Room temperature single GaN nanowire spin valves with FeCo/MgO tunnel contactsHyun Kum, Junseok Heo, Shafat Jahangir, Animesh Banerjee, Wei Guo, and Pallab Bhattacharya

Citation: Applied Physics Letters 100, 182407 (2012); doi: 10.1063/1.4711850 View online: http://dx.doi.org/10.1063/1.4711850 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/100/18?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Maximum magnitude in bias-dependent spin accumulation signals of CoFe/MgO/Si on insulator devices J. Appl. Phys. 114, 243904 (2013); 10.1063/1.4856955 Thermal spin injection and accumulation in CoFe/MgO tunnel contacts to n-type Si through Seebeck spintunneling Appl. Phys. Lett. 103, 142401 (2013); 10.1063/1.4823540 Transport of perpendicular spin in a semiconductor channel via a fully electrical method Appl. Phys. Lett. 102, 062412 (2013); 10.1063/1.4792690 Effect of the interface resistance of CoFe/MgO contacts on spin accumulation in silicon Appl. Phys. Lett. 100, 252404 (2012); 10.1063/1.4728117 Non-local detection of spin-polarized electrons at room temperature in Co50Fe50/GaAs Schottky tunnel junctions Appl. Phys. Lett. 99, 082108 (2011); 10.1063/1.3630032

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

109.148.186.166 On: Fri, 28 Mar 2014 18:39:55

Room temperature single GaN nanowire spin valves withFeCo/MgO tunnel contacts

Hyun Kum,1 Junseok Heo,1 Shafat Jahangir,1 Animesh Banerjee,1 Wei Guo,2

and Pallab Bhattacharya1,a)

1Center for Photonic and Multiscale Nanomaterials, Department of Electrical Engineering and ComputerScience, University of Michigan, Ann Arbor, Michigan 48109-2122, USA2Department of Electrical and Computer Engineering, University of Michigan-Dearborn, 4901 EvergreenRoad, Dearborn, Michigan 48128-2406, USA

(Received 15 April 2012; accepted 20 April 2012; published online 4 May 2012)

We report the direct measurement of spin transport characteristics in a GaN spin valve, with a

relatively defect-free single GaN nanowire (NW) as the channel and FeCo/MgO as the tunnel

barrier spin contact. Hanle spin precession and non-local transport measurements are made in

an unintentionally doped nanowire spin valves. Spin diffusion length and spin lifetime values

of 260 nm and 100 ps, respectively, are derived. Appropriate control measurements have been

made to verify spin injection, transport, and detection. VC 2012 American Institute of Physics.

[http://dx.doi.org/10.1063/1.4711850]

Wide bandgap semiconductors such as GaN and their

alloys are important for high-power electronics, solid-state

lighting, and more recently for studies on strong coupling

and polariton lasing.1–4 GaN crystallizes in the wurtzite or

zincblende forms and has inversion asymmetry. It is also

characterized by a weak spin-orbit coupling (SOC), which

makes it attractive for high temperature spintronics.5,6 Pre-

dictions of long spin lifetimes in GaN have been made from

theoretical calculations.7 Measurements of electron spin life-

times in bulk wurtzite GaN have been made by time-

resolved Kerr-rotation and time-resolved Faraday rotation

spectroscopy, and values of the parameter at room tempera-

ture ranging from 35 to 75 ps have been derived.8,9 It has

been shown recently by several groups, including ours, that

wurtzite GaN nanowires (NWs) can be grown epitaxially on

(001) or (111) silicon substrate with almost a complete ab-

sence of extended defects such as dislocations, stacking

faults, and twins.10–14 Measurements of fundamental mate-

rial parameters have been made with such nanowires and

they have been incorporated as the active region in the

design and fabrication of photon and polariton lasers, light-

emitting diodes, and electronic devices.13,15–18 The measure-

ment of spin transport parameters in the nanowires would

yield the intrinsic values of the parameters in GaN and would

serve as a standard for future reference.

