intervalley up-transfer for electrons in type-11 gaas ... · index term-charge carrier processes,...

IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 41, NO. 3, MARCH 2005 337

Evidence of Strong Phonon-Assisted Resonant . ''

Intervalley Up-Transfer for Electrons in ..'

Type-11 GaAs-A1 As Superlattices Xiaodong Mu, Yujie J. Ding, Zhiming Wang, and Gregory J. Salamo

Abstract-We demonstrate that interface optical phonons can efficiently pump electrons from the quasi-X states to the quasi-r states in short-period type-II GakeALb superlattices. As a result, peculiar behaviors on these superlattices have been observed. First, photoluminescence intensity for the quasi&ct transition drastically increases as the tempera- or pump power increases. Second, the dependence of the integrated photoluminescence intensity on the pump power exhibits a square power law.

Index Term-Charge carrier processes, phonons, scattering, semiconductor superlattices.

S EMICONDUCTOR superlattices (SLs) usually have high luminescence efficiencies [I]. [2]. However, for short-pe-

riod GaAs-AlAs SLs with the thickness of the alternating layers less than 12 monolayers (MLs), the X-states of AlAs mix with the !?-states of GaAs. As a result, the lowest electronic state in the conduction band becomes a quasi-X state (type-I1 SLs). In this case, the type-II SLs have effective indirect gaps [3]-[lo]. For example, nonequilibrium populations of both confined and interface phonons generated by picosecond laser pulses in ultrathin GaAs-AMs multiple quantum wells (MQWs) were studied in detail [l 11. In most of the previous experiments on type-II GaAs-AlAs SLs, a typical photoluminescence (PL) spectrum is dominated by a strong quasi-indirect transition peak with an extremely weak quasi-direct emission peak appearing as a shoulder in the spectrum [I], [3]-[lo]. It was demonstrated that the PL intensity for the quasi-direct peak could be slightly increased (less than a factor of 2) by raising the temperature of the type-I1 (GaAs)lo4AlAs)l~ SLs to 180 K [6]. Such a slight increase in the PL intensity was attributed to the fact that the electrons started to thermally populate the quasi-I? states from the quasi-X states as the temperature increased [6]. Obviously, such a mechanism dominates only when the lowest quasi-r state is close to the lowest quasi-X state by the thermal energy ~ B T .

In this paper, we demonstrate that the quasi-direct emission intensity in short-period type-I1 GaAs-AlAs SLs can be dramatically enhanced by raising the SL temperature or pump

Manuscript received July 8.2004; revised October 11.2004. This work was supported by the Air Force Office of Scientific Research and Army Research Laboratory.

X. Mu and Y. J. Ding are with the Department of Electrical and Com- puter Engineering, Lehigh University, Bethlehem, PA 18015 USA (e-mail: yud263lehigh.edu). 2. Wang and G. J. Salamo arc with thc DcpPNncnt of Physics, University of

Arkansas. Fayetteville, AR 72701 USA. Dgital Object Identifier 10.1109/JQE.2004.841613

power even if the energy difference between the lowest quasi-r state and quasi-X state is much larger than the thermal energr. Furthermore, we have found that the enhanced quasi-direct PL intensity exhibits a quadratic dependence on the pump intensity for the type-II SLs. We argue that our experimsntal results cannot simply be explained by the mechanism' of thermal transfer proposed in [6]. Instead, we propobi; 1. . . - vs r~echai.iisnl of phonon-assisted resonant up-transfer of electrc~n: :~uni -'?:

quasi-X state to the quasi-!? state. Based on this mechanism, we have derived the quasi-direct emission rate and comiar!d it with our experimental results.

vpe-II structures could be used to reduce threshold9 foi lasers [12] or to achieve passive Q-switching [l.i]. It is also conceivable to use phonon-assisted up-transfer for co?ling semiconductor devices by efficiently removing interface .optical phonons. So far, optical cooling has been ohserved in GaAs4al-,Al,As QWs with a pump beam wit11 the photon energy resonating with the energy of heavy-hole exciton"and emission at the energy of light-hole exciton [14]. Ln prinijple, the conduction band electrons have much longer decay time in type-II structures than in type-I structures. Therefore, the . con- .

