Self-polarized spin-nanolasers.

ARTICLES PUBLISHED ONLINE: 21 SEPTEMBER 2014 | DOI: 10.1038/NNANO.2014.195

Self-polarized spin-nanolasers Ju-Ying Chen, Tong-Ming Wong, Che-Wei Chang, Chen-Yuan Dong and Yang-Fang Chen* Besides adding a new functionality to conventional lasers, spin-polarized lasers can, potentially, offer lower threshold currents and reach higher emission intensities. However, to achieve spin-polarized lasing emission a material should possess a slow spin relaxation and a high propensity to be injected with spin-polarized currents. These are stringent requirements that, so far, have limited the choice of candidate materials for spin-lasers. Here we show that these requirements can be relaxed by using a new self-polarized spin mechanism. Fe3O4 nanoparticles are coupled to GaN nanorods to form an energy-band structure that induces the selective charge transfer of electrons with opposite spins. In turn, this selection mechanism generates the population imbalance between spin-up and spin-down electrons in the emitter’s energy levels without an external bias. Using this principle, we demonstrate laser emission from GaN nanorods with spin polarization up to 28.2% at room temperature under a low magnetic field of 0.35 T. As the spin-selection mechanism relies entirely on the relative energy-band alignment between the iron oxide nanoparticles and the emitter and requires neither optical pumping with circularly polarized light nor electrical pumping with magnetic electrodes, potentially a wide range of semiconductors can be used as spin-nanolasers.

A

s they are of academic interest and have practical applications, spin-related devices have attracted a great deal of attention1–3. Spin-polarized light sources, especially spinpolarized lasers (spin-lasers), are among the most intriguing spindriven devices4–12. Spin-lasers can provide unique properties superior to those of their conventional counterparts. For instance, in addition to the study of spin characteristics in semiconductor heterostructures, they promise reduced threshold currents5,6,8, enhanced emission intensities9, polarization control10,11 and signal modulation with extremely high frequencies and enhanced bandwidths12–14. Therefore, spin-lasers possess an excellent potential for mass application, including secure communication7,15, chirality studies16, highspeed modulators17, reconfigurable optical interconnections and advanced optical devices6. Nowadays, spin-lasers require spin injection using electrical pumping by a magnetic electrode or optical pumping by a circularly polarized light source7,10,18. Most of these studies focused mainly on quantum wells based on GaAs because of its large spin–orbit splitting19. The use of other materials in spin-lasers is limited by their weak spin–orbit interaction. For example, wurtzite semiconductors (such as GaN and ZnO) have a smaller output spin polarization than those of zinc blende semiconductors (such as GaAs and GaP)20. This difficulty limits the construction of spin-lasers based on nitride semiconductors, although they represent one of the most important material systems in the lighting industry21. The largest output from the spin-polarized spontaneous emission of GaN is predicted to be about 3% even if the injected electrons are completely spin polarized20. Here, we propose to overcome these limitations and achieve nitride semiconductor spin-nanolasers using an approach based on a self-polarized spin orientation mechanism.

Design and materials characteristics The newly designed spin-lasers, which do not need electrical pumping by a ferromagnetic spin aligner or optical pumping by circularly polarized light source22,23, are made with a composite that consists of nanostructured semiconductors (GaN nanorods) and half-metal nanoparticles (Fe3O4 nanoparticles). Owing to the unique band structures between GaN nanorods and Fe3O4 nanoparticles, different band alignments for electrons with spin-up

and -down states are created, as shown in Fig. 1. As a result, selective transfer of spin-polarized electrons between GaN nanorods and Fe3O4 nanoparticles can be achieved. It is known that halfmetallic materials have a unique feature in that they behave as a good conductor for electrons with one particular spin orientation, but as a good insulator for electrons with the opposite spin orientation24–26. Thus, the population imbalance of spin-up and -down electrons can be generated in light-emitting materials. Moreover, the transfer of spin-down electrons from the GaN nanorods to Fe3O4 nanoparticles will build up an electrostatic field that will induce the transfer of holes. As the effective mass of heavy holes is much larger than that of light holes perpendicular to c axis in the wurtzite structure, Fe3O4 nanoparticles that have a suitable energy-band alignment can cause the transfer of light holes from GaN nanorods to nanoparticles much more easily. Thus, the population of spin-down electrons and heavy holes in GaN nanorods increases drastically. Notably, the self-polarized spin effect shown here for introducing spin polarization to initially spin-unpolarized material arises from the spin-dependent band alignment of a magnetic material, which is in contrast to spin injection or extraction by magnetic electrode and optical pumping by circularly polarized light. Spin injection and spin extraction across a heterostructure have also been studied theoretically in magnetic/non-magnetic semiconductor p–n junctions27, in which for a spin injection or extraction into/from the non-magnetic region, the system still needs an external bias, similar to the injection of a charge carrier in a typical semiconductor p–n junction. Nevertheless, this study supports the feasibility of a spin-charge transfer across a heterojunction with no external bias. Additionally, spin-filtering effects have been studied intensively recently, including trapping of spin carriers via magnetic centres28,29. For instance, the spin-filtering effect has been achieved through the mediation of paramagnetic defect centres29. When the defect captures a majority spin electron, it selectively filters out the minority spin electrons according to the Pauli exclusion principle. For the spin-filtering effect to occur, injection of spinpolarized carriers is required. Among the semiconductors based on group III nitrides, GaN serves as the pillar of the group III nitride family30. Compared with ternary InGaN alloys, the growth

