Rare-earth doped III-nitride semiconductors for semiconductor spintronics

Journal of Luminescence 132 (2012) 3136–3140

Contents lists available at SciVerse ScienceDirect

Journal of Luminescence journal homepage: www.elsevier.com/locate/jlumin

Rare-earth doped III-nitride semiconductors for semiconductor spintronics Hajime Asahi n, Shigehiko Hasegawa, Yi-Kai Zhou, Shuichi Emura The Institute of Scientific and Industrial Research, Osaka University, 8-1 Mihogaoka, Ibaraki, Osaka 567-0047, Japan

a r t i c l e i n f o

a b s t r a c t

Available online 8 December 2011

InGaGdN layers and InGaGdN/GaN superlattice (SL) structures were grown by plasma-assisted molecular beam epitaxy. InGaGdN layers exhibited photoluminescence emission at room temperature and its peak wavelength was red-shifted with the increase of In composition. Clear hysteresis and saturation were observed in the magnetization versus magnetic field curves at room temperature for the InGaGdN layers. Si co-doping into InGaGdN layers increased the electron carrier concentration and enhanced the magnetization. In the InGaGdN/GaN SL samples, enhanced magnetization was also observed. Si doping into wide bandgap GaN layers in these SL structures further increased the magnetization, where InGaGdN layers were not doped with Si. All these results can be understood with the carrier-mediated ferromagnetism. & 2011 Elsevier B.V. All rights reserved.

Keywords: InGaN-based diluted magnetic semiconductor Rare-earth-doping Ferromagnetism Photoluminescence Molecular beam epitaxy

1. Introduction Rare-earth (RE) doped semiconductors are widely studied for the application to photonic devices. However, some of RE-doped semiconductors exhibit ferromagnetic characteristics (diluted magnetic semiconductors (DMSs)) and are also important for the application to semiconductor spintonic devices such as tunnel magnetoresistance (TMR) devices, circular polarized laser diodes (CP-LDs) and spin field effect transistors (FETs). CP-LDs are one of the important devices to realize the optical communication system that is sturdy against the wiretapping. CP-LDs can be fabricated using the DMS as an active layer or a cladding layer. CP-LDs are also important devices to construct the information processing devices system by the combination of circular-polarized light controlled TMR devices. Bhattacharya et al. [1] reported the fabrication of CP-LDs using vertical-cavity surface-emitting laser (VCSEL) structure with GaMnAs DMS layer as spin-polarized carrier (hole) supplying layer and the observation of circularpolarized light at a low temperature of 80 K because of the low Curie temperature of GaMnAs. To fabricate practical CP-LDs, room temperature ferromagnetic DMS is a requisite. We have reported, for the first time, the observation of ferromagnetic characteristics even at high temperatures ( 4400 K) for the rare-earth Gd-doped GaN (GaGdN) grown by molecular beam epitaxy (MBE) [2]. Since then, extensive studies have been conducted theoretically [3–7] and experimentally on the magnetic semiconductor GaGdN prepared so far by various techniques, which include MBE [8–11], metalorganic vapor phase

n

Corresponding author. Tel.: þ81 6 6879 8405; fax: þ81 6 6879 8409. E-mail address: [email protected] (H. Asahi).

0022-2313/$ - see front matter & 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jlumin.2011.11.033

epitaxy (MOVPE) [12] as well as via Gd ion implantation into GaN [13,14]. In addition, ferromagnetism with even incredibly high values of Curie temperature (TC) 700 K was reported in GaGdN films [15]. On the other hand, it is known that InGaN can cover the wavelength range from ultraviolet to infrared. By combining these, the fabrication of CP-LDs in the optical communication wavelengths is considered to be possible. In this paper, we report on the growth of Gd-doped InGaN (InGaGdN) and their superlattice (SL) and multi-quantum well (MQW) structures and the observation of room temperature ferromagnetic characteristics as well as photoluminescence (PL) emission.

