the group II site, resulting in a high concentration of magnetic ions. Sec-

ond, when a portion of the incorporated Mn substitutes for the group IV

site, it will readily act as dopants. Then, through the hole-meditated mecha-

nisms, ferromagnetism is expected. Based on this postulation, there followed

the successful fabrication of CdGeP2:Mn [2.3, 4], ZnGeP2:Mn [2.5, 6], and

CdSnP2:Mn, all of which show ferromagnetism above room temperature.

Considering this idea from a more general point of view, one realizes

that this is the unique character achieved in the ternary chalcopyrite semi-

conductors but not in binary semiconductors. The latter has only one cation

site, while the former has two cation sites. This is the distinct property of

the ternaries that allows the possibility to functionalize each cation site in

a different way, e.g., magnetic ion doping at the group II site and acceptor

doping at the group IV site.

In addition to this fascinating property of the chalcopirites, they have

many natural defects as discussed in section 2.1. As for ZnGeP2, these de-

fects were the hamper in the performance of optical parametric oscilator,

but when used as a host semiconductor of DMS, these formerly unwanted

defects may act as hole producing centers which provide a favorable mech-

anism for the realization of ferromagnetic DMS. To this end, it is clear that

the ternary chalcopyrite-based DMSs are not just an extension of the con-

ventional binary-based DMSs, but rather, have their own new possibilities.

All these rich and interesting possibilities in the chalcopyrite-based

DMSs have attracted considerable interest as a result. In fact, many the-

oretical calculations on the chalcopyrite-based DMSs have come into issue

[2.13, 14, 15, 16] after the first report of ferromagnetism in CdGeP2:Mn.

The main conclusions from the calculations are summarized below.

• The ground state is antiferromagnetic for Cd1−xMnxGeP2 [2.13].

• Mn doped CdGeP2 would be ferromagnetic when Mn atoms partially

occupy the Ge site as well as the Cd site [2.14].

• Substitution of P for S (correspinding to electron doping) would sta-

bilize the ferromagnetic state [2.15]. The importance of the carriers

meditating the Mn impurites is indicated.

• Under the presence of hole-producing defects (cation vacancies and Zn

on Ge site), ferromagnetic state is always preferred in (Zn,Mn)GeP2

[2.16].

All the calculations above seems to converge into the general conclusion that,

simultaneous carrier doping with the Mn incorporation into the chalcopyrites

2.3. Synthesis of II-IV-V2:Mn 17

stabilizes ferromagnetism. However, the detailed mechanism of carrier dop-

ing is still controversial whether it is the MnGe (Mn on Ge site), the cation

vacancy, the anion substitution of some group-IV ions, or its complex.

2.3 Synthesis of II-IV-V2:Mn

Figure 2.3(a) shows the synsthesis of chalcopyrite-based DMSs following

Medvedkin et al.[2.3, 4, 5, 6] The schematic illustration of the Mn incorpo-

ration into II-IV-V2 chalcopyrite semiconductors is carried out by depositing

Mn metal on the annealed surface of the host semiconductors in an ultrahigh

vacuum (UV) chamber. This results in the diffusion of Mn atoms into the

substrate, yielding a densely Mn incorporated interface II-IV-V2:Mn. Fig.

2.3(b) shows the concentration profile of the CdGeP2:Mn as a function of

depth, clearly showing Mn atoms diffusing into the bulk. The optimum an-

nealing temperature for the preparation of ZnGeP2:Mn was reported to be

400◦C.

Figure 2.3: Mn incorporation into chalcopyrite semiconductors. (a)

Schematic illustration of the sample preparation. (b) Depth profile of the

atomic concentrations in CdGeP2:Mn obtained by SEM [2.3]. Hatched re-

gion is the sample surface.

The host chalcopyrite semiconductors can be either a single crystal or

polycrystalline powder. Surprisingly, it was reported that, when the sin-

gle crystal chalcopyrites were used as the substrate, reflection high-energy


electron diffraction (RHEED) pattern of the host semiconductor preserved

even after the sample synthesis [2.5, 6]. This indicates that the densly Mn

incorporated II-IV-V2:Mn interface retains the chalcopyrite structure. Such

preservation of the RHEED pattern was not observed in the Mn deposition

on InP, which is a zinc-blende binary semiconductor [2.6]. Therefore, the

Mn incorporation process introduced above seems to be a unique method

realized in the chalcopyrite semiconductors.

2.4 Magnetization of II-IV-V2:Mn and bulk

Zn1−xMnxGeP2

After the Mn incorporation into chalcopyrite semiconductors, the interface

shows ferromagnetism above the room temperature. Some of the reported

magnetization curves of the II-IV-V2:Mn interface are shown in Fig. 2.4.

Figure 2.4(a) and (b) show the M(T ) curve and the M −H curve at 298 K,

respectively, of the Mn-containing polycrystalline CdGeP2 powders. They

clearly show the ferromanetism in CdGeP2:Mn above room temperature.

Figure 2.4(c) is the M -H curve of ZnGeP2:Mn single crystal showing hyster-

sis at 350 K. It should be noted that InP:Mn prepared in a similar method

did not show ferromagnetism, corresponding to the disappearance of the

RHEED pattern after the sample preparation. In Fig. 2.4(b), the diamag-

netic behaviour of the undoped CdGeP2 is also shown. As is characteristic

of wide-gap semiconductors, undoped ZnGeP2 also shows a diamagnetic be-

haviour.

Recently, ferromagnetism in bulk Zn1−xMnxGeP2 was reported by Cho

et al. [2.17] The single crystal (x = 0.013 and 0.03) and policrystalline sam-

ples (x = 0.045, 0.056, and 0.2) were synthesized by loading the chemical

composition of each sample in to walled quartz ampoules, and then heated

up to 1130◦C. This temperature was above the optimum preparation tem-

perature of ZnGeP2:Mn (400◦C) and even above the melting temperature

of ZnGeP2 (1022◦C). These samples also showed ferromagnetism above the

room temperature (TC = 312 K, Fig. 2.5) However, at temperatures below

47 K, the samples with x = 0.056 and 0.2 (polycrystalline samples) were

antiferromagnetic. The antiferromagnetic ground state of Zn1−xMnxGeP2 is

consistent with the theoretical calculations by Zhao, et al. [2.13] though the

relation between the ZnGeP2:Mn interfacial compound is unclear. All the

samples of bulk Zn1−xMnxGeP2 were insulators, which is consistent with the

isovalent substitution of Mn2+ for Zn2+.

