Novel Mn-doped chalcopyrites

K. Sato[*], G.A. Medvedkin1, T. Ishibashi, S. Mitani2, K. Takanashi2, Y. Ishida3, D. D. Sarma4, J. Okabayashi3, A. Fujimori3, T. Kamatani5, H. Akai5

Department of Applied Physics, Tokyo University of Agriculture and Technology, Tokyo 184-8588, Japan

1Ioffe Physico-Technical Institute, St. Petersburg 194021, Russia

2Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan

3Department of Physics, University of Tokyo, Tokyo 113-0033, Japan

4Indian Institute of Science, Bangalore 560-012, India

5Department of Physics, Osaka University, Toyonaka, Osaka 560-0043, Japan

Heavily Mn-doped II-VI-V2 semiconductors, such as CdGeP2 and ZnGeP2 have been prepared by depositing Mn on single crystalline substrate at nearly 400‹C in an ultra high vacuum chamber. Well-defined ferromagnetic hysteresis with a saturation behavior appears in the magnetization curve up to above room temperature. The chemical states of the ZnGeP2:Mn interface has been clarified by a careful in-situ photoemission spectroscopy. The as-prepared surface consists of Ge-rich, metallic Mn compound. In and below the sub-surface region, dilute divalent Mn species as precursors of the DMS phase exist. No MnP phase was observed at any stage of the depth profile. Theoretical band-calculation suggests that the system with vacancies (Cd, Vc, Mn)GeP2 or a non-stoichiometric (Cd, Ge, Mn)GeP2 are ferromagnetic and energetically stable although ferromagnetism is not stable in a stoichiometric compound (Cd, Mn)GeP2.

Keywords: magnetic semiconductor, II-IV-V2 semiconductors, manganese-doped crystal, photoelectron spectroscopy, ab-initio calculation

1. Introduction

Spintronics or electronics using spin-related phenomena has been attracting attention because of its potential applicability to new functional devices combining transport and magnetic properties.[1]) Magnetic semiconductors and ferromagnet/semiconductor hybrid structures are now the most important topics of investigation in the field of new functional semiconductor devices.

There is a long history of research on this category of materials.[2] The first-generation materials are europium chalcogenides[3] and ternary chalcogenides of chromium with spinel-type crystal structures[4], which were studied intensively in the late 60fs and early 70fs. Although important physical properties of magnetic semiconductors were discovered at that time, researcher lost interest in these materials because of their low Curie temperatures and difficulty in growth of good-quality single crystals.

The second-generation materials are II-VI-based diluted magnetic semiconductors (DMSs),[5] among which Cd1-xMnxTe was the focus of most attention due to its capability to accommodate a high percentage of Mn atoms (as high as 77%) and its appropriate energy gap for optical application. The magnetic properties of most of these materials are either paramagnetic or spin-glass. Although the controllability of transport properties is relatively poor, the material shows a good optical property that led to its application to optical isolators.

The third-generation materials are III-V-based diluted semiconductors, in which magnetic properties are strongly dependent on the carrier concentration in the material.[6] These compounds have opened a door into the ferromagnetic realm and a number of exciting new properties such as spin injection, carrier-induced and optically controlled ferromagnetism have rapidly been discovered.[7],[8],[9], Unfortunately, the maximum Curie temperature reported to date for III-V based ferromagnetic semiconductors, namely (Ga,Mn)As, is limited at TC = 110 K.[10]

Researchers are eagerly awaiting new functional devices based on ferromagnetic semiconductors working at room temperature. Recently, we disclosed  room-temperature ferromagnetism in another diamond-like semiconductor II-IV-V2:Mn with chalcopyrite structure. II-IV-V2 chalcopyrites are close analogs to the well-developed III-V materials[11] and could be used in heterostructures based on them. Some of these materials belonging to this family show both p and n type conduction with mobility as high as III-V compounds. Compared with III-V semiconductors, in which Mn2+ should occupy the group III sites, Mn2+ can easily substitute for the group II site without any sacrifice of electrical neutrality. If a part of the Mn atoms occupy the group IV site, they will act as acceptors to supply holes to the Mn 3d band making the Mn ions partially trivalent, which may realize the Mn2+-Mn3+ double exchange mechanism. Based on this postulation, we tried an incorporation of Mn atoms into the ternary chalcopyrite type semiconductor and carried out crystallographic as well as magnetic and magneto-optical characterization. We have succeeded in incorporating high concentration of Mn atoms into both CdGeP2, and ZnGeP2 belonging to the ternary II-IV-V2 compounds, and discovered ferromagnetism at room temperature. [12],[13] After our discovery, there occurs a rush of reports on possible high-Tc magnetic semiconductors, such as ZnO:Co,[14] TiO2:Co[15], GaN:Mn[16] and ZnTe:Cr.[17]

In this paper, we first summarize the sample-preparation method and results of magnetic and magneto-optical characterizations. We next introduce some of theoretical approaches including an ab-initio calculation on band structures and magnetism. We also describe our recent experimental results on in-situ photoelectron spectroscopy.