Department of Physics, National Taiwan University, Taipei 106, Taiwan. * e-mail: [email protected] NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology

© 2014 Macmillan Publishers Limited. All rights reserved.

1

ARTICLES

NATURE NANOTECHNOLOGY a

b

1 eV

e−

e−

e−

EF

e−

GaN

e− EF

Spin-polarized light hν

3.2 eV

HH+

DOI: 10.1038/NNANO.2014.195

HH+

LH+ Fe3O4

LH+

HH+

GaN

Fe3O4

Figure 1 | Schematics of band alignment of spin-down and spin-up electrons between GaN and Fe3O4. a, Band alignment of spin-down electrons between GaN and Fe3O4. The imperfection of spin orientation of the electrons at the Fermi energy of Fe3O4 is shown explicitly to reflect the true situation. b, Band alignment of spin-up electrons between GaN and Fe3O4. Owing to the population imbalance of spin alignment, the recombination of electron and hole generates spin-polarized emissions. EF , Fermi energy; up and down arrows, spin orientation of the electrons; shaded areas, energy states occupied by electrons; HH+, heavy holes; HH−, light holes.

of good-quality GaN films that can produce strong ultraviolet light emission is relatively easy. Among half-metallic materials, Fe3O4 has received considerable interest because of its relatively simple crystallographic structure in comparison to those of other potential candidates31,32. Based on spin-resolved photoelectron spectroscopy, it was found that the observed polarization at the Fermi energy of Fe3O4 is extremely sensitive to the morphology of the surface. A 40% spin polarization has been detected for samples prepared in situ31,32. The band alignment shown in Fig. 1 is plotted according to work functions and band structures published in previous reports33–35. Finally, the nanoscale hexagonal cross-section of GaN nanorods, as shown in Fig. 2, serves as a natural laser cavity for the creation of ultraviolet whispering gallery modes (WGMs)36.

Spin-laser action Figure 3a shows the emission behaviour of orderly GaN nanorod arrays filled with Fe3O4 nanoparticles under excitation energy from 67 µJ to 87 µJ. The spectra only reveal one single mode at 371 nm with a full-width at half-maximum around 1.2 nm. Figure 3b shows the σ+ and σ− emission spectra under a magnetic field of +0.35 T and an excitation energy of 87 µJ. Quite surprisingly, the σ+ and σ− emission intensities have a very pronounced difference with the degree of circular polarization up to a high value of

