phys. stat. sol. (b) 243, No. 7, 1643–1646 (2006) / DOI 10.1002/pssb.200565271

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Original

Paper

Fast spin relaxation in InGaN/GaN multiple quantum wells

J. Brown1, J.-P. R. Wells*, 1, S. A. Hashemizadeh1, P. J. Parbrook2, T. Wang2, A. M. Fox1,

D. J. Mowbray1, and M. S. Skolnick1

1 Low Dimensional Structures and Devices Group, Department of Physics and Astronomy,

University of Sheffield, Hicks Building, Hounsfield Road, Sheffield, S3 7RH, United Kingdom 2 EPSRC National Centre for III–V Technologies, University of Sheffield, Sheffield, S1 3JD,

United Kingdom

Received 31 July 2005, revised 25 January 2006, accepted 31 January 2006

Published online 9 June 2006

PACS 42.50.Md, 71.70.Gm, 73.21.Fg, 78.47.+p, 78.55.Cr

We report measurements of a fast spin relaxation in InGaN multi-quantum wells (MQW) using femtosec-

ond circularly polarised pump–probe spectroscopy. These reveal quantum beats in the spin polarisation

arising from a 1.6 meV fine structure splitting of the excitonic levels. We attribute this splitting to ex-

change interactions in the narrow quantum wells studied. The 5 K spin coherence time was measured to

be 450 fs, a value which decreases by a factor of approximately two by 50 K.

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction

There is huge current interest in using electron spin rather than charge to store information, with the

potential to reinvent parts of the electronics industry. The key to successful implementation of these

ideas is the development of spin injectors as well as the transfer and detection of electron and carrier spin

at room temperature. Amongst the potential materials to be used as spin injectors are materials such as

GaMnN [1, 2] for which it is apparently possible to have a Curie temperature as high as 370 K. The

obvious compatibility of GaN based non-magnetic semiconductors with these materials and the possibil-

ity of long spin lifetimes due to the weak spin–orbit interaction makes studies of the spin lifetime in

wide gap nitride nanostructures a subject of clear interest.

Reliable measurements of carrier spin relaxation became possible with the advent of ultrafast tech-

niques, with early reports concentrating on GaAs based systems [3]. The earliest report investigating a

GaN based system, studied spin relaxation in InGaN multiple quantum wells (MQWs) [4]. Spin relaxa-

tion times close to 100 ps were inferred from time resolved photoluminescence (TRPL). The first femto-

second pump–probe measurements performed to measure spin relaxation in nitrides, studied InGaN

epilayers [5] and concluded that it was not possible to generate an observable spin polarisation due to the

5–10 meV splitting of the light and heavy hole valence band states and the near 1:1 ratio of the transi-

tion moments for transitions involving those states. Despite this, the same authors demonstrated fast spin

decay in high quality GaN epilayers using a circularly polarised pump–probe technique [6]. The spin

decay time at a temperature of 150 K was measured to be 470 fs, at least an order of magnitude faster

than other III–V semiconductors. It was determined that the spin relaxation was governed by Elliot–

Yafet (momentum scattering) processes. Most recently, spin lifetimes as long as 220 ps have been meas-

ured for InGaN/GaN MQWs, with a strong dependence on the indium concentration demonstrated;

longer decay times observed for higher molar fractions up to ~0.11 [7].

* Corresponding author: e-mail: j.p.wells@sheffield.ac.uk, Phone: +44 (0) 114 222 4348, Fax: +44 (0) 114 222 3555

1644 J. Brown et al.: Fast spin relaxation in InGaN/GaN multiple quantum wells

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com

In this work, we report on femtosecond spin relaxation observed in an InGaN/GaN MQW using a

generic, circularly polarised pump–probe technique. We observe the presence of spin quantum beats at

cryogenic temperatures which wash out above 50 K. From these measurements we infer the presence of

fine structure splittings of the electronic eigenstates and the associated temperature dependent spin co-

herence times.

2 Experimental details

The sample consists of a 1 nm thick, 10 period In0.1Ga0.9N/GaN MQW having a 7.5 nm thick GaN bar-

rier. All investigated samples were grown on (0001) sapphire, double polished substrates by metalor-

ganic chemical vapor deposition (MOCVD) system. The substrate was treated in H2 at 1150 °C, followed

by a 25 nm GaN low temperature nucleation layer prior to preparation of the buffer and quantum well

layers. The PL spectra were recorded with a He–Cd laser excitation source, the PL dispersed in a Spex

1403 double monochromator and detected with an RCA C31034 photomultiplier tube. PLE spectra were

gathered using a Xenon lamp with wavelength selectivity/tunability provided by a Spex minimate mono-

chromator. Pump–probe spectroscopy was performed using a Spectra-Physics ‘Tsunami’ Ti-Sapphire

oscillator yielding 150 fs pulses, tunable between 700 nm and 1 µm at a repetition rate of 80 MHz. To

excite the interband transitions of InGaN MQWs, the oscillator output was directed to a Spectra-Physics

model 3290 frequency doubler providing the required excitation wavelengths between 350 and 450 nm.