2. Sample preparation[18]

CdGeP2 single crystals were grown by directional crystallization of the stoichiometric melt in a quartz ampoule or graphite crucible (diameter 1cm, length 10 cm), in which 8 to 10 grams of chemicals was charged. The method was developed at the Ioffe Institute.[19] A large block crystalline ingot was obtained, from which bulk single crystal of up to 100 mm3 in volume and free of visible defects were cut. The crystals were oriented by optic asterism figures or X-ray diffraction method and cut to a rectangular shape of a convenient orientation. The crystals showed highly compensated n-type conductivity.

The vertical Bridgman technique was employed to grow single crystal bulk ingots of ZnGeP2 with a size of 28 mm in diameter and 150 mm in length. Single crystal plates with a required crystallographic orientation were cut from the ingots. The samples show p-type conductivity, high resistivity and may be prepared with a controlled optical transparency in the infrared spectral range.

Oriented single crystals of CdGeP2 {112} and ZnGeP2 {001} with polished and etched faces were employed as host substrates. Mn was evaporated from a Knudsen cell and was deposited onto the crystal surface of these single crystals in an MBE chamber with base pressure of 10-9 Torr. In the case of CdGeP2, the temperature of the substrate during deposition was kept at 380‹C. The total thickness of the deposited Mn-layer was 30 nm. After deposition, the sample was annealed in situ at 500‹C for 30 min to assist a diffusion of Mn accompanied by the solid-phase chemical reaction with the host crystal. The process was monitored in situ using reflection high-energy electron diffraction (RHEED) equipment. RHEED patterns for the case of CdGeP2:Mn are shown in Fig. 1: (a) Before deposition of Mn, the pattern with well-defined diffraction spots due to chalcopyrite lattice was observed. (b) After deposition of Mn, the spots became obscured. (c) During annealing process at 500‹C, spotty patterns became prominent.[20] RHEED pattern of the material at initial and final stages of this reaction demonstrate a set of RHEED spots similar to each other and sharing the same chalcopyrite-type lattice.

On the other hand, in the case of ZnGeP2, the substrate temperature was elevated to 400‹C during deposition.  In this case, the deposition of Mn and subsequent solid-state reaction were simultaneously undertaken. Figure 2 illustrates RHEED patterns showing the evolution of ZnGeP2 single crystal surface (a) before, (b) during and (c) after the solid-phase chemical reaction with manganese. The RHEED pattern before deposition shows a Laue spot with well-defined Kikuchi lines suggesting a high quality nature of the crystal surface. The pattern changes to a streaky image during Mn deposition and to a spotty pattern showing a roughened surface at the finishing stage. The streaky pattern observed during deposition implies the appearance of flat terraces on the crystal surface during Mn-deposition, suggesting chalcopyrite phase is sustained even when the Mn covers the surface. We believe that the deposited Mn atoms diffuse into the lattice as soon as it reaches at the surface of the crystal, because if the substrate was kept at room temperature during deposition, we observed no streak patterns as shown in Fig. 3.

3. Characterization

EDX depth profile of the Mn/Cd ratio were measured using the cleaved portion of the Mn-deposited crystal, assuming that most of the Mn occupies the divalent Cd-site taking into account the ionic radii of Mn, Cd and Ge. The ratio at the surface reaches 53.4% and drops rapidly with depth, the value being 12.7% at 0.6 mm and 0.9% at 2.5 mm. The average Mn/Cd ratio is determined as 20% for effective thickness 0.5 mm.

Crystallographic properties of Mn-deposited single crystals of CdGeP2 and ZnGeP2 were analyzed by a Rigaku type RAD-IIC X-ray diffractometer (XRD). The XRD patterns of these single crystals show strong chalcopyrite peaks. No observable traces of second phase compounds were observed in the conventional XRD studies. In CdGeP2:Mn, slight shift of the chalcopyrite peaks to larger angles 2ƒÆ with increasing Mn-content was observed in a careful XRD experiment. It was found that the lattice constant changes as a = 5.741Å ® 5.695Å by incorporation of Mn.