Measurements on a device such as a spin valve involve

successful spin injection and detection in the semiconductor

channel via ferromagnetic contacts. Schottky and oxide tun-

nel contacts, a solution for the impedance mismatch prob-

lem19 proposed by Rashba,20,21 have been very successful

for injecting spin polarized carriers in GaAs-, InP-, Si-, and

Ge-based spintronic devices.22–29 An ultra thin tunnel barrier

between the ferromagnet and semiconductor provides a large

interface resistance with high spin asymmetry—a require-

ment for efficient spin injection. Ferromagnetic FeCo, to-

gether with MgO as the tunnel barrier, have demonstrated

the highest spin injection efficiency of 32% into GaAs at

300 K.30,31 We have used this tunnel contact in single wurt-

zite GaN NW spin valves. Measurements have been made as

a function of transport length in the nanowire in the longitu-

dinal direction between the ferromagnetic contacts. Analysis

of the data yields a longitudinal spin relaxation time as high

as �100 ps and a corresponding spin diffusion length of

�260 nm at 300 K. Four-terminal non-local magnetoresist-

ance (MR) and Hanle spin precession measurements were

performed to confirm spin injection into the nanowires.

Several GaN NW samples were epitaxially grown on

(001) Si substrate in a plasma-assisted molecular beam epi-

taxy (PA-MBE) system, with length and diameter ranging

from 2–4 lm and 50–80 nm, respectively. The general

growth steps and conditions are as follows. First, the surface

oxide on the Si substrate is removed in a solution of HF-H2O

and annealed in the growth chamber at a temperature of

900 �C, after which the temperature is lowered to 800 �C and

a few monolayers of Ga are deposited with a Ga flux of

1.5� 10�7 Torr in the absence of nitrogen. The GaN NW

growth is then initiated at the same substrate temperature at

a rate of 300 nm/h under nitrogen-rich conditions. Steady Ga

flux and nitrogen flow rate are maintained at 1.5� 10�7 Torr

and 1 sccm, respectively. A scanning electron microscope

(SEM) image of a grown sample and a high-resolution

transmission electron microscopy (HR-TEM) of a single

nanowire are shown in Figs. 1(a) and 1(b), respectively. The

nanowires were not intentionally doped. However,

capacitance-voltage (C-V) measurements indicate a back-

ground n-doping density of �1� 1017 cm�3.32 The nano-

wires are dispersed by drop casting a low density mixture of

isopropyl alcohol and nanowires on a silicon wafer covered

with 200 nm SiO2 formed by thermal oxidation at the surface

of the wafer. Single nanowires are identified with the help of

a grid mask with alignment marks and SEM imaging. Four

contact regions are defined on a single nanowire by electron-

beam lithography. Finally, 1 nm MgO tunnel barrier and

60 nm FeCo are deposited by e-beam evaporation to form

the ferromagnetic tunnel contacts. The sample is then

annealed at a temperature of 400 �C for approximately 2 h.a)E-mail: pkb@eecs.umich.edu.

0003-6951/2012/100(18)/182407/4/$30.00 VC 2012 American Institute of Physics100, 182407-1

APPLIED PHYSICS LETTERS 100, 182407 (2012)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

109.148.186.166 On: Fri, 28 Mar 2014 18:39:55

This annealing step was found to be crucial in forming an

appropriate tunnel barrier for spin injection. Several ferro-

magnet/NW/ferromagnet (F/N/F) spin valves were fabricated

with varying channel lengths ranging from 200 nm to

3.5 lm. A SEM of a completely fabricated device is shown

in Fig. 1(c). Control devices with non-ferromagnetic Ti/Au

detector contacts (F/N/N) were also fabricated. The magnet-

ization characteristics of the FeCo films were investigated by

magneto-optic Kerr effect (MOKE) measurements. Since

this measurement made directly on the FeCo contact pads

would not yield accurate results (laser spot size of MOKE is

much larger than the size of the pads), it was done, instead,

on a �70 nm FeCo film deposited on SiO2. The measured

data are shown in Fig. 1(d). The applied magnetic field is

swept in-plane, parallel to the film surface. The hysteresis

exhibits sharp magnetization switching characteristics and a

coercivity of �100 Oe. The latter value is dependent on the

thickness and lateral dimensions of the FeCo film.

Figure 2(a) shows the typically measured two-terminal

I-V characteristics of a single nanowire at different tempera-

tures. Also shown are the I-V characteristics of the SiO2

insulating layer. A non-linear variation of bias-dependent

current through the NW is observed. The zero-bias resistance

(ZBR), R0(T)/R0(300 K), exhibits weak insulator-like de-

pendence on temperature, verifying spin injection into the

NW via single step tunneling (Fig. 2(b)). The temperature

dependence of ZBR is known to be a reliable indicator of

tunneling transport.33 The SiO2 layer, which does not exhibit

any conductive breakdown up to 80 V, provides a good insu-

lating platform for the nanowire spin valve and our

experiments.