1 μm

17.5%, which is far beyond the previously reported value20. Meanwhile, the degree of photoluminescence (PL) circular polarization is negligible under zero magnetic field (Supplementary Fig. 1). This observation demonstrates that the circular polarization we measured does not arise from polarization effects of our experimental apparatus. Figure 3c shows the dependence of the integrated emission intensity on the excitation energy of GaN nanorod arrays filled with Fe3O4 nanoparticles. We can clearly see an abrupt change of the slope, which indicates the occurrence of stimulated emission. Indeed, if we carefully examine the evolution of the PL spectra with increasing pumping energy, a drastic enhancement of PL intensity can be observed (Supplementary Fig. 2), which provides a clear signature for stimulated emission. Interestingly, the σ− integrated emission intensity is larger than that of σ+. Also, the trend of the σ+ integrated lasing intensity caused by the transition of spindown electrons increases sublinearly as the excitation energy increases, but that of σ− caused by the transition of spin-up electrons increases linearly. This result shows that the degree of circular polarization increases with the excitation intensity. In addition, the threshold of the σ− emission intensity is lower than that of the σ+ emission intensity. Figure 3d shows the corresponding degrees of circular polarizations under different magnetic fields. Two phenomena were found. One is that the degree of circular polarization increases when the external magnetic field increases, and the second is that the rising trend saturates when the magnetic field exceeds 0.35 T. These two phenomena are similar to the magnetization hysteresis loop of Fe3O4 nanoparticles shown in Supplementary Fig. 3c. Furthermore, the handedness of the circular polarization changes with magnetic field reversal, which indicates that the observed behaviour arises from a spin effect. The observations described above illustrate that the spin-injection mechanism does not arise from electrical or optical pumping. To explore the observed laser action, we examined several optical microcavities, including WGM, quasi-WGM and two kinds of Fabry–Pérot, as shown in Supplementary Fig. 4a–d, respectively. Here the WGM modes represent the fact that the light waves are almost perfectly guided a round by total internal reflection and interfere with themselves when they have completed one full circulation within the GaN nanorod cavity. It was found that the WGM can be used to describe our result quite well. Supplementary Fig. 3e shows the classical plane-wave model for the WGM cavity37. When the mode number N = 1, the resonance wavelength is 371 nm, which agrees with the peak position shown in Fig. 3a.

Underlying mechanism of self-polarized spin-laser action Figure 2 | Scanning electron microscope image of the top view of GaN nanorod arrays. 2

To understand the spin-laser characteristics and the underlying mechanism, we utilize the random lasing in cluttered GaN

NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology


NATURE NANOTECHNOLOGY a

365

370 Wavelength (nm)

375

d Degree of circular polarization (%)

σ−det σ+det

14.0 PL intensity ×103 (a.u.)

360

380

18.0 16.0

σ−det σ+det PL intensity (a.u.)

PL intensity (a.u.)

c

b

87 μJ 81 μJ 79 μJ 75 μJ 73 μJ 70 μJ 67 μJ

360

ARTICLES

DOI: 10.1038/NNANO.2014.195

12.0 10.0 8.0 6.0 4.0 2.0 0.0

365

370 Wavelength (nm)

375

380

20 15 10 5 0 −5 −10 −15 −20

60

65

70

75

80

85

90

−6

−5

−4

−3

−2

−1

0

1

2

3

4

5

Magnetic field (kOe)

Power (μJ)

Figure 3 | Characteristics of spin-polarized laser action in GaN nanorods filled with Fe3O4 nanoparticles at room temperature. a, PL spectra of GaN nanorod arrays filled with Fe3O4 nanoparticles above and near the lasing threshold. b, PL spectra of GaN nanorod arrays filled with Fe3O4 nanoparticles under a magnetic field of 0.35 T and 87 µJ pumping energy. c, The dependence of integrated PL intensity on the excitation energy for GaN nanorod arrays filled with Fe3O4 nanoparticles under different pumping energies. The arrows indicate the different threshold pumping energies for σ− and σ+ circular polarization. The solid lines are linear fits. d, The degree of circular polarization of the spin-laser under different magnetic fields. a.u., arbitrary units; σ −det, emission with σ− polarization; σ+det, emission with σ+ polarization.

nanorocks grown by metal organic chemical vapour deposition (MOCVD) as the reference sample. The top-view scanning electron microscope (SEM) image of the cluttered GaN nanorocks is shown in Supplementary Fig. 5 and the corresponding random laser action is displayed in Fig. 4a,b. The SEM image indicates that the GaN nanorocks have an irregular shape with a large size distribution. The maximum emission peak is located at around 373 nm, but the circularly polarized phenomenon does not occur for the GaN nanorocks filled with Fe3O4 nanoparticles under a magnetic field of 0.35 T. According to the working principle of random lasing, the degree of circularly polarized light will be suppressed by the multiple reflection processes during the formation of a random cavity for the laser action to occur. In addition, the disappearance of the circular polarization of the emission below the pumping threshold can be attributed to the irregular shape of GaN nanorocks with a large size variation and multiple reflections. Therefore, this result can serve as additional evidence that the measured circular polarization is not caused by systematic errors arising from our experimental measurements. It also confirms the importance of the periodicity of GaN nanorods for the detected polarization. Based on this result, we can also exclude the possibility that the magnetic circular dichroism of Fe3O4 nanoparticles is responsible for the measured spin-polarized signal. This is because, before being collected, the light beam has to pass through the magnetic nanoparticles that cover the GaN nanorocks. To rule out the influence of the magnetic circular dichroism of Fe3O4 nanoparticles conclusively, we prepared a sample with a Fe3O4 nanoparticle