2. Experimental InGaGdN layers and InGaGdN/GaN and InGaN/GaGdN superlattice (SL) structures were grown on (0 0 0 1) sapphire substrates or 2-mm-thick MOVPE-grown GaN templates at substrate temperature of 450–500 1C by radio frequency plasma-assisted MBE. These layers were grown after the initial growth of  8-nm-thick GaN buffer layers and InGaN layers at 700 and 500 1C, respectively. Si co-doping into InGaGdN layers and GaN barrier layers of SL structures were also carried out. Fig. 1(a) and (b) shows the schematic sample structures for epilayers and SLs, respectively. Ga and In beam equivalent pressure (BEP) were kept at around  5.8  10  8 and  1.5  10  8 Torr, respectively. Gd cell temperature was 1070 1150 1C. Nitrogen flow rate was 1.5 sccm with fixed rf plasma power of 180 W. The grown InGaGdN layers and SLs were investigated through various techniques including crystalline quality by X-ray diffraction (XRD) measurements, local structure study around Gd

H. Asahi et al. / Journal of Luminescence 132 (2012) 3136–3140

3137

A GaN (~3 nm)

B

InGaGdN

Gd-N Gd-Ga

7-15 periods

Sample P1

InGaGdN (~100 nm)

T Gd = 1070°C

InGaN (~8 nm)

QW GaN (~6 nm)

GaNbuffer (~8 nm)

GaN buffer (~3 nm)

MOVPE-grown GaN (~2 nm)

MOVPE-grown GaN (~2 nm)

Sapphire (0001)

Sapphire (0001)

FT Intensity (arb. units)

InGaGdN (~2 nm) Gd-N

Sample P2

Gd-Ga

T Gd = 1100°C

Reference

Gd-Ga

Gd-N

T Gd

GaGdN = 1070°C

Gd-N

Fig. 1. Schematic illustrations of (a) InGaGdN layer and (b) InGaGdN/GaN superlattice samples.

Gd-Gd

LIII-edge by X-ray absorption fine structure (XAFS) measurements performed around Gd LIII-edge in a fluorescence mode at room temperature. PL measurements were performed using excitation by 325 nm line (3.8 eV) of a He  Cd laser. The magnetic properties were obtained by means of superconducting quantum interference device (SQUID) magnetometer.

GdN

Gd-Gd

Gd metal 0

1

2 3 4 Radial Distance (Å)

5

6

Fig. 2. (a) Radial distribution functions (RDFs) measured from the Gd ion for the InGaGdN single layers. (b) RDFs for the GaGdN, GaN and Gd metal as reference

3. Results and discussion

Laser pump

300 K

In 0.28 Ga 0.72 GdN

Normalized PL Intensity (arb.units)

XRD profiles (data not shown) indicated no phase separation for all samples, where the estimated InN molar fractions are about 28–35% and 5–8% for the single-layer and SL samples, respectively. The incorporation of Gd into InGaN layer was initially confirmed by the EPMA and the local structure around Gd atoms was studied by XAFS. Fig. 2(a) shows the radial distribution functions (RDFs) measured from the Gd ion [16]. Results from the GaGdN, GdN and Gd foil are also represented as reference in Fig. 2(b). Dashed lines A and B in Fig. 2(a) and (b) are assigned to denote the first nearest neighbor (NN) shell and the second NN shell, respectively. Two main peaks (A, B) from the samples InGaGdN and GaGdN seen in this figure correspond to the first NN of Gd–N bond and the second NN of Gd–Ga bond, respectively. Further fine analysis using a curve fitting method indicated that the Gd atoms in InGaN are about at the same radial distance as Gd of GaGdN in Fig. 2(b), which are at around 2.2 and 3.3 A˚ for Gd-N and Gd-Ga interatomic distances, respectively. As analyzed from the XAFS spectra, it was found that the Gd atoms in both GaN and InGaN host materials are surrounded by the similar atomic shells, namely by four N ions as the first NN and twelve Ga ions as the second NN. Therefore, it can be concluded that the Gd atoms substitutionally occupy the Ga sites in InGaN. Fig. 3 shows typical room temperature PL spectra taken for the two InGaGdN layers of different In compositions. Both samples contain about  1% of Gd concentrations as confirmed by EPMA. For the samples with the In composition of 28% and 35%, the emission peaks were observed at 2.38 and 2.19 eV, respectively. Thus, it was confirmed that the peak energy was red-shifted according to their InN molar fractions from these two samples. Fig. 4 shows the PL spectra for the InGaGdN (In: 23%; Gd: 1%) sample as a function of temperature. PL emission was clearly observed at all measuring temperatures (10–300 K). The temperature variation of PL peak energy exhibits S-shaped behavior (redshift blueshift redshift) with increasing temperature, which is attributed to localized excitonic emission due to inhomogeneity

In 0.35 Ga 0.65 GdN

x5

2.0

2.5 3.0 Photon energy (eV)

3.5

Fig. 3. Room temperature PL spectra taken for the InGaGdN layers of different InN molar fractions (InN molar fraction of 35%, broken line and InN molar fraction of 28%, solid line).

and carrier localization in InGaGdN alloys with the consideration of the band-tail approach. This temperature variation is similar to those for the Gd-undoped InGaN layers.