Magnetoresistance and spin accumulation measurements

were made on the GaN nanowire devices with contacts in

non-local spin valve configuration. The measurements were

made, using a standard four-probe ac lock-in technique, as a

function of channel length at 300 K in a closed loop He cryo-

stat. Samples of different channel lengths are loaded in the

cryostat, which is then mounted between the poles of an

electromagnet such that the magnetic field is applied in-

plane and orthogonal to the direction of spin transport. A

FIG. 1. (a) Scanning electron microscopy

(SEM) image of epitaxially grown nanowires

by PA-MBE on a (001) Si substrate. (b) HR-

TEM image of a single NW. Inset shows the

diffraction pattern of the nanowire. (c) A top-

down SEM view of the lateral spin valve fabri-

cated on a �4 lm long nanowire using e-beam

lithography. (d) Ferromagnetic hysteresis meas-

ured by the MOKE on a bulk 70 nm thick FeCo

film e-beam evaporated on SiO2. The magnetic

field is swept in-plane, parallel to the film

surface.

FIG. 2. (a) Two-terminal I-V characteristics of a single NW at various tem-

peratures (black line). Non-linear I-V characteristics indicate tunneling

transport from the FeCo into the GaN NW through the MgO barrier. (b)

ZBR as a function of temperature. The weak temperature dependence (less

than an order of magnitude) of the ZBR is a good indication of the tunneling

nature of the FeCo/MgO contacts.

182407-2 Kum et al. Appl. Phys. Lett. 100, 182407 (2012)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

109.148.186.166 On: Fri, 28 Mar 2014 18:39:55

schematic illustration of the measurement scheme is shown

in Fig. 3. Non-local MR measurements were made to elimi-

nate the response from possible spurious effects such as ani-

sotropic magnetoresistance (AMR) and local Hall effects,

which may resemble MR behavior arising from true spin

injection. Results from non-local measurements at T¼ 300 K

for channel lengths L¼ 0.7 lm and 1.5 lm are shown in

Figs. 4(a) and 4(b). The peak accumulation corresponds to a

voltage change (DV� 0.2 mV and 8 mV for L¼ 1.5 lm and

0.7 lm, respectively), and a significant nonlocal baseline re-

sistance was not observed. In contrast, the peak accumula-

tion in the control F/N/N device is negligible.

To further ascertain spin injection in the channel, Hanle

spin precession measurements were made with sample A for

two different nanowire channel lengths at T¼ 300 K. The

Hanle effect is manifested as a change in the non-local

voltage due to the precession (at a Larmor frequency

xL¼ glBBz/�h, where g is the g-factor, lB is the Bohr magne-

ton, and �h is reduced Plank’s constant), and suppression of

spin that is subject to a transverse magnetic field (Bz). It is

measured by first setting the magnetization of contacts 2 and

3 (Fig. 3) either in parallel or antiparallel state, then sweep-

ing an out-of-plane magnetic field while measuring the non-

local voltage (contacts 3 and 4). A constant current is flown

through contacts 1 and 2. Clear precession of spin in the

channel was observed in the spin valves, as shown in Fig. 5,

where the top and bottom branches correspond to parallel

and antiparallel magnetization of contacts 2 and 3. Due to

the relatively short length of the epitaxially grown nanowires

(�4 lm max length) and thus a short channel, we were not

able to observe a full 3p/2 precession. To obtain an estima-

tion of the transverse spin relaxation time, T2, the Hanle data

for a channel length of L¼ 1.5 lm (solid lines in Fig. 5)

were fitted using the equation22

VNL

IInject/ 6

ð10

1ffiffiffiffiffiffiffiffiffiffi4pDtp exp � L2

4Dt

� �cosðxLtÞexp � t

ssf

� �dt;

(1)

where D is the diffusion constant, ssf is the spin lifetime, L is

the distance between the injector and detector electrodes,

and þ (�) sign indicates parallel (antiparallel) magnetization

state of the FM electrodes. Here, D can be approximated by

using the mobility values measured on a bulk GaN film

grown in our lab with the same doping level as the nano-

wires. Invoking Einstein’s relation D¼lkBT/q, we obtain a

diffusion constant of D¼ 10 cm2/V-s. Using a g-factor of 2

for GaN, we derive the spin lifetime ssf� 100 ps, which

translates to a spin diffusion length of ksf� 260 nm at

T¼ 300 K using the relation ksf¼ (Dssf)1/2. Similar results

were obtained for the 0.7 lm channel device.