layer of thickness about 3 µm, which is much larger than the thickness of nanorocks. No indication of output spin polarization was found. To explain the origin of the circular polarization in the composite of the periodic GaN nanorod array and Fe3O4 nanoparticles, let us consider the band alignment as shown in Fig. 1. First, we discuss the behaviour of electrons. In an n-type semiconductor, the flow of electrons dominates the current. It is known that, as a result of the unique band structure of Fe3O4 , for spin-down electrons Fe3O4 acts as a conductor, but for spin-up electrons it behaves as an insulator. Therefore, according to the band alignment between GaN nanorods and Fe3O4 nanoparticles25,26,33, when extra electrons in GaN nanorods are caused by external perturbations, spin-down electrons can overcome the small Schottky barrier between the Fe3O4 nanoparticles and GaN nanorods and flow into the Fe3O4 nanoparticles, but spin-up electrons remain in the GaN nanorods. As a result, the population of spin-up electrons in the nanorods is more than that of spin-down electrons, and circularly polarized radiation can be generated. At the same time, this mechanism can also be used to explain the sublinear and linear behaviours of the σ+ and σ− integrated lasing intensities shown in Fig. 3c. Based on the band alignment shown above, the number of spin-down electrons that flow into the Fe3O4 nanoparticles increases as the excitation energy increases, but the number of spin-down electrons that stay in GaN nanorods increases linearly with the excitation energy. Therefore, the trend of the σ+ integrated lasing intensity increases sublinearly, and that of σ− light intensity increases linearly.



3

ARTICLES

NATURE NANOTECHNOLOGY

a

GaN/Fe3O4-σ−det GaN/Fe3O4-σ+det GaN

1.0 Normalized PL intensity (a.u.)

PL intensity (a.u.)

74 μJ 78 μJ 85 μJ 97 μJ

DOI: 10.1038/NNANO.2014.195

0.8 0.6 0.4 0.2 0.0

360

b

370 375 Wavelength (nm)

380

385

σ+det

12.0 10.0 8.0 6.0 4.0 2.0 0.0 60

70

80 Power (μJ)

90

100

Figure 4 | Characteristics of random laser action in GaN nanorocks filled with Fe3O4 nanoparticles. a, PL spectra of GaN nanorocks filled with Fe3O4 nanoparticles above the lasing threshold. b, Dependence of the integrated emission intensity on the excitation energy of GaN nanorocks filled with Fe3O4 nanoparticles.

Consequently, the degree of circular polarization increases as the excitation energy increases. For a conventional spin-laser, the operation is based on an optical and/or electrical spin injection10,18. For the operation regime above the threshold, the carrier density is strongly clamped and is more or less constant with respect to the excitation energy. Therefore, it is expected that the output polarization will increase and then decrease after the threshold for the lower gain mode is crossed11. However, in our case the population imbalance of spin-down and spin-up electrons arises from the selective spin-charge transfer of the unique band alignment between the active layer and magnetic nanoparticles, which is different from conventional optical and electrical spin injections. It is known that the charge transfer across the interface of a heterojunction strongly depends on the built-in electric field. The built-in electric field can be screened effectively by the photogenerated electrons and holes because of the optical excitation, which has been reported in several previous publications for nitride semiconductors38. As a result of the reduction of the potential barrier for the charge transfer because of optical screening, the number of spin-down electrons that flow into the Fe3O4 nanoparticles therefore increases with increasing excitation energy. Next, we discuss the behaviour of holes. The transfer of spin-down electrons from the GaN nanorods to Fe3O4 nanoparticles builds up an electrostatic field that will induce the transfer of holes. As the effective mass of heavy holes is much larger than that of light holes perpendicular to the nanorods, light holes can transfer from the nanorods to the nanoparticles much more easily than heavy holes, which increases the population of heavy holes in the nanorods. This result 4

−0.2

0.0

0.2 0.4 Time (ns)

0.6

0.8

Figure 5 | Normalized time-resolved photoluminescence spectra of pure GaN nanorods and GaN nanorods filled with Fe3O4 nanoparticles. A magnetic field of 0.35 T was used.