3138

H. Asahi et al. / Journal of Luminescence 132 (2012) 3136–3140

C6

PL Peak intensity (arb. units)

C4

Ea: 34 meV

1 0 -1

-400 -200 0

0

20

40

60

80 100

1000/T(K-1)

200 400

Fig. 5. Room temperature M  H curves for the InGaGdN films with different Gd concentrations.

Arrhenius plot

300K Si-doped InGaGdN

TGd = 1100°C

Magnetization (emu/cm3)

2

Peak energy (eV)

2.43 2.42 2.41 2.40 2.39

TSi = 1130°C

Undoped InGaGdN

0 1 0

-2

2.38 2.37

-1

0

50

100

150

200

250

300

-800

-400

0

400

800

Temperature (K) -4000 Fig. 4. (a) PL spectra for the InGaGdN sample (In: 28%; Gd:  1%) as a function of temperature. (b) Temperature variation of PL peak energy.

By means of a SQUID magnetometer, magnetic properties were measured at 300 K for all the prepared Gd-doped InGaN epilayers and SL samples. During measurements, the magnetic field was applied parallel to the sample surface, i.e., perpendicular to the c-axis. All the data presented here were corrected for the diamagnetic background of the substrate. Magnetization versus magnetic field curves for the InGaGdN single-layer samples with different Gd concentrations exhibit clear hysteresis and saturation magnetization (MS) measured at room temperature, as shown in Fig. 5 [17]. Even clear hysteresis can be further observed from the expanded curve of both samples as shown in the inset of Fig. 5. It shows that the saturation magnetization increases with the increase in Gd concentration. With the increase of magnetic Gd atoms, the ferromagnetic interaction is increased. Fig. 6 shows the magnetization versus magnetic field M H curves of InGaGdN sample and Si co-doped InGaGdN sample. Both samples show clear hysteresis and saturation characteristics measured at 300 K. With the increase in Si cell temperature (that is the increase of Si concentration), an increase in the saturation magnetization was also observed [18]. M H measurements were also carried out on GaN template and InGaN sample to confirm that the

-2000

0

2000

4000

Magnetic Field (Oe) Fig. 6. M  H curves measured in-plane at 300 K for Si-doped InGaGdN and undoped InGaGdN layers. Inset indicates the expanded view of the M  H curves.

magnetic behavior of the InGaGdN layers is indeed a result of Gd doping. Although such finding implied that the doping of magnetic ions (i.e. Gd) alone could induce the magnetic properties in InGaN, it is essential if the magnetism can be further improved through incorporation of extra shallow donors such as Si. These magnetic properties (Fig. 6) confirmed that by intentionally co-doped InGaGdN with Si could promote the enhancement in carrier density as determined by Hall measurement (Table 1) at room temperature as well as an increase in saturation magnetization (MS), which can be clearly seen from the room temperature M–H curves for the Si codoped InGaGdN sample. The increase in magnetization for the Si codoped sample is closely related to carrier-mediated ferromagnetism, that is, the Si co-doping increases the electron concentrations in InGaGdN layers as also inferred previously in few reports on the Si co-doped GaGdN samples and GaGdN/GaN superlattice samples [9,11,14]. In the InGaGdN/GaN SL structures, enhanced magnetization was also observed compared with that for the InGaGdN singlelayer (thickness: 100 nm), as shown in Fig. 7. Both samples were

H. Asahi et al. / Journal of Luminescence 132 (2012) 3136–3140

Table 1 Magnetic and electrical properties obtained at room temperature for the undoped and Si-doped InGaGdN samples shown in Fig. 6. Electron Resistivity conc. (cm  3) (O cm)

Undoped InGaGdN Si-doped InGaGdN

3.7 3.0

1.1  1017 4.7  1018

Magnetization (emu/cm3)

120 60

0.01 0.003

300K

20 15

1.8 4.0

U1

InGaGdN/GaN SL InGaGdN layer

Si-doped barriers Undoped barriers

Magnetization (emu/cm3)

Gd (%) Ms Hc (emu/cm  3) (Oe)

S1

300K 20

Sample

3139

10

U1

-10

10 5

-20

A2

0 -5

-4000

-10

-2000

0

2000

4000

Magnetic field (Oe)

-15

Fig. 8. M  H curves measured in-plane at 300 K for the InGaGdN/GaN SLs with and without Si doping into wide bandgap GaN layers.