duction band electrons in type-JI structures will have a .higher probability to absorb phonons and then to up-transfer to higher energy levels, e.g., from the X-subband to the r-su6lqnd. Moreover, one can design a novel structure to us,: the density of interface optical phonons to control the pipt..,ai.".~.! 2 f the electrons in the !?- and X-subbands and, therefore, tu achiev. d

Q-switched high-power output [15]. This paper is organized as follows. The sample slructures,and

the experiments carried out by us are described in Section II': The experimental results are reported in Section III. 4 ph~non-assisted up-transfer model for conduction band electrons i s ' bn - structed in Section 111 to explain our experimental results;..Then, in Section IV, we present a simplified theory and discuss opr results. Finally, in Section V, we draw our conclusions.

U. DESCRIP~IONS OF SAMPLES AND EXPERLMENTS '

Each of the five samples of the short-period ~ a A s - ~ i i - \ < SLs, studied here (see Table I) consists of 40 periods of G ~ A S and AlAs layers grown by molecular-beam epitaxy at 580 on a (100) semi-insulating GaAs substrate typically 400 /J,m thick. The growth rate is 1 MUs for both GaAs and AlAs layers. The layer thicknesses for Samples 1, 3, and 4 c-' ';lr.~lcs 2 znc! 5 were measured by X-ray diffraction and refle~tiori":li~!t-.:l.(.;~~~ electron diffraction, respectively. I :

.I' :

.00 0 2005 IEEE

Semrconduc~ors, V d 3% No. I , 2005, pp 127 131. Translaredfmm Fr:rka r Trkhnrka Polupmvodnrkov, Vol. 39, No. 1. 200j . pp. I40 -144. Orrgrnal Hus.\ron TCXI Copyright C 200j by Valakh Slwlchuk Kulomys. Mazur: Wan& ,Yrao. Salamo.

AMORPHOUS, VITREOUS, AND POROUS SEMICONDUCTORS

R

Resonant Raman Scattering and Atomic Force Microscopv .

of InGaAsJGaAs Multilayer Nanostructures with Quantum Dots M. Ya. Valakh*", V. V. Strelchuk*, A. F. Kolornys*, Yu. I. Mazur**,

Z. M. Wang**, M. Xiao**, and G. J. Salarno** *Institute ofSemiconductor Physics, National Academy qf Sciences qf Ukraine, Kiev, 03028 Ukraine

"e-mail: valakh@isp. kiev. ua **University of Arkansas, Department of Physics, 72701 Arkansas, USA

Submitted June 1,2004; accepted for publication June 16, 2004

Abstract-The transition from twdirr,ensima\ (ID) p~uhmorg,h:~c g~,ro.#tLh to the +hreeairn,ensio~~~\ (3D: (nanoisland) growth in In,Gal -,As/GaAs multilayer structures grown by molecular-beam epitaxy was investigated by atomic force microscopy, photoluminescence, and Raman scattering. The nominal In content x i n InxGal -,As was varied from 0.20 to 0.50. The thicknesses of the deposited In,Ga, -,As and GaAs layers were I4 and 70 monolayers, respectively. It is shown that, at these thicknesses, the 2D-3D transition occurs at x 2 0.27. It is ascertained that the formation of quantum dots (nanoislands) does not follow the classical Stranski-Krastanoc mechanism but is significantly modified by the processes of vertical segregation of In atoms and interdiffusion of Ga atoms. As a result, the In,Ga, -,As layer can be modeled by a 2D layer with a low In content (x < 0.20), which undergoes a transition into a thin layer containing nanoislands enriched with In (x > 0.60). For multilayer In,Ga, _,rAs

structures, lateral alignment of quantum dots into chains oriented along the [ i 101 direction can be implemented and the homogeneity of the sizes of quantum dots can be improved. O ZOOS Pleiades Publishing, Inc.

1. INTRODUCTION

An important line of development of fundamental and applied solid-state physics is investigating the processes of self-assembled formation of semiconductor quantum dots (QD) upon molecular-beam growth of strained heterostructures. It is believed that this process follows the Stranski-Krastanov mechanism; i.e., when the thickness of a deposited layer attains some critical thickness, elastic strain relaxation occurs with the formation of three-dimensional (3D) nanoislands (QDs) on a thin (several monolayers) two-dimensional (2D) wetting layer.