σ−det

14.0 PL intensity ×103 (a.u.)

365

therefore breaks the previously reported value of 3% for the degree of circular polarization in nitride semiconductors21. To support the mechanism proposed above, we performed timeresolved photoluminescence (TRPL) measurements. Figure 5 shows the TRPL spectra of pure GaN nanorods and GaN nanorods filled with Fe3O4 nanoparticles under a magnetic field of 0.35 T excited by a 260 nm pulse laser with a spectra resolution of about 10 ps. Quite interestingly, the TRPL spectra also reveal a pronounced degree of circular polarization. We can clearly see that the lifetime (40 ps) of σ+ emission is much shorter than that of σ− emission (95 ps). Also, the lifetime of σ− emission is very similar to that of the emission from pure GaN nanorod array. This result reflects that spin-up and spin-down electrons experience a very different environment, and provides concrete evidence for our proposed mechanism described in Fig. 1, in that there exists an additional channel for spin-down electrons to flow out. Supplementary Fig. 6a,b shows PL spectra of GaN nanorods with and without Fe3O4 nanoparticles, respectively, excited by a circularly polarized laser under a magnetic field of 0.35 T. The PL spectra in Supplementary Fig. 6a reveal a large degree of circular polarization for the bandgap as well as defect emission. The origin of the peak around 550 nm can be attributed to the optical transition between the conduction electrons and nitrogen antisite complex defects39. The presence of such a peak indicates that the number of spin-up electrons remaining in GaN nanorods excited by a σ− pulse laser is larger than the number of spin-down electrons remaining in GaN nanorods excited by a σ+ pulse laser when GaN nanorods are covered with Fe3O4 nanoparticles. This is because the spindown electrons excited by a σ+ pulse laser can easily flow from GaN nanorods to Fe3O4 nanoparticles, a phenomenon that does not occur in pure GaN nanorods, as shown in Supplementary Fig. 6b. The very small difference in the bandgap emission in pure GaN nanorods excited by σ− and σ+ pulse lasers can be attributed to the difference in the oscillator strength of σ− and σ+ optical transitions in the GaN material20. Meanwhile, the completely negligible spin polarization for the defect emission arises from the indistinguishable optical defect states caused by the spin-flip scattering of the inherent nature of defects40.

Electrically tunable output circular polarization Additional evidence to support our proposed mechanism is given in Fig. 6. Figure 6a shows a schematic for the applied bias to the measure the PL spectra for GaN nanorod arrays filled with Fe3O4 nanoparticles. Figure 6b shows the PL spectra under different biases. It is clear that the intensity of PL can be controlled by an external bias. With a positive bias, the PL intensity increases, but




ARTICLES

DOI: 10.1038/NNANO.2014.195

a

b

Earth

10 V 0V

1,500 Fe3O4 nanoparticle PL intensity (a.u.)

−10 V

GaN nanorod

1,000

500

0 360

365

370

375

380

385

Wavelength (nm)

c σ−det

2,000

d

e−

1,500

1,000

500

Fe3O4

GaN

σ−det σ+det

1,000 PL intensity (a.u.)

PL intensity (a.u.)

σ+det

e−

500

Fe3O4

GaN 0

0 350

355

360

365 370 375 Wavelength (nm)

380

385

350

355

360

365 370 375 Wavelength (nm)

380

385

Figure 6 | Spin-polarized emission spectra under different external biases. a, Schematic of the applied bias for the measurement of PL spectra of GaN nanorod arrays filled with Fe3O4 nanoparticles. b, PL spectra under a different external bias and a magnetic field of 0.35 T. c,d, PL spectra and the band bending (inset) of GaN nanorod arrays filled with Fe3O nanoparticles under +10 V (c) and –10 V (d) and a magnetic field of 0.35 T. Spin polarization is 8.66% in c and 28.2% in d.

the PL intensity decreases under a negative bias. This behaviour can be understood based on our proposed mechanism. The external bias can be used to tune the band bending upwards or downwards across the interface between the GaN nanorods and Fe3O4 nanoparticles and therefore control the electron flow and manipulate the emission intensity. In addition, as a consequence of the flow of electrons with selective spin, the polarization of PL spectra can also be manipulated, as shown in Fig. 6a,b. When an external bias of –10 V is provided to bend the band alignment, the degree of circular polarization can reach up to 28.2%. We stress that the output spin polarization and intensity tunable by an external bias provides an excellent route to control and modulate the characteristics of spin-lasers. This offers a very attractive feature that makes our developed spin-lasers more feasible for several practical applications, such as secure communications, high-speed modulators and chiroptical spectroscopies41, when compared with those derived from electrical or optical spin injections. Additionally, the tunable feature of the output spin polarization by an external bias can also be used to rule out that the magnetic circular dichroism of Fe3O4 nanoparticles is responsible for our observed spin effect.