To scrutinize presence of small amount of extraneous phases, high sensitive XRD measurements using a Rigaku type RINT-RAPID system with a glancing angle incidence of X-ray were carried out. In the XRD pattern of CdGeP2:Mn crystals, dominant spots can be assigned to chalcopyrite structure, while very weak diffraction rings that can be assigned to polycrystalline MnP were observed. On the other hand, in the XRD pattern of ZnGeP2:Mn crystals, only chalcopyrite spots were observed except for one obscure ring which could not be identified to any of known binary Mn-P or Mn-Ge phases.

Accurate X-ray analysis of ZnGeP2:Mn was conducted using XRD system with a four-axis goniometer. The XRD patterns of 008 and 112 reflections are shown in Figs. 4(a) and 4(b). As shown in Fig. 4(a), a sub-peak diffraction was observed in the higher-angle side of the main 112 reflection. On the contrary no trace of sub-peak was observed in 008 reflections, as shown in Fig. 4(b). These experimental results suggest that lattice constant of ZnGeP2:Ge becomes slightly smaller along 112 direction compared with that of host materials.

Magnetic properties of CdGeP2:Mn were measured using Toei type VSM-5 vibrating sample magnetometer (VSM) in the temperature range 80~423 K. Well-defined magnetic hysteresis curves were observed in CdGeP2:Mn system throughout the temperature range of the measurements as shown in Fig. 5. Room temperature M-H curves were clearly composed of diamagnetic and ferromagnetic components. Applying suitable corrections for diamagnetism and demagnetization field, the saturation field Hs and coercivity Hc were determined as 3 kOe and 0.4 kOe, respectively.[21] Assuming that the deposited Mn of 30 nm in thickness on the 3L5 mm2 surface area was completely incorporated into the host semiconductor, the magnetization per atomic unit was evaluated as 0.956L10-20 emu/atom, from which the gS value was determined as 1.03 mB. (S~1/2) On the other hand, magnetization of ZnGeP2:Mn was found to be very weak compared with the case of CdGeP2:Mn. The M-H curves measured at IMR, Tohoku University using SQUID magnetometer are shown in Fig. 7 (a)-(c). The saturation behavior persists up to 350 K. Estimated magnetic transition temperature is considerably higher than 350 K, taking into
account the negligibly small temperature variation.

In CdGeP2:Mn, magneto-optical spectra have been measured. Kerr ellipticity took maximum around 1.75eV, where band edge of the host material exists.21 It should also be noted that well-defined stripe pattern was observed in CdGeP2:Mn. The electrical conductivity was measured on the surface of CdGeP2:Mn, which manifests itself a typical metallic conduction behavior.[22]

4. In-situ photoelectron spectroscopy[23]

In order to clarify the chemical states of the densely Mn deposited ZnGeP2:Mn interface, in-situ ultraviolet and x-ray photoemission measurements were performed at BL-18A of the Photon Factory. Mn metal of the nominal thickness 130 Å was deposited onto the ZnGeP2 single-crystal surface annealed at 400‹C. Spectra were taken while the synthesized ZnGeP2:Mn was gradually Ar+-sputter-etched (1.5 kV).

Figure 7 shows the relevant core-level spectra taken in the sputtering series, and Fig. 8 shows their intensities as functions of sputtering time. The sputtering ratio was roughly estimated to be 2 Å/min. The as-prepared surface shows Ge and P signals as well as Mn signals. The Ge and P signals were observed even for the nominal 500 Å -Mn -deposition (not shown), indicating the outdiffusion of Ge and P atoms to the surface region. The line-shape and the energy position of the Mn 2p core-level spectrum is that of a metallic Mn compound, indicating that the compound in the surface region is a metallic Mn compound. After 20 min sputtering, the Mn 2p core-level spectra start so show a shoulder structure at EB. = 641.6 eV in addition to the main metallic peak at EB. = 638.8 eV. The systematic increase of the shoulder peak and the decrease of the main peak between 20-100 min sputtering suggest that these signals are originated from different Mn compounds, the former peak being attributed to the divalent signal of a DMS-like compound. After 100 min sputtering, the intensity ratio of Zn, Ge, and P becomes fairly stable and approaches the value of the ZnGeP2 substrate (Fig. 8). This suggests that after removing the intermediate sub-surface layer, the matrix of Zn, Ge, and P has come to the chalcopyrite structure with Mn incorporated in it. After 230 min sputtering, only the divalent Mn signal is observed, the intensity of which decreases systematically until the Mn signal finally disappears (Fig. 7).