Most attention has been paid to InAs QDs, which are formed as a result of the 2D-3D transition under epitaxial growth of strained InAsIGaAs heterostructures. Some recent results indicate that the initiation and the growth of InAs QDs cannot be described in terms of the classical Stranski-Krastanov mechanism. It was indicated that these processes may be affected by the vertical segregation of In atoms and interdiffision of Ga atoms [I]. For In,Ga, -,As QDs formed in a GaAs matrix, the situation is even more complicated due to the simultaneous deposition of cations of two types. In addition, despite the fact that arrays with high densities (-10" ~ m - ~ ) of InxGal -,As QDs have been obtained, the spread of sizes and shapes of QDs hinder their wide application in practice. The use of multilayer structures makes it possible to solve in general the problem of vertical alignment of QDs along the growth

direction and improve the homogeneity of Llir,i 5,; F [2, 31, but lateral alignment (in the interface plar,e) is still a problem [4-6]. In the case of multilayer rjanois- land structures In,Ga, -,As/GaAs, the alignment critically depends on the surface elastic anisotropy of the matrix material [7] and the crystallographic orientation of the surface [8]. Previously, we showed for a multi- . layer system that, with the use of growth interrilption upon deposition of a separating GaAs layer, lateral alignment of QDs into a line can be implemented with an increasing number of layers. The QD parametel s and the features of their spatial alignment are cont~olled by the epitaxial growth conditions.

In this study, we investigated the formation and optical properties of QDs in In,Ga, -,As multilaye~ structures by methods of atomic force microscopy. resonant " Raman scattering, and photoluminescence. It is shown that laterally ordered QD arrays can be formed in such ' structures upon deposition of Ino 'Ga, 5 A ~ solio S O I U - tion onto a (1 00) plane.

2. EXPERIMENTAL

In,Ga, -,As/GaAs multilayer structures with quantum wells (QWs) and QDs were grown on semi-insulating GaAs(100) substrates by molecular-beam epitaxy. After removal of the oxide layer from the surface, a buffer GaAs layer 0.5 pm thick was grown at a rate of one monolayer (ML) per second. All samples were grown at a constant As vapor pressure (1 0-' Torr).

O 2005 Pleiades Publishing, Inc.

QELS (IEEE Cat. No. 05CH37696), p 1609-11 vol. 3,2005

JThE16 2005 Quantum Electronics and Laser Science Conference (QELS)

Observation of a ID and 2D dark notch formed in a quadratic photorefractive medium

Yongan Tang, Aqiang Guo, and Gregory Salamo Physics Department, Univenify ofArknnras, Fayetteville, AR 08544

[email protected]

Jason W. Fleischer Electrical Engineering Deparbnent. Princeton Universify, Princeton, NJ 08544

Abstract: We report on the observation of a 1D and 2D dark notch in an unpoled SBN:75 crystal using the quadratic electro-optic effect. Both the 1D and 2D dark notches are observed to be fixed in the crystal. Q 2005 Optical Society of America OCIS codes: 190.4400 and 050.1940

Optical dark notch and vortex solitons form when self-defocusing balances diffraction. Vortex solitons were predicted [I] and observed [2] in Kerr-type media and in saturable self-defocusing nonlinear media [3]. In these investigations, the nonlinearity that gives rise to the solitons was isotropic and therefore was expected to support dark soliton stripes, dark soliton crosses and grids [4], as well as circular vortex solitons [2, 31. Following the first prediction of photorefiactive spatial solitons and the corresponding observation of quasi-steady-state photorefractive spatial solitons [5], self-trapping of an optical dark notch and vortex was also demonstrated in these inherently anisotropic nonlinear media, such as, photorefractive SBN:60 crystals [6] and photovoltaic LiNb03 crystals [7]. Vortex solitons observed in photorefractive materials [6, 71 differ in their physical origin from those observed in the references [2] and [3]. These solitons are driven by the linear electrwptic effect and form despite

+ E-polarized input

/ Fig.1 Results fiom a Michelson Interferometer measurement using ordinary (diamond) and extraordinary (circle) polarized Argon laser beam. Interference fringe movement is measured as a function of positive and negative applied field. The solid lines are a quadratic fit to the data.

the anisotropy of the photorehctive nonlinearity. In addition to quasi-steady-state and photovoltaic solitons, another important type of photorefiactive soliton is the screening photorefiactive soliton [8]. Screening solitons are also generated in biased photorefractive media, but unlike quasi-steady-state solitons they exist in steady state when the applied field is nearly completely screened [9-111. For this reason, they are of most interest today. Both dark notch and vortex solitons have also been observed 112, 131 as screening solitons.