2.4. Magnetization of II-IV-V2:Mn and bulk Zn1−xMnxGeP2 19

(a)

(c)

CdMnGeP powder

2

CdGeP2

CdMnGeP powder

2

ZnMnGePsingle crystal

2

(b)

Figure 2.4: Magnetization curves of II-IV-V2:Mn interface. (a) Magneti-

zation of Mn-containing polycrystalline materials of CdGeP2:Mn. Average

Mn concentration is 20 %. (b) Magnetization of polycrystalline CdGeP2:Mn

and undoped CdGeP2 at 298 K. Undoped CdGeP2 shows diamagnetic be-

haviour, while CdGeP2:Mn shows clear hysterisis [2.6]. (c) Magnetization of

Mn incorporated ZnGeP2 single crystal at 350 K. Inset shows the magnified

plot indicating a hysterisis behaviour [2.5].


Temperature (K)

Mag

netiz

atio

n (e

mu/

g)

AF FM

PM

x = 0.2 (poly)

0.056 (poly)

0.045 (poly)

0.03 (single)

0.013 (single)

x

0.0001

0.001

0.01

0.1

1

10

0.200.150.100.050.00

1E-3

1E-4

H = 100 Oe200 K

Zn1-xMnxGeP2

Mag

netiz

atio

n (e

mu/

g)(a) (b)

Figure 2.5: Magnetization of bulk Zn1−xMnxGeP2 [2.17]. (a)Temperature-

dependent magnetization in 100 Oe magnetic field. (b)Magnetization at 200

K as functions of the concentration of Mn (x).

References

[2.1] M.D. Lind and R.W. Grant, J. Chem. Phys. 52, 5607 (1970).

[2.2] G.A. Verozubova, A.I. Gribenyukov, V.V. Korotkova, O. Semchinova,

and D. Uffmann, J. Cryst. Growth 213, 334 (2000).

[2.3] G.A. Medvedkin, T. Ishibashi, T. Nishi, K. Hayata, Y. Hasegawa, and

K. Sato, Jpn. J. Appl. Phys. 39 L949 (2000).

[2.4] K. Sato, G.A. Medvedkin, T. Nishi, Y. Hasegawa, R. Misawa, K.

Hirose, T. Ishibashi, J. Appl. Phys. 88, 7027 (2001).

[2.5] G.A. Medvedkin, K. Hirose, T. Ishibashi, T. Nishi, V.G. Voevodin,

and K. Sato, J. Cryst. Growth 236, 609 (2002).

[2.6] K. Sato, G.A. Medvedkin, and T. Ishibashi, J. Cryst. Growth 237-

239, 1363 (2002).

[2.7] H. Ohno, Science 281, 951 (1998); H. Ohno, J. Mag. Mag. Mat. 200,

110 (1999).

[2.8] A.M. Nazmul, S. Sugahara, and M. Tanaka, preprint, cond-

mat/0208299.

[2.9] J.K. Furdyna, J. Appl. Phys. 64, R29 (1988).

[2.10] T. Dietl, A. Haury, and Y. Merle d’Aubigne, Phys. Rev. B 55, R3347

(1997).

[2.11] A. Haury, A. Wasiela, A. Arnoult, J. Cibert, S. Tatarenko T. Dietl,

and Y. Merle d’Aubigne, Phys. Rev. Lett. 79, 511 (1997).

[2.12] D. Ferrand, J. Cibert, A. Wasiela, C. Bourgognon, S. Tatarenko, G.

Fishman, T. Andrearczyk, J. Jaroszynski, S. Kolesnik, T. Dietl, B.

Barbara, and D. Dufeu, Phys. Rev. B 63, 085201.

[2.13] Y.-J. Zhao, W.T. Geng, A.J. Freeman, and T. Oguchi, Phys. Rev. B

63, 201202 (2001).

22 References

[2.14] P. Mahadevan and A. Zunger, Phys. Rev. Lett. 88, 047205 (2002).

[2.15] Y.-J. Zhao, S. Picozzi, A. Continenza, W.T. Geng, and A.J. Freeman,

Phys. Rev. B 65, 094415 (2002).

[2.16] T. Kamatani and H. Akai, Phase Transition, in press.



Rev. Lett. 88, 257203 (2002).

Chapter 3

Results and discussions

3.1 Experimental

Ultraviolet photoemission (UPS) and x-ray photoemission (XPS) measure-

ments were performed at BL-18A of Photon Factory, Institute for Material

Structure Studies, National High Energy Accerelator Reserch Organization.

Photoelectrons were collected using a VG CLAM hemispherical analyzer in

the angle-integrated mode. The overall energy resolution was 800 meV for

XPS and 200 meV for UPS at room temperature.

The schematic illustration of the experimental setup is shown in Fig.

3.1. The photoemission spectrometer was equipped with a Mn evaporator, a

quartz thickness monitor, and an ion gun, together with the annealing sys-

tem at the sample mount. This setup allowed Mn deposition onto ZnGeP2 in

situ in the photoemission spectrometer. In situ preparation of ZnGeP2:Mn

is desirable for photoemission experiments, since it is less bothered with

unwanted contamination by gases which severely damages the surface and

hence the photoemission spectra. Mn incorporation was carried out by evap-

orating Mn metal (99.999 %) from the Mn evaporator onto the (001) surface

of single crystal ZnGeP2. The pre-synthesized ZnGeP2 substrate was cleaned

by Ar+-ion sputtering at 1.5 kV before the Mn deposition. Surface cleanli-

ness of the substrate was checked by core-level XPS. During Mn deposition,

the substrate temperature was kept at 400◦C following the sample synthesis

of Medvedkin et al. [3.1, 2] The Mn deposition rate was 3 A/min, which

was determined by the quartz thickness monitor. After the Mn deposition,

the sample was post-annealed for 5 min and then cooled down to room

temperature.

In order to see the depth profile, the in situ prepared ZnGeP2:Mn was

sputter-etched by Ar+-ion at 1.5 kV. The incidence angle of the ion beam

was set to 45◦ relative to the substrate surface normal in order to reduce

24 Chapter 3. Results and discussions

VG CLAM analyzer

Mn evaporator(Mn 99.999 %)

Ion gun(1.5 kV Ar-ion sputtering)

Quartz thickness monitorO

hmic heater

1 cmTherm

ocouple

Synchrotron radiation (hν = 30 - 70 eV)Mg Kα x-ray

Figure 3.1: Experimental setup at BL18-A of Photon Factory

the Ar+-ion implantation during the sputtering. Sputter-etching rate was

approximately 2 A/min under this configuration.