The channel length dependent non-local measurement

data can be fit independently to estimate ksf using the

equation34

DV

IInject¼ P2qksf

Aexp�L

ksf

� �; (2)

FIG. 3. Schematic illustration of the four-terminal non-local measurement

scheme.

FIG. 4. Nonlocal magnetoresistance characteristics for channel length of (a)

1.5 lm and (b) 0.7 lm for sample A at room temperature for a constant cur-

rent bias Iinject¼ 100 nA.

FIG. 5. Four-terminal Hanle precession curves measured for sample A for a

channel length of L¼ 1.5 lm at T¼ 300 K. Inset shows a fitting of the chan-

nel length dependent four-terminal non-local data.

182407-3 Kum et al. Appl. Phys. Lett. 100, 182407 (2012)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

109.148.186.166 On: Fri, 28 Mar 2014 18:39:55

where P is the polarization, q is the resistivity of the nano-

wire, A is the cross sectional area of the nanowire, and L is

the distance between the injector/detector. Due to the good

uniformity of the nanowire and the electrodes, a reasonable

spin diffusion length can be estimated even with two data

points. A good fit is made with values P� 0.7% and

ksf� 220 nm, as shown in the inset of Fig. 5. This value is

in good agreement with the spin diffusion length derived

from the analysis of the Hanle data, within experimental

and fitting error. Finally, two-terminal spin valves in the

local geometry were also fabricated and measured. A MR

value of 10.5% was measured at room temperature with the

values of jHj for peak MR coinciding with those for peak

spin accumulation. However, due to inconclusiveness of

spin precession in the local geometry, the data are not pre-

sented here.

In conclusion, we have investigated spin injection, trans-

port, and detection in defect-free single GaN nanowire spin

valves with FeCo/MgO tunnel contacts. Measurements have

been made with spin valves having different nanowire chan-

nel lengths. Spin injection is confirmed by non-local MR and

Hanle measurements. Analysis of the temperature-dependent

MR data confirms diffusive spin transport in the nanowires.

The spin diffusion length and spin lifetime in unintentionally

doped GaN nanowires are 260 nm and 100 ps, respectively,

at room temperature.

The work is supported by the National Science Founda-

tion (MRSEC program) under Grant 0968346.

1U. K. Mishra, P. Parikh, and Y.-F. Wu, Proc. IEEE 90, 1022–1031 (2002).2E. F. Schubert and J. K. Kim, Science 308, 1274–1278 (2005).3R. Butte, G. C. E. Feltin, J.-F. Carlin, M. Mosca, M. Ilegems N. G. Rand-

jean Phys. Rev. B 73, 033315 (2006).4S. Christopoulos, G. B. H. von Hogersthal, A. J. D. Grundy, P. G. Lagou-

dakis, A. V. Kavokin, J. J. Baumberg, G. Christmann, R. Butte, E. Feltin,

J.-F. Carlin, and N. Grandjean, Phys. Rev. Lett. 98, 126405 (2007).5J. H. Bub, J. Rudolph, F. Natali, F. S. D. Hagele, Appl. Phys. Lett. 95,

192107 (2009).6A. Banerjee, F. Dogan, J. Heo, A. Manchon, W. Guo, and P. Bhattacharya,

Nano Lett. 11, 5396 (2011).7S. Krishnamurthy, M. V. Schilfgaarde, and N. Newman, Appl. Phys. Lett.

83, 1761 (2003).

8B. Beschoten, E. Johnston-Halperin, D. K. Young, M. Poggio, J. E. Gri-

maldi, S. Keller, S. P. DenBaars, U. K. Mishra, E. L. Hu, and D. D.