Conclusions In summary, we describe here an efficient spin-laser that, unlike all previous reports5,7,41,42, requires neither spin injection by a magnetic electrode nor circularly polarized light. Rather, our system relies on hexagonal GaN nanorods, which serve as a whispering gallery mode cavity, and the different energy-band alignments for spin-up and spin-down electrons between the nanorods and Fe3O4 nanoparticles to increase spontaneously the number of electrons with a particular spin orientation in the semiconductor emitter. We report a spin polarization of 28.2%, which needs to be improved before practical

applications are considered. In this respect, previous work6,11,13,14 showed that external perturbations near or exceeding threshold conditions can induce a large gain anisotropy because of the nonlinear spin dynamics in the active region. This can be understood easily when the injection of spin-polarized electrons is slightly below threshold, as a perturbation that leads to the injection of spin-polarized electrons will favour enhancement of the gain of one polarized emission above threshold, and the gain of the other polarized emission will remain subthreshold. These effects should provide some control to manipulate the light polarization in spin-lasers using, for example, birefringence, strain, temperature and geometric dimensionality, and should improve spin polarization. In addition, our spin-laser allows an external bias to control and modulate the output spin polarization and intensity. The requirement of an external magnetic field may be removed by depositing a ferromagnetic thin film near the half-metallic material. Under this circumstance, the major role played by the ferromagnetic film would not be to serve as a spin injector, but rather to provide the necessary magnetic field for the operation of the spin-laser. Finally, the approach demonstrated here can be extended to other composite systems43 and, with a better choice of higher spin polarization of half-metallic materials42, it is expected that the efficiency of the selective spin-carrier transfer and the output polarization of spin-lasers can be enhanced further.

Methods

Optical measurement. To investigate their laser action, the devices were optically excited by a Q-switched Nd:yttrium aluminium garnet laser (266 nm, 35 ns pulse, 10 Hz) focused to a beam size of about 200 µm in diameter. All PL and lasing spectra were measured at room temperature, and the magnetic field was provided by a magnet of strength 0.35 T. Unless specified otherwise, all the PL spectra were excited by a linearly polarized light beam. We define the +z direction to be parallel to the direction of light propagation and the magnetic field, and perpendicular to the



5

ARTICLES


sample plane. In this arrangement, the spectra were measured in the Faraday geometry. Let σ+ and σ− represent the helicity of light arriving in the +z direction. For outgoing light in the –z direction, the helicity is reversed. The degree of the polarization of emission is defined by P = (Iσ+ – Iσ−)/(Iσ+ + Iσ−), where Iσ+ (right circular polarization) and Iσ− (left circular polarization) are the integrated intensities of the emission peaks for the σ+ and σ− circular polarization, respectively. Sample fabrication. The growth of periodic GaN nanorod arrays was patterned with a nanoimprint technique on a 2 µm thick GaN template, which was deposited by MOCVD on a c-plane sapphire substrate44. The top view of the GaN nanorods recorded by an SEM (JSM-6500F, JEOL) is shown in Fig. 2. The nanorods were estimated to have an average diameter of 250 nm and an average length of 1 µm. A more detailed description of the synthesis process is given elsewhere44. The vacancies between the GaN nanorods were filled by Fe3O4 nanoparticles. Monodispersed hydrophobic Fe3O4 nanoparticles were synthesized by a seed-growth method using the high-temperature solution-phase reaction of Fe(acac)3 in benzyl ether in the presence of 1,2-dodecanediol, oleic acid and oleylamine. A more detailed description of the synthesis process is given elsewhere45. Supplementary Fig. 3a shows the morphology of the Fe3O4 nanoparticles taken by a transmission electron microscope (JSM-1200 EX II, JEOL) operated at 100 kV, and the diameter of the Fe3O4 nanoparticles was approximately 10 nm. The top view of the composite that consists of a GaN nanorod array and Fe3O4 nanoparticles taken by SEM is shown in Supplementary Fig. 3b. It reveals that GaN nanorods were almost covered by Fe3O4 nanoparticles. The magnetization hysteresis loop of the Fe3O4 nanoparticles was measured at room temperature, and is shown in Supplementary Fig. 3c. This indicates that the Fe3O4 nanoparticles have a pronounced superparamagnetic characteristic.