All the spectra, except the annealed substrate surface spectra, were

taken at room temperature. During the collection of photoelectrons at the

elevated temperatures, the heater voltage was switched off in order to avoid

the voltage fluctuations of the substrate which was in electrical contact with

the heater. The pressure was below 7 × 10−10 Torr during the measurements,

and the photoemission spectra did not change while collecting the data under

this condition. The magnetization of the sample was measured ex situ using

a SQUID magnetometer (Quantum design). For comparison, photoemission

spectra of polycrystalline MnP (TC = 290 K [3.3]) was measured by in situ

scraping.

3.2. Photoemission spectra of ZnGeP2 substrate 25

3.2 Photoemission spectra of ZnGeP2 sub-

strate

Prior to the Mn deposition, the effect of annealing and sputtering on the

ZnGeP2 substrate surface was studied. Figure 3.2(a) shows the UPS spectra

(hν = 60 eV) of ZnGeP2 covering the valence-band region and the shallow

core levels of Zn 3d (EB ∼ 10 eV) and Ge 3d (EB ∼ 30 eV). Figure 3.2(b)

shows the core-level intensity of Zn 3d and Ge 3d as functions of substrate

temperature. Starting from the as-sputtered surface, the temperature of

ZnGeP2 was gradually elevated up to ∼ 400◦C. When the temperature ex-

ceeded ∼ 300◦C, the Zn 3d intensity started to decrease rapidly, while that

of Ge 3d slightly increased. At 400◦C the Zn 3d intensity was ∼ 30 % of

the initial intensity. When cooling the sample from ∼ 400◦C down to room

temperature, on the other hand, there was no change in the core-level inten-

sities. Then after 3 min Ar+-ion sputtering (corresponding to ∼ 6 A removal

of the surface region), the spectrum recovered the as-sputtered one. [See also

Fig. 3.3(b), where the core-level intensities for the as-sputtered surface have

been normalized to Zn:Ge:P = 1:1:2 and the deviation from this is seen for

the 400◦C-annealed surface before the Mn deposition.]

The decrease of Zn content by annealing is attributed to the diffusion of

Zn atoms from the surface region and some Ge atoms to the surface region

above 300◦C. The reported optimum substrate temperature during the Mn

deposition 400◦C is thus plausible since the Mn may easily diffuse into the

sufficiently Zn-deficient surface region at this temperature. The Zn-depleted

layer in the surface region in ZnGeP2 resembles the cases reported for I-III-

VI2 chalcopyrite surfaces, e.g., CuInSe2, where the surface layer generally

exhibits CuIn3Se5 stoichiometry [3.4].

Further sputtering after the removal of Zn-depleted layer in the surface

region did not change the core-level intensities of the substrate within the

accuracy of 3 %, indicating the selective sputtering of the light P atoms

compared to the heavy Zn and Ge atoms can be ignored. Therefore, we refer

to the core-level intensities of this as-sputtered ZnGeP2 as revealing the bulk

chemical composition of Zn:Ge:P = 1:1:2 in the following normalizations of

the core-level intensities.

3.3 Mn deposition on ZnGeP2

Next we show a series of spectra for Mn deposition on the 400◦C-annealed

substrate. The surface with the nominal Mn thickness up to 500 A was

studied sequentially. Each time after having taken a set of spectra for one


30 25 20 15 10 5 0

Room temp.(after sputtering)

218 - 187oC

110 - 93oC

312 - 296oC

410 - 387oC

385 - 363oC

353 - 328oC

377 - 358oC

400 - 373oC

Ge 3d Zn 3d

Valence band

Binding energy (eV)

Int

ensi

ty (

arb.

uni

ts)

ZnGeP2 hν = 60 eV(a)

(b)1.5

1.0

0.5

04003002001000

Temperature (°C)

Inte

nsity

(ar

b. u

nits

)

Ge 3d Zn 3d

Figure 3.2: Photoemission spectra of ZnGeP2 (hν = 60 eV) at elevated tem-

peratures. (a) Raw spectra at elevated temperatures. Spectrum at room

temperature is the as-sputtered ZnGeP2. The intensities have been nor-

malized to the photon flux. (b) Core-level intensities of Zn 3d and Ge 3d

as functions of substrate temperatures. The finite temperature ranging for

each spectrum is the consequence of cooling down while switching off the

heater.

3.3. Mn deposition on ZnGeP2 27

Mn coverage, the deposited Mn was completely sputtered out and then Mn

was newly deposited.

Figure 3.3(a) shows the core-level spectra of the Mn-deposited sur-

face, and Fig. 3.3(b) shows their intensities normalized to the as-sputtered

ZnGeP2. The Mn 2p intensity has been normalized using the Mn 2p and P

2p intensity ratio of MnP. In the region of Mn thickness d < 64 A, one can

see a monotonic increase and decrease of the Mn and Zn signals, respectively,

without significant changes in the Ge and P intensities. This behaviour sug-

gests that Mn atoms primarily substitutes for Zn. In the Mn 2p3/2 spectra

below d = 16 A, one can see some signal on the higher binding energy side of

the dominant metallic peak at EB. = 638.7 eV. This indicates that a portion

of the incorporated Mn atoms have chemically reacted with the substrate

and presumably became Mn2+, which is the isovalent substitution of Zn2+.

In going from d = 64 to 130 A, the Ge and P intensities suddenly changed.

The saturation in the Mn intensity with the sizable signals of Ge and P

above d = 130 A is a clear signature of atomic diffusion associated with the

chemical reaction for such a thick Mn overlayer. The surface region above d

= 130 A consists of Ge-rich, ternary metallic compound(s) of Mn, Ge, and

P with possible inhomogeneities and/or phase mixture.

The valence-band spectra taken in the Mn 3p-3d core-excitation region

are shown in Fig. 3.4. The spectra have been normalized to the post focusing

Au mirror current. The Mn 3d-derived spectra were obtained in the follow-

ing way: the on- (51 eV) and off-resonance (48 eV) spectra were further

normalized to the valence band intensities of d = 0 A (undoped substrate)

at the corresponding photon energies. Then the subtraction was carried out

for these normalized spectra to obtain the Mn 3d-drived difference spectra.