Awschalom, Phys. Rev. B 63, 121202(R) (2001).9J. H. Bub, J. Rudolph, F. Natali, F. Semond, and D. Hagele, Phys. Rev. B

81, 155216 (2010).10A. K. Bertness, A. Roshko, A. N. Sanford, J. M. Barker, and A. V. Davy-

dov, J. Cryst. Growth 287, 522 (2006).11L. Cerutti, J. Ristic, S. Fernandez-Garrido, E. Calleja, A. Trampert, K. H.

Ploog, S. Lazic, and J. M. Calleja, Appl. Phys. Lett. 83, 213114 (2006).12K. Kishino, A. Kikuchi, H. Sekiguchi, and S. Ishizawa, Proc. SPIE 6473,

64730T (2007).13W. Guo, M. Zhang, A. Banerjee, and P. Bhattacharya, Nano Lett. 10, 3355

(2010).14H. P. T. Nguyen, S. Zhang, K. Cui, X. Han, S. Fathololoumi, M. Couillard,

G. A. Botton, and Z. Mi, Nano Lett. 11, 1919 (2011).15J. Heo, W. Guo, and P. Bhattacharya, Appl. Phys. Lett. 98, 021110 (2011).16A. Das, J. Heo, M. Jankowski, W. Guo, L. Zhang, H. Deng, and P. Bhatta-

charya, Phys. Rev. Lett. 107, 066405 (2011).17T.-C. Lu, J.-R. Chen, S.-C. Lin, S.-W. Huang, S.-C. Wang, and Y. Yama-

moto, Nano Lett. 11, 2791 (2011).18W. Guo, A. Banerjee, M. Zhang, and P. Bhattacharya, Appl. Phys. Lett.

98, 183116 (2011).19M. Johnson and R. H. Silsbee, Phys. Rev. B 35, 4959 (1987).20E. I. Rashba, Phys. Rev. B 62, 16267I (2000).21G. Schmidt, D. Ferrand, L. W. Molenkamp, A. T. Filip, and B. J. J. van

Wees, Phys. Rev. B 62, R4790 (2000).22X. Lou, C. Adelmann, S. A. Crooker, E. S. Garlid, J. Zhang, K. S. M.

Reddy, S. D. Flexner, C. J. Palmstrom, and P. A. Crowell, Nat. Phys. 3,

197–202 (2007).23H. Kum, D. Basu, P. Bhattacharya, and W. Guo, Appl. Phys. Lett. 95,

212503 (2009).24S. P. Dash, S. Sharma, R. S. Patel, M. P. de Jong, and R. Jansen, Nature

(London) 462, 491–494 (2009).25D. Saha, D. Basu, and P. Bhattacharya, Phys. Rev. B 82, 205309 (2010).26T. Suzuki, T. Sasaki, T. Oikawa, M. Shiraishi, Y. Suzuki, and K. Noguchi,

Appl. Phys. Exp. 4, 023003 (2011).27C. H. Li, O. M. J. van’t Erve, and B. T. Jonker, Nat. Comm. 2, 245 (2011).28J. Tarun, S. Huang, Y. Fukuma, H. Idzuchi, Y. Otani, N. Fukata, K. Ishiba-

shi, and S. Oda, J. Appl. Phys. 109, 07C508 (2011).29E. Liu, J. Nah, K. M. Varahramyan, and E. Tutuc, Nano Lett. 10,

3297–3301 (2010).30S. S. P. Parkin, C. Kaiser, A. Panchula, P. M. Rice, B. Hughes, M. Samant,

and S. Yang, Nature Mater. 3, 862 (2004).31X. Jiang, R. Wang, R. M. Shelby, R. M. Macfarlane, S. R. Bank, J. S. Har-

ris, and S. S. P. Parkin, Phys. Rev. Lett. 94, 056601 (2005).32W. Guo, A. Banerjee, P. Bhattacharya, and B. S. Ooi, Appl. Phys. Lett. 98,

193102 (2011).33B. J. Jonsson-Akerman, R. Escudero, C. Leighton, S. Kim, I. K. Schuller,

and D. A. Rabson, Appl. Phys. Lett. 77, 12 (2000).34F. J. Jedema, M. V. Costache, H. B. Heersche, J. J. A. Baselmans, and B.

J. van Wees, Appl. Phys. Lett. 81, 5162 (2002).

182407-4 Kum et al. Appl. Phys. Lett. 100, 182407 (2012)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

109.148.186.166 On: Fri, 28 Mar 2014 18:39:55