-20 -4000

-2000

0

2000

4000

Fig. 7. M  H curves measured in-plane at 300 K for InGaGdN layers and InGaGdN/GaN SLs.

grown under similar growth conditions except for the sample structure. The magnetic properties of these samples show clear saturation magnetization at about 16 and 7 emu/cm3 for the SL sample and single-layer sample, respectively. In the SLs, electron carriers can flow into narrow bandgap InGaGdN well layers from the wider bandgap GaN layers. These increased carrier concentrations further improve the magnetization. Similar observations were also seen in the GaGdN/GaN and GaGdN/AlGaN SLs in which the enhancement of Ms was also explained by the carriermediated ferromagnetism [9,11]. Carriers (electrons) in GaN or AlGaN layers can flow into and accumulate in the GaGdN layers, resulting in the fact that the higher electron concentrations exist in the narrower bandgap GaGdN layers. Moreover, Si doping into wide bandgap GaN barrier layers in the SL structures was also found to be able to increase the magnetization as clearly observed from the room temperature M  H curves shown in Fig. 8. One reason that could induce the enhancement of magnetization is also the carrier-(electron)mediated ferromagnetism as previously observed in the Si co-doped InGaGdN single-layer. However, in the InGaN/GaGdN SL structures the magnetization was decreased. Such observation is indicated in the M  H curves in Fig. 9, between the InGaGdN/GaN SL and InGaN/GaGdN set of samples measured at room temperature. In the InGaN/GaGdN SL structure, carriers (electrons) flow out of the Gd-containing GaGdN layers and into the narrow bandgap InGaN layers. So the carrier concentrations in the GaGdN layers are decreased, resulting in the suppression of the magnetization. With the interesting magnetic and optical properties observed for the Gd-doped InGaN-based DMS layers, this study hopes to lay the foundation of InGaN-based DMS material that will someday enable the fabrication of longer wavelength spin-based electronic (spintronic) devices such as circular-polarized light emitting diodes (LEDs) and circular-polarized laser diodes. 3 nm-InGaGdN

Magnetization (emu/cm3)

Magnetic field (Oe)

Magnetic Field (Oe) Fig. 9. M  H curves measured in-plane at 300 K for InGaGdN/GaN SL and InGaN/ GaGdN SL.

(In:  3.9% and  6%, Gd: 1%))/9 nm-GaN MQW LED structures sandwiched with Si-doped n-type and Mg-doped p-type GaN layers were grown on n-type GaN template. From these MQW LED structure samples, PL emission from the InGaGdN quantum well layers was observed at around 400–500 nm at room temperature as shown in Fig. 10. Emission from GaN layers as well as defect-related broad band emission is also observed.

4. Conclusion InGaGdN layers and InGaGdN/GaN and InGaN/GaGdN SL structures were grown on (0 0 0 1) sapphire substrates or MOVPE-grown GaN templates by plasma-assisted molecular beam epitaxy. XRD profiles for the grown InGaGdN layers indicated no phase separation. The Gd incorporation in InGaN was verified with the XAFS measurement. PL emission was observed and PL peak energy was

3140

H. Asahi et al. / Journal of Luminescence 132 (2012) 3136–3140

from wider bandgap GaN layers. Thus the increased carrier concentrations further enhance the magnetization.

Acknowledgments

In ~ 3.9% PL Intensity (a.u.)

This work was supported in part by the Grant-in-Aid for Creative Scientific Research, Grant-in-Aid for Scientific Research, Grant-in-Aid for Scientific Research on Priority Area and the Management Expenses Grants for National Universities Corporations from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan.