This way of subtraction virtually eliminates the effect of the photon en-

ergy dependence of the host sp3 band and allows us to obtain the correct

representation of the Mn 3p partial density of states (PDOS). In fact, as

demonstrated in Fig. 3.3(a), using this method, the difference spectrum of

the undoped substrate (which should be ideally zero) is minimized.

In the valence band spectra for d = 4 and 16 A, one can see a clear peak

located ∼ 4 eV below EF in going through hν = 48 to 51 eV. This spectral

behaviour is similar to the previous results of the Mn incorporated II-VI- and

III-V-based DMSs [3.5, 6, 7, 8] and thus is attributed to the localized nature

of Mn 3d electrons. Surprisingly, however, strong Mn M3L45L45 Auger peak

above the core-excitation energy replaces the ∼ 4 eV peak for further Mn

deposition above d = 32 A. This indicates that the Mn 3d electrons became

itinerant in the surface region. The difference in the decay processes after

the 3p-3d core absorption, whether it leads to Auger-electron emission or

the direct recombination leading to resonance enhancement of the ∼ 4 eV


2.5

2.0

1.5

1.0

0.5

0.0

12 3 4 5 6

102 3 4 5 6

1002 3 4 5

Mn 2p

Zn 2p3/2

Ge 3d P 2p

656 652 648 644 640 636

Mn 2p1/2

Mn 2p3/2

02

48

1632

64130

260

d(Å)= 510

32 28

Ge 3d

1024 1020

Zn 2p3/2

132 128 124 120 116

P 2p Ge 3p

Binding energy (eV)

Inte

nsity

(ar

b. u

nits

)

Rel

ativ

e ch

emic

al c

ompo

sitio

n

Deposited Mn thickness (Å)

(a)

(b)

0

Figure 3.3: Core-level spectra of ZnGeP2:Mn for various Mn thickness de-

posited at 400◦C. (a) Raw spectra. Ther vertical scale is counts per second.

0 A corresponds to the spectra of the sputtered and annealed ZnGeP2 sub-

strate. (b) Core-level intensities as functions of deposited Mn thickness. For

normalization, see text.

3.3. Mn deposition on ZnGeP2 29

10 5 0

d = 0Å

difference

47

48

49

50

51on-

off-

Zn 3d

hν=52

(a)

10 5 0

d = 16Å

51

50

49

48

47

difference

off-

on-

Zn 3d

hν=52

(c)

10 5 0

d = 32Å

hν=54

53

5251

48

47

49

difference

Zn 3d

off-

on-

50

(d)

10 5 0

d = 4Å

difference

47

48

49

50

51on-

off-

Zn 3d

hν=52

(b)

Binding energy (eV)

Inte

nsity

(ar

b. u

nits

)

Figure 3.4: Valence-band spectra of ZnGeP2:Mn in the 3p-3d core excitation

region. On- and off-resonance energies are 51 eV and 48 eV, respectively.

Arrows in (d) indicate the constant kinetic energy of the Mn M3L4,5L4,5

signal. See also Fig. 3.6 (a) for d = 150A.

peak, reflects the different degrees of the 3d electron localozation. Strictly

speaking, however, it does not directly reflect the degree of localization in the

ground state, but rather the degree of localization in the intermediate state

after the 3p-3d core excitation. In other words, it depends on the lifetime

of the core exiton-like intermediate state, i.e., the state which the excited

electron is bound to the core hole at 3p. If the photoexcited electron from

the 3p orbital resides in the localized 3d orbital due to the strong interaction

between the excited electron and the core-hole, the core-hole excitation is

followed by the direct recombination of the excited electron with the 3p core-

hole, hence the ∼ 4 eV peak. The extinction of the ∼ 4 eV peak above d

= 32 A is in accordance with the Mn 2p core-level spectrum [Fig. 3.3(a)],

where the (localized) divalent Mn signal disappeared above d = 32 A and

the highly asymmetric line shape characteristic of a metallic Mn compound

appeared. As for the d = 16 A spectrum, both a clear Fermi edge and the

∼ 4 eV peak is observed. The line shape of the difference spectrum of d

= 16 A is almost a superposition of d = 4 A and 32 A difference spectra.

Together with the fact that Mn 2p core-level spectrum was showing metallic


and divalent signals simultaneously at d = 16 A [Fig. 3.3(a)], it is plausible

that the localized divalent Mn species and the metallic Mn compound(s)

coexisted at d = 16 A.

3.4 Depth profile of ZnGeP2:Mn interface

So far, the surface region of ZnGeP2:Mn interface has been studied exten-

sively, and we have seen that the surface region became metallic after a

sizable amount of Mn was deposited and diffused. Now, interest arises on

the chemical and electronic states formed underneath the surface metal-

lic compound(s) (cf. Fig. 2.3). In order to clarify the depth profile of the

interface, ZnGeP2:Mn of the nominal Mn thickness of 150 A was repeat-

edly sputter-etched and studied by PES until the core-level signal of Mn 2p

disappeared. In this series, the sputter-etched surface was not annealed be-

fore taking the photoemission spectrum, otherwise the exposed surface after

sputtering quickly became Zn depleted, which is the similar behaviour to

that observed in the surface region of the ZnGeP2 substrate (Section 3.2).

Figure 3.5(a) shows core-level spectra taken in the sputter-etching se-

ries and Fig. 3.5(b) shows their intensities [same normalization as in Fig.

3.3(b)] as function of sputtering time. The dramatic change in the first 20

min sputtering is attributed to the removal of the Zn-depeted layer of the

as-grown outer-surface. Subsequently, the Mn 2p3/2 core level starts to show

a shoulder structure at EB. = 641.7 eV, attributed to ionic Mn (Mn2+ most

likely). The systematic increase of this shoulder structure and the decrease

of the metallic main peak at EB = 638.7 eV between 20 to 70 min sput-

tering indicate that these signals are originated from chemically different

Mn species. After 100 min sputtering, the relative chemical composition

became Zn:Ge:P ∼ 1:1:2 suggesting the chalcopyrite-type matrix of Zn, Ge,

and P plus dilute Mn was exposed [Fig. 3.5 (b)]. After 230 min sputter-

ing, the Mn signal became totally of the ionic one [Fig. 3.5 (a)], indicating

that the Mn-diluted phase finally appeared. Now since the binding energy

of the ionic Mn 2p signal in 20 -100 min sputtering (in the intermediate

phase) corresponds to those after 230 min sputtering, the ionic Mn com-

pounds in the intermediate region may be attributed to precursors of the

Mn-diluted phase exposed after 230 min sputtering. However, around 200

min sputtering, there is a thin layer separating the intermediate phase and

the Mn-diluted phase. This may indicate that Mn diffusion from the surface

to bulk regions took place during the fabrication of this interface. However,

the origin of this discontinuity is unclear and further elaboration may be

necessary.