A 1 µm resolution translation stage was used to provide the optical delay for the probe beam line with

circular polarisation of the pump and probe achieved using Alphalas tunable zero order, quarter wave

plates. The experiments were peformed in a transmission geometry with 1–2 kHz modulation of the

pump beam providing the reference frequency for lock-in detection of the probe beam signal, monitored

by a photomultiplier tube. The spot size of the probe beam was comparable but slightly smaller than that

of the pump (~50 µm). For all experiments reported here the pump beam provides an excitation density

of 5.6 × 1012 cm–2 per pulse at the sample surface.

3 Results

Figure 1 shows the 5 K photoluminescence (PL) and photoluminescence excitation (PLE) spectra of the

1 nm MQW sample. A single PL peak is observed at 3.25 eV with clear evidence of the lowest e1-hh1

excitonic transition apparent in the PLE spectrum. The wavelength at which the pump–probe measure-

ments were performed is resonant with the e1–hh1 transition.

In Fig. 2(a) we present the same and opposite circularly polarised (SCP – I+, OCP – I

-

) traces for

excitation at 375.5 nm with a sample temperature of 5 K. A striking feature of this data is the presence of

a relatively weak oscillatory signal, which is superimposed on a background trace due to incoherent

carrier recombination. Both the SCP and the OCP traces show these oscillations, having a phase shift of

π radians relative to one another. Figure 2(b) plots the degree of (or net) circular polarisation defined in

3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7

e1-hh1PLE

PL

Intens

ity(arb.u

nits)

Photon energy (eV)

Fig. 1 5 K PL and PLE spectra for the 1 nm In0.1

Ga0.9

N/GaN MQW.

The lowest energy transition within the quantum well is observable.

phys. stat. sol. (b) 243, No. 7 (2006) 1645

www.pss-b.com © 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Original

Paper

0 1 2 3 4

(a)

Transm

ission

Chan

ge(arb.un

its)

Probe Pulse Delay (ps)

Ι+

Ι−

0.0 0.5 1.0 1.5 2.0 2.5

0

10

20

30

40

(b)

Π(%

)

Probe Pulse Delay (ps)

Fig. 2 a) SCP (I+) and OCP (I

-) traces measured at 5 K. b) the net spin polarised signal (Π).

the usual fashion as ( )/( )I I I IΠ+ - + -

= - + . Such behaviour has been previously observed in a variety

of quantum dots systems from In(Ga)As [8, 9] to InP [10]. In the case of InP dots, the quantum beats

understandably occur in the presence of a magnetic field. However in the In(Ga)As system, quantum

beats are present for neutral dots with no external fields. These have been identified as arising from fine

structure splittings of the ground state exciton [8, 9]. Following these works, we suggest that the spin

quantum beat behaviour we observe is likely to arise from a fine structure splitting of the exciton states.

The simultaneous coherent excitation of these modes results in a propagating third order polarisation

which oscillates in time with a period equal to the magnitude of the fine structure splitting, and which we

sample as a function of the probe beam delay. To analytically account for this effect we use the frame-

work provided by Ref. [11] which describes quantum beat behaviour in an idealised three level system

having a split excited state. From this analytical model we obtain the expression:

( )2 2

coh/( ) (0) e e cos ,

t T tt t

δΠ Π ω

- -

= (1)

where Π(0) is the net spin polarisation at t = 0, Tcoh is the spin coherence time, δ is the width of a Gaus-

sian distribution of fine structure splittings and FS/Eω �= , with EFS being the magnitude of the fine struc-

ture splitting. At 5 K, we measure a spin coherence time (Tcoh) of 450 fs with an oscillation frequency of

2.45 THz, corresponding to a fine structure splitting of approximately 1.6 meV. The oscillations in our

signal are heavily damped due to the fast spin coherence decay as well as the presence of variations in

the magnitude of the splitting due to inhomogeneity. In fact the damping of the quantum beats is domi-

nated by the spin coherence decay. This is clearly evidenced by the temperature dependent decoherence

rate (defined as 1/Tcoh) which is shown in Fig. 3. Although an exact trend is not obvious from the data,

the spin decoherence rate is observed to increase immediately as the temperature is raised above 5 K.

Temperature (K)

0 30 60 90

Dec

oheren

ceRate(THz)

2

3

4

5

Fig. 3 Temperature dependence of the spin decoherence rate (1/Tcoh

).