In ~ 6.0% References

350

400

450

500

550

600

650

700

Wavelength (nm) Fig. 10. Room temperature PL spectra from the 3 nm-InGaGdN (In:  3.9% and  6%, Gd:  1%)) /9 nm-GaN MQW LED structures.

red-shifted according to their InN molar fractions. Clear hysteresis and saturation were observed in the M H curves at room temperature for the grown InGaGdN layers. Si co-doping into InGaGdN layers enhanced the magnetization. In the InGaGdN/GaN SL samples, enhanced magnetization was also observed. Si doping into wide bandgap GaN layers in these SL structures further increased the magnetization. However, in the InGaN/GaGdN SL structures the magnetization was decreased. Such results can be understood with the carrier-mediated ferromagnetism. That is, the Si co-doping increases the electron concentrations in InGaGdN layers. In the SLs, electron carriers flow into narrow bandgap InGaGdN layers

[1] M. Holub, J. Shin, S. Chakrabarti, P. Bhattacharya, Appl. Phys. Lett. 87 (2005) 091108. [2] N. Teraguchi, A. Suzuki, Y. Nanishi, Y.K. Zhou, M. Hashimoto, H. Asahi, Solid State Commun. 122 (2002) 651. [3] M. Hashimoto, K. Araki, K. Sato, H. Asahi, H. Katayama-Yoshida, Extended Abstracts of the 8th Symposium on the Physics and Application of Spinrelated Phenomena in Semiconductors, Sendai, Japan, 2002, pp.165–168. [4] G.M. Dalpian, S.H. Wei, Phys. Rev. B 72 (2005) 115201. [5] Y. Gohda, A. Oshiyama, Phys. Rev. B 78 (2008) 161201. [6] L. Liu, P.Y. Yu, Z. Ma, S.S. Mao, Phys. Rev. Lett. 100 (2008) 127203. [7] V.I. Litvinov, V.K. Dugaev, Appl. Phys. Lett. 94 (2009) 212506. [8] S. Dhar, L. Pe´rez, O. Brandt, A. Trampert, K.H. Ploog, J. Keller, B. Beschoten, Phys. Rev. B 72 (2005) 245203. [9] S.W. Choi, Y.K. Zhou, M.S. Kim, S. Kimura, S. Emura, S. Hasegawa, H. Asahi, Phys. Status Solidi (a) 203 (2006) 2774. [10] J.K. Hite, R.M. Frazier, R. Davies, G.T. Thaler, C.R. Abernathy, S.J. Pearton, Appl. Phys. Lett. 89 (2006) 092119. [11] Y.K. Zhou, S.W. Choi, S. Kimura, S. Emura, S. Hasegawa, H. Asahi, Thin Solid Films 518 (2010) 5659. [12] I.H. Hung, Y.H. Lai, Z.C. Feng, S. Gupta, T. Zaidi, I. Ferguson, W. Lu, Proc. SPIE 7784 (2010) 77840H. [13] S.Y. Han, J.K. Hite, G.T. Thaler, R.M. Frazier, C.R. Abernathy, S.J. Pearton, H.K. Choi, W.O. Lee, Y.D. Park, J.M. Zavada, Appl. Phys. Lett. 88 (2006) 042102. [14] J.K. Hite, R.M. Frazier, R.P. Davies, G.T. Thaler, C.R. Abernathy, S.J. Pearton, ¨ J.M. Zavada, E. Brown, U. Hommerich, J. Electron. Mater. 36 (2007) 391. [15] J. Hejtma´nek, K. Knı´zˇek, M. Maryˇsko, Z. Jira´k, D. Sedmidubsky´, Z. Sofer, V. Peˇrina, H. Hardtdegen, C. Buchal, J. Appl. Phys. 103 (2008) 07D107. [16] S.N.M. Tawil, R. Kakimi, D. Krishnamurthy, M. Ishimaru, S. Emura, H. Tambo, S. Hasegawa, H. Asahi, Phys. Stat. Sol. Rap. Res. Lett. 4 (2010) 308. [17] S.N.M. Tawil, D. Krishnamurthy, R. Kakimi, S. Emura, S. Hasegawa, H. Asahi, J. Cryst. Growth 323 (2011) 351. [18] S.N.M. Tawil, D. Krishnamurthy, R. Kakimi, M. Ishimaru, S. Emura, S. Hasegawa, H. Asahi, Phys. Status Solidi C 8 (2011) 491.