3.4. Depth profile of ZnGeP2:Mn interface 31

2.0

1.5

1.0

0.5

05004003002001000

P 2p

Mn 2p3/2

Zn 2p3/2

Ge 3d

× 5

656 652 648 644 640 636

Mn 2p3/2Mn 2p1/2t(min) =

20406070

230270330390450540

100130190210

0

132 128 124 120 116

P 2p Ge 3p

1024 1020

Zn 2p3/2

34 32 30 28 26

Ge 3d

Binding energy (eV)

Inte

nsity

(ar

b. u

nits

) (a)

(b)

Sputtering time (min)

Rel

ativ

e ch

emic

al c

ompo

sitio

n

Figure 3.5: Core-level spectra of ZnGeP2:Mn in the depth profile. (a) Raw

spectra. The vertical scale is counts per second. (b) Core-level intensities as

functions of sputtering time.


10 5 0

40

4648

49.5

51

hν=54(a)

ZnGeP2:Mnd = 150 Å, t = 0 min

10 5 0

404648

51

hν=54

Zn 3d

t = 130 min

difference

(x3)

on-

off-

(c)

10 5 0

(b)

Zn 3d

40

46

49.5

48

51

52

hν=54

t = 40 min

10 5 0

40

5148

hν=54

t = 330 min

Zn 3d difference

(x3)

on-

off-

(d)

Inte

nsity

(ar

b. u

nits

)

Binding energy (eV)

1.0 0.5 0Binding energy (eV)

Inte

nsity

(a.

u.)

hν = 48 eV

t(min) = 230330450

540510

Figure 3.6: Valence-band spectra of ZnGeP2:Mn (150 A) in the sputter-

etching series. Arrows in (a) and (b) denote the M3L4,5L4,5 Auger peak.

Inset shows the valence-band spectra near EF after 230 min sputtering.

Figure 3.6 shows the valence-band spectra in the sputter etching series

for photon energies in the Mn 3p-3d core excitation region. There was a

change in the decay process following the Mn 3p-3d core-hole decay process at

80 min sputtering, before which the M3L4,5L4,5 Auger process was dominant

[Fig. 3.6 (a) and (b)] and after which resonant photoemission is dominant

[Fig. 3.6 (c) and (d)]. This indicates again that the Mn 3d states changed

their character from the itinerant to localized states along the depth profile.

A peak at ∼ 4 eV peak was seen after 80 min sputtering, and hence Mn

was divalent deep below the surface. This is consistent with the recent EPR

measurements (6-fold degeneracy of Mn2+ being split by Zeeman effect was

observed) [3.9], suggesting that the diffused Mn is predominantly divalent.

However, the Mn 3d-drived intensity near EF in the difference spectra is

weak compared to those in Fig. 3.4 (b) and (c). This indicates that the Mn

3d states are more strongly localized in the deep region than the 3d states

of Mn substituting Zn in the early stage of the series of Mn deposition

(Fig. 3.4(b)). It also indicates that there is less coexisting metallic Mn

compound(s) compared to Fig. 3.4(c) where the spectral weight in the Mn-

drived difference spectra near EF was attributed to the coexisting metallic

3.5. ZnGeP2:Mn prepared at low temperatures 33

Mn compound(s). Inset of Fig. 3.6 shows valence-band spectra near EF

after 230 min sputtering, where Mn has fully reacted with the substrate

[Fig. 3.5 (a)]. They clearly show a Fermi edge. Since the Mn 3d-drived

spectra was supressed near EF , this Fermi edge comes from the valence band

of the host semiconductor which was somehow doped with metallic charge

carriers. Since isovalent substitution of Mn2+ for Zn2+ cannot dope the

system with carriers, Mn2+ may have substituted for the Ge site and/or Mn

incorporation simultaneously induced defects such as, e.g., VZn and ZnGe,

all of which produce hole carriers. The Fermi edge became obsecure after

540 min sputtering, in accordance with the diminishing Mn 2p core-level

intensity (Fig. 3.5).

3.5 ZnGeP2:Mn prepared at

low temperatures

So far, ZnGeP2:Mn prepared at 400◦C, the reported optimum temperature,

has been studied. It will be interesting to study ZnGeP2:Mn prepared un-

der other conditions. In this section, results from Mn deposition at room

temperature and a sputtering series of ZnGeP2:Mn prepared at 200◦C is

presented.

Figure 3.7(a) shows the Mn deposition series on the substrate kept at

room temperature. As Mn metal was deposited, the Mn 2p signal monoton-

ically increased while the other core-level signals decreased. The line shape

of Mn 2p was highly asymmetric, which is characteristic of the core-level

spectrum of a metallic sample. Correspondingly, no satellite structure was

observed on the higher binding energy side of the metallic Mn 2p peak. This

behaviour is simply interpreted as the accumulation of Mn metal on top of

the substrate. The incorporation of Mn atoms into the substrate was not

appreciable at this temperature. This is consistent with the results of the

surface study of ZnGeP2 (Section 3.2), where the ZnGeP2 surface was inert

at room temperature. One can still see some weak signals of Zn, Ge, and P

even at d = 60 A. These are attributed to the photoelectron signals from the

ZnGeP2 substrate below the Mn metal coverage (the photoelectron escape

depth in the XPS region is ∼ 20 A). It is expected that these signals are on

way of vanishing with some more Mn metal coverage. However, there are

no further data.

Figure 3.7(b) shows the Mn 2p3/2 core-level intensities as functions of

the deposited amount of Mn. Both Mn deposition at room temperature (RT

series) and 400◦C (400◦C series) are plotted. One can see the rapid rise of

the Mn signal in the early stage of RT series compared to that in the 400◦C


650 640

Mn 2p1/2

Mn 2p3/2

32 28

Ge 3d

130 120

P 2p Ge 3p

1028 1020

Zn 2p3/2

0 2468

1015

30

d (Å) =60

20

Binding energy (eV)

Inte

nsity

(ar

b. u

nits

)

ZnGeP2:MnMg Kα

1

0806040200

Mn

2p3/

2 in

tens

ity (

arb.

uni

ts)

Deposited Mn thickness (Å)

ZnGeP2:Mn

RT

400ºC

(a)

(b)

Figure 3.7: Mn deposition series on ZnGeP2 at room temperature (RT).