1646 J. Brown et al.: Fast spin relaxation in InGaN/GaN multiple quantum wells

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com

This happens because after the system is prepared in a defined spin state by the exciting pump pulse, a

fast loss of amplitude occurs due to fluctuations in the precessional frequency via phonon and impurity

scattering. This is unlike the case for InGaAs quantum dots [9] where the data shows no change for tem-

peratures below 50 K because of the dominance of inhomogeneous effects. The apparent plateau in our

data above 60 K is most likely due to the fundamental time resolution of the experiment.

It is possible that the magnitude of the exciton fine structure splitting of 1.6 meV can be explained by

the large electron–hole exchange interaction. Exchange splittings are known to be sensitive function of

both quantum well width and barrier height due to their dependence on the electron–hole overlap [12].

In particular, narrow wells are therefore expected to have increased splittings, which can be larger than in

the bulk. Splittings as large as 6 meV have been reported for GaAs/Ga1–xAlxAs quantum wells [13]. This

also provides a most likely explanation for our inability to clearly resolve spin quantum beats in 2 nm

thick InGaN MQWs of the same composition, where any increase of the beat period (due to a reduction

of the splitting in wider wells) would then exceed the spin coherence time and be difficult to observe.

Ultimately, this is a consequence of the observation above that the beats are damped by the decay of

coherence and not by inhomogeneous effects. An additional possibility for the quantum beat behaviour

has been suggested to arise from the heavy hole – light hole splitting however this is likely to be close to

10 meV in our samples and as such cannot account for the observed effects.

4 Conclusions

Circularly polarised pump–probe measurements reveal spin quantum beats due to fine structure split-

tings of the excitonic states. From the period of the oscillation, we infer a fine structure splitting of

1.6 meV. A fast spin coherence of 450 fs is measured which decreases rapidly as the sample temperature

is increased, reaching the limiting time resolution of the measurements by approximately 60 K. Fine

structure splittings due to exchange interactions are expected to be enhanced for narrow quantum wells

and this is consistent with our observations.

Acknowledgements This work has been supported by the Engineering and Physical Sciences Research Council

(EPSRC) of the United Kingdom by grant GR/S24251/01. JB is supported by an EPSRC funded DTA award.

References

[1] T. Dietl, H. Ohno, F. Matsukura, J. Cilbert, and D. Ferrand, Science 287, 1019 (2000).

[2] T. Dietl, phys. stat. sol. (b) 240(2), 433 (2003).

[3] A. Tackeuchi, S. Muto, T. Inata, and T. Fujii, Appl. Phys. Lett. 56, 2213 (1990).

[4] M. Julier, A. Vinattieri, M. Colocci, P. Lefebvre, B. Gil, D. Scalbert, C. A. Tran, R. F. Karlicek Jr., and J.-P.

Lascaray, phys. stat. sol. (b) 216, 341 (1999).

[5] A. Tackeuchi, T. Kuroda, A. Shikanai, T. Sota, A. Kuramata, and K. Domen, Physica E 7, 1011 (2000).

[6] T. Kuroda, Y. Yabushita, T. Kosuge, A. Tackeuchi, K. Taniguchi, T. Chinone, and N. Horio, Appl. Phys. Lett.

85(15), 3116 (2004).

[7] S. Nagahara, M. Arita, and Y. Arakawa, Appl. Phys. Lett. 86, 242103 (2005).

[8] A. S. Lenihan, M. V. Gurudev Dutt, D. G. Steel, S. Ghosh, and P. K. Bhattacharya, Phys. Rev. Lett. 88(22),

223601 (2002).

[9] A. I. Tartakovskii, J. Cahill, M. N. Makhonin, D. M. Whittaker, J.-P. R. Wells, A. M. Fox, D. J. Mowbray,

M. S. Skolnick, K. M. Groom, M. J. Steer, and M. Hopkinson, Phys. Rev. Lett. 93(5), 057401 (2004).

[10] I. A. Yugova, I. Ya. Gerlovin, V. G. Davydov, I. V. Ignatiev, I. E. Kozin, H. W. Ren, M. Sugisaki, S. Sugou,

and Y. Masumoto, Phys. Rev. B 66, 235312 (2002).

[11] M. Mitsunaga and C. L. Tang, Phys. Rev. A 35(4), 1720 (1987).

[12] E. Blackwood, M. J. Snelling, R. T. Harley, S. R. Andrews, and C. T. B. Foxon, Phys. Rev. B 50(19), 14246

(1994).

[13] Y. Chen, B. Gil, P. Lefebvre, and H. Mathieu, Phys. Rev. B 37(11), 6429 (1988).