Corresponding spectra in the valence-band region have been obtained. There was a clear change in the decay process after the Mn 3p-3d resonant absorption at 80 min sputtering, before which the Mn M23VV Auger process dominates the decay, and after which the super-Coster-Krönig decay is dominant. This indicates that the Mn 3d have the successively changed from the itinerant Mn 3d to the localized Mn 3d along the depth profile.

During the whole sputtering series, no signal at EB. = 639.2 eV, which is the peak position of Mn 2p3/2 of MnP, was observed. From this, we excluded the possibility of MnP (TC = 290 K) being the origin of the room-temperature ferromagnetism. We have also studied ZnGeP2:Mn annealed at 200‹C, but no DMS-like signals were seen in the sputtering series. This indicates that the high temperature of 400‹C is necessary for Mn to be incorporated as divalent cations in the DMS-like compound.

5. Ab-initio calculation[24]

The origin of the ferromagnetism in DMSs has been investigated using the first-principles electronic structure calculations by Akai et al.[25],[26] In these systems, the effective exchange interactions are mainly determined by the competition between the double-exchange and superexchange interactions. The electronic structure calculation based on the local density approximation (LDA) usually takes into account the basic process producing the double-exchange and superexchange, which makes it possible to discuss the magnetic structure of the system semi-quantitatively. To examine the relative stability of the magnetic states, the total energy difference between the ferromagnetic and spinglass-like states were calculated, from which stability of ferromagnetic or spinglass-like state is judged at 0 K. Moreover, when the ferromagnetic state is more stable than the spinglass-like state, the difference gives an estimate of the ferromagnetic Curie temperature if suitably normalized by the help of the mean-field theory.

When Cd or Zn atoms are substituted by Mn atoms, the ground state magnetic structure is spinglass-like. This result is consistent with previous calculation by Zhao et al.[27] This is because d-states are nearly half-filled and the superexchange prevails. On the other hand if Mn atoms substitute Ge atoms, the system becomes ferromagnetic due to the double-exchange due to d-holes. However, the calculation of the formation energies shows that the latter is not energetically favorable.

It is shown that the system with vacancies (Cd, Vc, Mn)GeP2 or non-stoichiometric (Cd, Ge, Mn)GeP2 are also ferromagnetic and energetically favorable compared with other systems. Figure 9 shows the energy difference between the ferromagnetic and spinglass-like state as a function of the vacancy concentration, where the positive DE means that the ferromagnetic state is more stable than the spinglass-like state. Though we do not exclude the possibility that some other unknown magnetic phases exist in the matrix, we conclude at the moment that the above two are the most plausible candidates for the ferromagnetic phase observed experimentally in CdGeP2:Mn. We come to the same conclusion also for ZnGeP2:Mn, i.e., the existence of (Zn, Vc, Mn)GeP2 or (Zn, Ge, Mn)GeP2 seem to be the origin of the ferromagnetism of ZnGeP2:Mn. Existence of Zn vacancies in undoped ZnGeP2 has been observed experimentally.[28],[29]

All the results are summarized in Table 1. The stable magnetic state is ferromagnetic except for (Cd,Mn)GeP2 and Cd3MnGe4P8. The calculation of the formation energy shows that (Cd, Mn)GeP2 has a lower formation energy than Cd(Ge,Mn)P2, supporting a natural expectation that Mn will substitute Cd. On the other hand, for the non-stoichiometric cases, the calculation shows that Mn atoms substitute Cd atoms in the Ge-rich case and substitute Ge atoms in the Cd-rich cases. This result is consistent with previous calculation by Mahadevan and Zunger.[30] We assume that the chemical potentials for Cd and Ge are those of CdGeP2. Though this seems to be plausible, we are not quite confident whether this assumption corresponds to the real experimental situations or not. In this respect, the determination of the most plausible structure may need further elaboration.

6. Conclusion

Sample preparation and characterization of novel chalcopyrite-based magnetic semiconductors CdGeP2:Mn and ZnGeP2:Mn are summarized. In both materials, ferromagnetic properties have been observed up to temperatures considerably higher than room temperature. In-situ photoelectron spectroscopy revealed that the surface layer of ZnGeP2:Mn is metallic while sub-surface layer is non-metallic with well-defined Mn2+ signals, suggesting presence of DMS states. Ab-initio calculation on the electronic and magnetic states of ZnGeP2:Mn was performed. It is concluded that ferromagnetic states in Mn-doped chalcopyrites become stable if the presence of vacancies or nonstoichiometry is assumed. Further effort is necessary to prepare single phase crystal with uniform Mn concentration.