(a) Core-level spectra. Intensities in each window are counts per second.

Relative intensities between different windows are arbitrary. (b) Mn 2p3/2

intensity of Mn deposited at RT and 400◦C. Solid circles and open circles are

the Mn 2p3/2 intensities when Mn was deposited on RT and 400◦C substrate,

respectively. Abcissa is in the linear scale of the deposited Mn thickness.

Broken line is the expected saturation behaviour at RT.

3.5. ZnGeP2:Mn prepared at low temperatures 35

series. This strongly supports the diffusion of Mn atoms at 400◦C in the

surface region of the ZnGeP2 aubstrate but not at room temperature (cf.

Section 3.3). The apparent saturaton behaviour in the RT series at d = 60

A indicates that the thickness of the Mn coverage reached the photoelectron

escape depth. For Mn coverage above d = 60 A, the Mn signal intensity is

expected to saturate, as is drawn by the broken line in Fig. 3.7(b).

650 640

t(min)= 0

10507090110130170230270290310350390610650

770730

850890

Mn 2p1/2

Mn 2p3/2

1024 1020

Zn 2p3/2

32 28

Ge 3d

132 128 124 120 116

Ge 3p P 2pZnGeP2:Mn

Figure 3.8: Core-level spectra in the sputter-etching series of ZnGeP2:Mn (d

= 200 A) prepared at 200◦C. (a) Raw spectra. (b) Core-level intensities as

functions of sputtering time.

Next, the depth profile is shown for ZnGeP2:Mn of the 200 A nominal

thickness Mn synthesized at 200◦C. Figure 3.8 (a) shows core-level spectra

taken in the sputter-etching series. The change in the line shape in the

first 10 min sputtering (particularly the shift of the Ge 3d peak position

to the lower binding energy side) is again attributed to the removal of the


30 20 10 0

ZnGeP2:Mn(d = 200 Å)

hν (eV) = 47

4849

5051

54

60

70

Ge 3d

0 min

30 20 10 0

hν (eV) = 48

51

54

60

70

770 min

Ge 3d Zn 3d

30 20 10 0

hν (eV) = 48

51

60

70

350 min

Ge 3d

Zn 3d

Binding energy (eV)

Inte

nsity

(ar

b. u

nits

)

Figure 3.9: Valence-band spectra in the sputter-etching series of ZnGeP2:Mn

(d = 200 A) prepared at 200◦C. The triangles trace the constant kinetic

energy of M3L4,5L4,5 Auger electrons.

surface compound which is different from the sub-surface region. Up to

800 min sputtering, Mn signal was observed together with Zn, Ge, and P

signals, indicating that the Mn atoms diffused into the substrate already

at 200◦C. Surprisingly, however, the Mn 2p3/2 signal always appeared at

EB = 638.7 eV (the position of the metallic peak) and no divalent signal

was observed [compare the Mn 2p spectra of fig. 3.5(a)]. The valence-band

spectra in the sputtering series is shown in Fig. 3.9. Corresponding to the

Mn 2p spectra showing only the metallic signal, no localized nature of Mn

3d was observed in the behaviour of the valence-band spectra, and only

the Mn M3L4,5L4,5 Auger peak was observed. It is therefore concluded that

the annealing temperature of 200◦C was insufficient for the Mn atoms to be

chemically incorporated as divalent ions in the deep region of the ZnGeP2:Mn

interface, although the thermal diffusion of Mn atoms did take place.

3.6. Photoemission spectra of MnP 37

3.6 Photoemission spectra of MnP

In this section, we describe the PES results of a policrystalline MnP. MnP is

an itinerant-electron ferromagnet, whose TC is as high as 290 K [3.3]. Since

its percipitation is suspected as the origin of the room-temperature ferro-

magnetism in ZnGeP2:Mn, its photoemission spectra is worthy to evaluate.

15 10 5 0

40

43

44

45

46

47

48

49

50

51

53

55

60

hν(eV) = 70

AB

C D

Binding energy (eV)

'Inte

nsity

(ar

b. u

nits

)

135 130 125 120 115

P 2p

660 655 650 645 640 635

Mn 2p3/2

Mn 2p1/2

Binding energy (eV)

Inte

nsity

(ar

b. u

nits

) (a) MnP (c) MnP

(b) MnP

DO

S (s

tate

s/eV

cel

l)

PES

inte

nsity

(ar

b. u

nits

)

Binding energy (eV)

20

15

10

5

015 10 5 0

PES (hν = 45 eV) DOS

AB

CD

Figure 3.10: Photoemission spectra of polycrystaline MnP. (a) Core-level

spectra of Mn 2p and P 2p. (b) Valence-band spectrum obtained by Kak-

izaki et al. [3.10], and its comparison with the density-of-states curve calucu-

lated by Hasegawa and Yanase [3.11]. Major structures in the photoemission

spectrum is designated by A through D. (c) Valence-band spectra in the Mn

3p-3d core excitation region. Vertical dotted lines indicate the major struc-

tures in the valence-band spectra. Triangles indicate the constant kinetic

energy of the M 2,3L4,5L4,5 Auger signal.

Figure 3.10(a) shows the Mn 2p3/2 core-level spectrum of MnP. Mn

2p3/2 core level appeared at EB = 639.2 eV, which is different from the

observed Mn 2p3/2 core-level peak positions in ZnGeP2:Mn either in the Mn

deposition series or in the sputtering series. The valence-band spectra of

MnP in the 3p-3d core excitation region is shown in Fig. 3.10(c). The four


peak structures in the valence-band spectra (indicated by the vertical dotted

lines A to D) correspond to those in the previous MnP spectra by Kakizaki et

al. [Fig. 3.10(b)] [3.10]. These structures designated by A, B, C, and D were

assigned as the P 3s band, the bonding state of P 3p and Mn 3d orbitals, the

Mn 3d majority-spin band, and the Mn 3d minority-spin band, respectively,

through comparison with the calculated DOS [Fig. 3.10(b)] [3.11]. One can

see in Fig. 3.10 the M 2,3L4,5L4,5 Auger signal moving through the valence

band for photon energies larger than 50 eV. Correspondingly, there is no

signal of the EB ∼ 4 eV peak which was observed in the lightly Mn-doped

surface region [Fig. 3.4(a), (b)] and the Mn-diluted phase in the deep region

[Fig. 3.6(c), (d)] of ZnGeP2:Mn interface.