Received 24 February 2014; accepted 11 August 2014; published online 21 September 2014

References 1. Datta, S. & Das, B. Electronic anlog of the electro-optic modulator. Appl. Phys. Lett. 56, 665–667 (1990). 2. Prinz, G. A. Magnetoelectronics. Science 282, 1660–1663 (1998). 3. Awschalom, D. D., Samarth, N. & Loss, D. Semiconductor Spintronics and Quantum Computation (Springer, 2002). 4. Hallstein, S. et al. Manifestation of coherent spin precession in stimulated semiconductor emission dynamics. Phys. Rev. B 56, R7076–R7079 (1997). 5. Rudolph, J. et al. Room-temperature threshold reduction in vertical-cavity surface-emitting lasers by injection of spin-polarized electrons. Appl. Phys. Lett. 87, 241117 (2005). 6. Gothgen, C. et al. Analytical model of spin-polarized semiconductor lasers. Appl. Phys. Lett. 93, 042513 (2008). 7. Holub, M. & Bhattacharya, P. Spin-polarized light-emitting diodes and lasers. J. Phys. D 40, 179–203 (2007). 8. Holub, M. & Jonker, B. T. Threshold current reduction in spin-polarized lasers: role of strain and valence-band mixing Phys. Rev. B 83, 125309 (2011). 9. Basu, D. et al. Electrically injected InAsGaAs quantum dot spin-laser operating at 200K. Appl. Phys. Lett. 92, 091119 (2008). 10. Hovel, S. et al. Optical spin manipulation of electrically pumped vertical-cavity surface-emitting lasers Appl. Phys. Lett. 92, 041118 (2008). 11. Basu, D., Saha, D. & Bhattacharya, P. Optical polarization modulation and gain anisotropy in an electrically injected spin-laser. Phys. Rev. Lett. 102, 093904 (2009). 12. Li, M. Y. et al. Birefringence controlled room-temperature picosecond spin dynamics close to the threshold of vertical-cavity surface-emitting laser devices. Appl. Phys. Lett. 97, 191114 (2010). 13. Gerhardt, N. C. et al. Ultrafast spin-induced polarization oscillations with tunable lifetime in vertical-cavity surface-emitting lasers. Appl. Phys. Lett. 99, 15117 (2011). 14. Boéris, G., Lee, J., Výborný, K. & Žutić, I. Tailoring chirp in spin-lasers. Appl. Phys. Lett. 100, 121111 (2012). 15. Jonker, B. T. Progress toward electrical injection of spin-polarized electrons into semiconductors. IEEE Proc. 91, 727–740 (2003). 16. Blackmond, D. G. The origin of biological homochirality. Phil. Trans. R. Soc. B 366, 2878–2884 (2011). 17. Awschalom, D. D., Warnock, J. & von Molnár, S. Low-temperature magnetic spectroscopy of a dilute magnetic semiconductor. Phys. Rev. Lett. 58, 812–815 (1987). 18. Gerhardt, N. C. et al. Electron spin injection into GaAs from ferromagnetic contacts in remanence. Appl. Phys. Lett. 87, 032502 (2005). 19. Oestreich, M. et al. Spin injection, spin transport and spin coherence. Semicond. Sci. Technol. 17, 285–297 (2002). 20. Chen, W. M. et al. Efficient spin relaxation in InGaN/GaN and InGaN/GaMnN quantum wells: an obstacle to spin detection. Appl. Phys. Lett. 84, 192107 (2005). 21. Buyanova, I. A. et al. On the origin of spin loss in GaMnN/InGaN light-emitting diodes. Appl. Phys. Lett. 84, 2599–2601 (2004).