This work was carried out under the 21st Century COE program on "Future Nano-Material". This work has been supported in part by the Grant-in-Aid for Scientific Research (A) (Number 13305003).

[*] E-mail:


[1] G.A. Prinz: Science 250 (1990) 1092

[2] K.Sato: Crystal growth and characterization of magnetic semiconductors; Advances in Crystal Growth Research, eds. Ki. Sato, Y. Furukawa and K. Nakajima: Elsevier, Amsterdam, 2001, pp.303-319.

[3] T. Kasuya, A. Yanase, Rev. Mod. Phys. 40 (1968) 684

[4] R. P. Van Stapele, Ferromagnetic materials, vol. 3, ed. E. P. Wohlfarth, North-Holland Publ. Co, Amsterdam, 1982, Chap.8.

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

[6] H. Ohno, J. Magn. Magn. Mater. 200 (1999) 110

[7] Y. Ohno, D.K. Young, B. Beschoten, F. Matsukura, H. Ohno and D.D. Awschalom, Nature 402 (1999) 790

[8] S. Koshihara , A. Oiwa, M. Hirasawa, S. Katsumoto, Y. Iye, C. Urano, H. Takagi, and H. Munekata: Phys. Rev. Lett. 78  (1997) 4617.

[9] Ohno, H., Chiba, D., Matsukura, F., Omiya, T., Abe, E., Dietl, T., Ohno, Y. & Ohtani: Nature 408 (2000) 944.

[10] H.Ohno: J. Mag. Magn. Mater. 200 (1999) 110.

[11] N.A. Gorynova: Ternary Diamond-like Semiconductors (Soviet Radio, Moscow, 1968)

[12] G. A Medvedkin, T. Ishibashi, T. Nishi,K .Hayata, Y. Hasegawa and K. Sato, K: Jpn. J. Appl. Phys. 39 (2000) L949.

[13] G.A. Medvedkin, K. Hirose, T. Ishibashi, T. Nishi, V.G. Voevodin and K. Sato: J. Cryst. Growth 236 (2002) 609.

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

[15] Y. Matsumoto, R. Takahashi, M. Murakami, T. Koida, X.-J. Fan, T. Hasegawa, T. Fukumura, M. Kawasaki, S. Koshihara and H. Koinuma.: Jpn. J. Appl. Phys. 40 (2001) L1204.

[16] S. Sonoda, S. Shimizu, T. Sasaki, Y. Yamamoto, H. Hori: J. Cryst. Growth 237–239 (2002) 1358–1362

[17] H. Saito, V. Zayets, S. Yamagata and K. Ando: Phys. Rev. B66 (2002) in press.

[18] K. Sato, G. A. Medvedkin and Takayuki Ishibashi: J. Cryst. Growth 237-239 (2002) 1363.

[19] A. S. Borshchevskii, N. A. Goryunova, F. P. Kesamanly, D. N. Nasledov, Phys. Stat. Solidi 21 (1967) 9.

[20] K. Sato, G.A. Medvedkin T. Ishibashi, T. Nishi, R. Misawa, K. Yonemitsu, and K. Hirose, J. Magn. Soc. Jpn. 25 (2001) 735. (In Japanese)

[21] K. Sato, G.A. Medvedkin, K. Hayata, Y. Hasegawa, T. Nishi, R. Misawa and T. Ishibashi: J. Magn. Soc. Jpn. 25 (2001) 283.

[22] G.A. Medvedkin: private communication

[23] Y. Ishida, D.D. Sarma, K. Okazaki, J. Okabayashi, A. Fujimori, G.A. Medvedkin, T. Ishibashi and K. Sato: This conference.(P1-12)

[24] T. Kamatani and H. Akai: Phase Transitions (in press).

[25] H. Akai: Phys. Rev. Lett. 81 (1998) 3002.

[26] H. Akai, T. Kamatani, S. Watanabe: J. Phys. Soc. Jpn Suppl.A 69 (2000) 112.

[27] Y.-J. Zhao, W. T. Geng, .A. J. Freeman and T. Oguchi: Phys. Rev. B63 (2001) 201202(R)

[28] M.H.Ralowsky, W.K.Kuhn and W.J.Launderdale: Appl. Phys. Lett. 64 (1994) 1615.

[29] L.E. Halliburton, G.J. Edwards, M.P. Scripsick, M.H. Rakowsky, P.G. Schunemann and T.M. Pollak: Appl. Phys. Lett. 66 (1995) 2670.

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