The spectral feature of MnP mentioned above was never observed in

the present study of ZnGeP2:Mn prepared at 400◦C. We therefore exclude

the possibility that MnP is dominant in the interface of ZnGeP2:Mn.

3.7 Magnetization measurements of

ZnGeP2:Mn

The magnetization of ZnGeP2:Mn of the 150 A nominal Mn thickness pre-

pared at 400◦C was studied ex situ using a SQUID magnetometer. Suc-

cessive removal of surface layer by Ar+-ion sputtering from the synthesized

sample was performed to see the depth profile of the magnetization. These

measurements serve for both the cross-check of our sample synthesis, and

the determination of where the magnetization comes from, i.e., the surface

metallic Mn compound or the from the Mn-diluted phase deep in the deep

region.

Figure 3.11(a) shows the magnetization of the synthesized ZnGeP2:Mn

(d = 150 A at 400◦C) and the one after sputtering for 200 min to remove

the surface metallic Mn compound. In the 200 min-sputtered spectra, only

the divalent Mn signal was observed in the core-level spectra. First, one

can clearly see the hysterisis in the M -H curve at room temperature (RT),

indicating that the present sample is a RT ferromagnet, and that our sample

systhesis was in good agreement with the previous works [3.1, 2]. After 200

min sputtering, surprisingly, one can still see nearly the same RT hysterisis.

This suggests that the Mn-diluted phase in the deep region is a RT ferromag-

net. The marginal drop of the saturation magnetization indicates that the

magnetization from the surface region was relatively small. The magnetic

moment per Mn atom in the Mn-diluted phase deep in the interface was

crudely estimated 1.5 ± 0.5 µB . The extimation was made in the following

way. First, we know the nominal volume of the deposited Mn on the 15 mm2

3.7. Magnetization measurements of ZnGeP2:Mn 39

-0.4

0

0.4

150-150

-0.4-0.2

00.20.4

150-150-2

-1

0

1

2

T = 330 K 10 K

-3

-2

-1

0

1

2

3

T = 330 K 10 K T = 330 K 10 K

Field (x 10-4

T)

Mag

netiz

atio

n (x

10-4

em

u / 1

5 m

m2 )

M (x

10-4

em

u / 1

5 m

m2 )

M (x

10-4

em

u / 1

5 m

m2 )

Field (x 10-4

T)0

0

Field (T)

t = 0 min

t = 200 min

(a) ZnGeP2:Mn (d = 150 Å)

-2

-1

0

1

2

-0.6 -0.4 -0.2 0 0.2 0.4 0.6

T = 330 K 10 K

t = 400 min

-0.4-0.2

00.20.4

150-150Field (x 10

-4 T)M

(x10

-4em

u/15

mm

2 )

1.5

1.0

0.5

300200100

Temperature (K)

H = 0.001 T

100 min

200 min

Mag

netiz

atio

n (x

10-4

em

u / 1

5 m

m2 )

(b)

t = 0 min

ZnGeP2:Mn (d = 150 Å)

400 min

Figure 3.11: Magnetization curves of ZnGeP2:Mn (d = 150 A) prepared at

400◦C. (a) Magnetization curves at T = 10 K and 330 K of as-prepared

ZnGeP2:Mn (upper), those after having removed the surface metallic Mn

compound (middle), and further sputtering of total 400 min. (b) Magneti-

zation versus temperature in the sputtering series.

substrate surface (let it be X) using the quartz thickness monitor. Second,

from Fig. 3.5, we calculate the area of Mn chemical composition inbetween

0 - 200 min (corresponding to 0 - 600 A thickness of the interface). This

leeds to the volume of Mn incorporated in this region (let it be Y ). Then

X - Y makes the nominal volume of Mn in the dilute Mn phase deep in the

interface. Assuming that this volume corresponds to that of Mn metal, we

estimate the magnetic moment of Mn per atom. We note that Y / X was

∼ 0.2. Therefore the error bar of the magnetic moment of Mn per atom

mainly comes from the estimation of X , and the contribution of the surface


metallic Mn compound to the magnetization is, if at all, small.

Figure 3.11(b) shows the temperature dependence of mangetization,

clearly showing ferromagnetism up to 400 K. One can see a kink at ∼ 300

K, which seems to be slightly enhanced by the removal of the surface layer by

sputtering. This may correspond to the TC = 312 K of bulk Zn1−xMnxGeP2

[3.12] (Fig. 2.5). Another anomaly at 20 - 50 K may also correspond to

the 47 K anomaly of bulk Zn1−xMnxGeP2. It is therefore possible that

the present sample is a multiphase one, containing Zn1−xMnxGeP2 as one

component. Recently, ZnGeP2:Mn prepared at 550◦C was studied and the

M(T ) curve showed pronounced singularities at ∼ 318 K and 20 - 50 K,

and small magnetization taling above 318 K the behaviour of which is more

similar to bulk Zn1−xMnxGeP2 than the present sample [3.13]. We note

that Zn1−xMnxGeP2 was electrically insulating [3.12], while the sample in

the present study showed a metallic Fermi edge, which was attributed to the

hole-producing defects and/or antisites. From these works, it is conjectured

that the increase of the preparation temperature of ZnGeP2:Mn above 400◦Cwill supress the hole-producing defects and/or antisites, which makes the

interfacial compound insulating, supresses the magnetization above 320 K,

and makes the ∼ 310 K and 20 - 50 K anomalies in the M -T curve more

pronounced. Further studies are necessary, in particular to understand the

relationship between the magnetic behaviour, the carrier density, and the

preparation temperature.

References

[3.1] G.A. Medvedkin, K. Hirose, T. Ishibashi, T. Nishi, V.G. Voevodin,

and K. Sato, J. Cryst. Growth 236, 609 (2002).

[3.2] K. Sato, G.A. Medvedkin, and T. Ishibashi, J. Cryst. Growth 237-

239 1363 (2002)

[3.3] E.E. Huber Jr. and D.H. Ridgley, Phys. Rev. 135 A 1033.