6

DOI: 10.1038/NNANO.2014.195

22. Fujino, H. et al. Circularly polarized lasing in a (110)-oriented quantum well vertical-cavity surface-emitting laser under optical spin injection. Appl. Phys. Lett. 94, 131108 (2009). 23. Satoshi, I., Shinji, K., Kazuhiro, I. & Hitoshi, K. Room temperature circularly polarized lasing in an optically spin injected vertical-cavity surface-emitting laser with (110) GaAs quantum wells. Appl. Phys. Lett. 98, 081113 (2011). 24. Hu, G. & Suzuki, Y. Negative spin polarization of Fe3O4 in magnetite/manganite-based junctions. Phys. Rev. Lett. 89, 276601 (2002). 25. Akira, Y. & Noriaki, H. Electronic structure in high temperature phase of Fe3O4. J. Phys. Soc. Jpn 68, 1607–1613 (1999). 26. Tang, J., Myers, M., Bosnick, K. A. & Brus, L. E. Magnetite Fe3O4 nanocrystals: spectroscopic observation of aqueous oxidation kinetics. J. Phys. Chem. B 107, 9501–7506 (2003). 27. Žutić, I., Jaroslav, F. & Das S. D. Spin-polarized transport in inhomogeneous magnetic semiconductors: theory of magnetic/nonmagnetic p–n junctions. Phys. Rev. Lett. 88, 066603 (2002). 28. Rinaldi, C. et al. Ge-based spin-photodiodes for room-temperature integrated detection of photon helicity. Adv. Mater. 24, 3037–3041 (2012). 29. Puttisong, Y. et al. Efficient room-temperature spin detector based on GaNAs. J. Appl. Phys. 111, 07C303 (2012). 30. Michael, E. L., Sergey, L. R. & Michael, S. S. Properties of Advanced Semiconductor Materials: GaN, AIN, InN, BN, SiC, SiGe (John Wiley & Sons, 2001). 31. Alvarado, S. F. et al. Observation of spin-polarized electron levels in ferrites. Phys. Rev. Lett. 34, 319–322 (1975). 32. Morton, S. A. et al. Spin-resolved photoelectron spectroscopy of Fe3O4. Surf. Sci. 513, L451–L457 (2002). 33. Fonin, M. et al. Surface electronic structure of the Fe3O4(100): evidence of a half-metal to metal transition. Phys. Rev. B 72, 104436 (2005). 34. Casey, H. C. et al. Low interface trap density for remote plasma deposited SiO2 on n-type GaN. Appl. Phys. Lett. 68, 1850–1852 (1996). 35. Camphausen, D. L., Coey, J. M. D. & Chakraverty, B. K. One-electron energy levels in Fe3O4. Phys. Rev. Lett. 29, 657–660 (1972). 36. Kouno, T., Kishino, K. & Sakai, M. Lasing action on whispering gallery mode of self-organized GaN hexagonal microdisk crystal fabricated by RF-plasmaassisted molecular beam epitaxy. IEEE J. Sel. Top. Quant. Electron 47, 1565–1570 (2011). 37. Chen, R., Ling, B., Sun, X. W. & Sun, H. D. Room temperature excitonic whispering gallery mode lasing from high-quality hexagonal ZnO microdisks. Adv. Mater. 23, 2199–2204 (2011). 38. Wei, C. M., Shih, H. Y., Chen, Y. F. & Lin, T. Y. Optical detection of magnetoelectric effect in the composite consisting of InGaN/GaN multiple quantum wells and FeCo thin film. Appl. Phys. Lett. 98, 131913 (2011). 39. Gohda, Y. & Oshiyama, A. Intrinsic ferromagnetism due to cation vacancies in Gd-doped GaN: first-principles calculations. Phys. Rev. B 78, 161201 (2008). 40. Chen, H. M., Chen, Y. F., Lee, M. C. & Feng, M. S. Yellow luminescence in n-type GaN epitaxial films. Phys. Rev. B 56, 6942–6946 (1997). 41. Evgeny, Y. T. & Zutic, I. Handbook of Spin Transport and Magnetism (CRC Press, 2011). 42. Žutić, I., Fabian, J. & Das Sarma, S. Spintronics: fundamentals and applications. Rev. Mod. Phys. 76, 323–410 (2004). 43. Chen, J. Y. et al. Efficient spin-light emitting diodes based on InGaN/GaN quantum disks at room temperature: a new self-polarized paradigm. Nano Lett. 14, 3130 (2014). 44. Liao, C. H. et al. Geometry and composition comparisons between c-plane disc-like and m-plane core-shell InGaN/GaN quantum wells in a nitride nanorod. Opt. Express 20, 15859–15871 (2012). 45. Wu, S. H. et al. PEGylated silica nanoparticles encapsulating multiple magnetite nanocrystals for high-performance microscopic magnetic resonance angiography. J. Biomed. Mater. Res. 98B, 81–88 (2011).

Acknowledgements This work was supported by the National Science Council and the Ministry of Education of the Republic of China. We thank W. Chen and C-Y. Mou for providing the Fe3O4 nanoparticles, and C-H. Liao and C-C. Yang for the fabrication of GaN nanomaterials.

Author contributions J-Y.C. and Y-F.C. conceived the idea of spin self-polarized nanocomposites and designed the experiments. J-Y.C., T-M.W. and C-W.C. fabricated and measured the samples. All the authors were involved in analysing the data and writing the manuscript.

Additional information Supplementary information is available in the online version of the paper. Reprints and permissions information is available online at www.nature.com/reprints. Correspondence and requests for materials should be addressed to Y-F.C.

Competing financial interests

The authors declare no competing financial interests.



Self-polarized spin-nanolasers.

Recommend Documents

Follow Us