[3.4] U. Rao and H.W. Schock, Appl. Phys. A: Mater. Sci. Process. 69,

131 (1999) and references therein; I.M. Kotschau and H.W. Schock,

ICTMC13, Paris France (2002).

[3.5] L. Ley, M. Taniguchi, J. Ghijsen, R. L. Johnson, and A. Fujimori,

Phys. Rev. B 35, 2839 (1987).

[3.6] J. Okabayashi, A. Kimura, T. Mizokawa, A. Fujimori, T. Hayashi and

M. Tanaka, Phys. Rev. B 59, R2486 (1999).

[3.7] J. Okabayashi, T. Mizokawa, D. D. Sarma, A. Fujimori, T. Slupinski,

A. Oiwa, and H. Munekata, Phys. Rev. B 65 R161203 (2002).

[3.8] T. Mizokawa, T. Nambu, A. Fujimori, T. Fukumura and M. Kawasaki,

Phys. Rev. B 65 085209 (2002).

[3.9] P.G. Baranov, S.I. Goloshchapov, G.A. Medvedkin, T. Ishibashi, and

K. Sato, Abstract of 2nd Int. Conf. PASPS, Wurtzburg Germany

(2002); J. Superconductivity (special issue Incorporating Novel Mag-

netism) 2003 (in press).

[3.10] A. Kakizaki, H. Sugawara, I. Nagakura, and T. Ishii, J. Phys. Soc.

Jap., 49 2183 (1980)

[3.11] A. Yanase and A. Hasegawa, J. Phys. C 13 1899 (1980)



Rev. Lett. 88, 257203 (2002).

42 References

[3.13] G.A. Medvedkin, P.G. Baranov, and S.I. Goloshchapov, J. Phys.

Chem. Solids, in press.

Chapter 4

Summary

ZnGeP2:Mn interface was studied by photoemission spectroscopy. We ob-

served spectral features of localized Mn 3d states incorporated into the host

ZnGeP2 in the lightly Mn-doped surface region and in deep bulk below the

surface metallic Mn compound. A metallic Fermi edge of non-Mn 3d char-

acter was observed in the deep dilute Mn phase. This indicates that the

carrier doping ability in the dilute Mn phase of ZnGeP2:Mn is sufficiently

high compared to the II-VI and III-V-based DMSs, although the detailed

mechanism of doping is still unknown. Room-temperature ferromagnetism

was observed after removing the metallic Mn compound in the surface and

subsurface regions. This indicates that the ferromagnetism in ZnGeP2:Mn

is due to the dilute Mn phase deep in the deep region. No secondary phase

of MnP was observed in the photoemission spectra.

44 Chapter 4. Summary

Acknowledgements

First of all, I would like to express my deepest gratitude to Prof. Atsushi

Fujimori who introduced me into the field of photoemission spectroscopy.

His gentle and heartful advices, together with his deep insight into physics,

always encouraged me in a delightful way. Words for showing my gratitudes

to him always become insufficient, but I really thank him.

I was very lucky to collaborate with Prof. D.D. Sarma from my early

stage of the master course. Particularly, the opportunities that I partici-

pated in the synchrotron experiments together with him, and the exciting

discussions over the interpretation of the data, how to proceed to the next

step, how to extract the maximum results from the limited beamtime, etc.,

have definitely become my heritage. I really learned a lot from him. Besides,

it was his strong initiative that I started the study of ZnGeP2:Mn. Not to

mention of his charms, I had a very pleasurable time with him. I thank him

so much.

The experiments at the synchrotron were supported by a number of

people. I am particulary indebted to the members of Kinoshita group, Dr.

T. Okuda, Ms. A. Harasawa, Dr. T. Wakita, and Prof. T. Kinoshita, for

their valuable technical help during the beamtime. Prof. K. Ono and Prof.

M. Oshima have kindly provided us with pure Mn metal, Prof. M. Okusawa

and Prof. T. Komatsubara have let us measure MnP. All these supports were

indispensable and essential for the present thesis. Gratitude is expressed for

all of them.

Let me also thank Prof. K. Sato and his colleagues, Prof. G.A. Medved-

kin and Dr. T. Ishibashi. They provided us with such an interesting and

excellent samples of ZnGeP2 togeter with the valuable advice and discus-

sions. Not just the offer of ZnGeP2, but it was the kind advice on the op-

tical measurements of ferromagnets, which have provided me another new

enthusiasm. I also thank them for their heartful support during my stay in

Paris. The lonely journy became a wonderful one.

The life during the master course is commemorated with the joyful

members of the Fujimori group: Dr. J. Okabayashi, Dr. T. Yoshida, Mr. K.

Okazaki, Mr. K. Tanaka, Mr. S. Nawai, Mr. H. Yagi, Mr. J.I. Hwang, Mr.

46 Chapter 4. Summary

H. Wadachi, and Ms. Y. Shimazaki. From each of these members, I was

taught a lot ranging from physics to common sense, and strong interaction

with each of them lead to either the stimulating discussions, new ideas, or

the share of passion. I thank each of them for adding pleasures to my life.

I cannot miss the contribution of Mr. H. Ott during his short stay in

the Fujimori group. Especially, I would like to acknowledge the support and

the constructive suggestions during the beamtime. I learned from him the

German common-sence in performing experiments. The ideas were always

refreshing, and surely have broadened my point of view.

I gratefully acknowledge the valuable comments of Prof. T. Mizokawa.

It was one of the most pleasurable time when having stimulating disscussions

with him. I have been deeply inspired by his view of physics and his sincere

attitude towards research.

The following people have contributed either their magic, faith, time,

energy, vision, passion, support, or friendship in an important and appreci-

ated way: R. Anraku, H. Sakai, Y. Tanaka, J. Quilty, J.Y. Son, Y. Hitsuda,

D. Asakura, M. Ikeda, N. Ueda, M. Kurokawa, T. Tran, S. Hirata, Y. Aisaka,

K. Ideguchi, T. Otsuki, R. Kumazawa, H. Yoshida. K. Abe, and A. Ito. I

thank each of them very much.

Finally, I would like to express my special thanks to my brother and

my parents. This is more than special thanks.

January 2002,

Yukiaki Ishida

Photoemission study of in diluted magnetic semiconductor...

Documents

Transcript of Photoemission study of in diluted magnetic semiconductor...