In this paper, we report the observation of steady state photorefractive 1D and 2D dark notches in a photorehctive crystal driven by the quadratic electro-optic effect (Fig.1). We study the self-trapping effects for both the 1D and 2D dark notches as potential dark notch and vortex screening soliton in an unbiused SBN:75 crystal.

polysilicon with TiN-only contacts before pulse programming. The oc. Symp. Proc. Vol. 864 O 2005 Materials Research Society E5.7

defects in the TiN-only contacted polysilicon causes a large density o giving low current through the diode prior to a programming pulse heats the polysilicon to the melting point, reordering the material after quenc Nanoscale Deformation in GaAs(100): Towards Patterned Growth of Quantum the quenched polysilicon diode has fewer crystalline defects with a smaller Dots band gap states, giving higher forward current after the programming pulse, contacted diode before programming.

REFERENCES

1.H.A. Schafft, Proc. IEEE 55, 1272 (1967). 2. M. Tanimoto, J. Murota, Y. Ohmori, and N. Ieda, IEEE Trans. Elect. De 3. M. Tanimoto, J. Murota, M. Wada, T. Watanabe, K. Miura, and N. Ieda, 17,62 (1982). 4. T. Mano, K. Takeya, T. Watanabe, N. Ieda, K. Kiuchi, E. Arai, T. Ogawa, an IEEE J. Sol. St. Circ. 15,865 (1980). 5. J.E. Mahan,Appl. Phys. Lett. 41,479 (1985). 6. P.G. LeComber, A.E. Owen, W.E. Spear, J. Hajto, A.J. Snell, W.K. Reynolds, J. Non-Cryst. Sol. 77 and 78, 1373 (1985). 7. V. Malhotra, J.E. Mahan, and D.L. Ellsworth. IEEE Trans. Elect. Dev. ED-3 8. D.W. Greve, IEEE J. Sol. St. Circ. 17, 349 (1982). 9. S.B. Herner, A. Bandyopadhyay, S.V. Dunton, V. Eckert, J. Gu, K.J. Hsia, S. Kidwell, M. Konevecki, M. Mahajani, K. Park, C.J. Petri, S.R. Radigan, U. Rag Vienna, M.A. Vyvoda, IEEE Elect. Dev. Lett. 25,27 1 (2004). 10. K. Holloway and R. Sinclair, J. Appl. Phys. 61, 1359 (1984). 11. M.K. Hatalis and D.W. Greve, J. Appl. Phys. 63,2260 (1988). 12. J.B. Lasky, J.S. Nakos, O.J. Cain, and P.J. Geiss, IEEE Trans Elect. Dev. 38 13. C.A. Sukow and R.J. Nemanich, J. Muter. Res. 9, 1214 (1994). 14. J.A. Kittl, Z.-Z. Hong, M. Rodder, D.A. Prinslow, and G.R. Misium, Proce Symposium on VLSI Technol. Dig., 1996, p. 14. 15.K.S.Kim,Y.C. Jang,K.J. Kim,N.-E.Lee,S.P.Youn,K.J.Roh,andY.H. Tech. B 19, 1164 (2001). 16. Q. Tiang, X. Liu, T.I. Karnins, G.S. Solomon, and J.S. Hams,App (2002). 17. G.D. Cody, T. Tiedje, B. Abeles, B. Brooks, a n d Y. Goldstein, Phys. R (1981). 18. G. Fortunato a n d P. Migliorato, Appl. Phys. Lett. 49,1025 (1986). Also, in order to integrate quantum dots with existing micro-devices, the

able to be positioned and patterned at precise locations. Therefore, there exists for the development of simple, inexpensive, and robust techniques for

ation of GaAs due to nanoindentation at extremely small dimensions (< 200 nm).

175

Available online at www.sciencedirect,com

ELSEVIER

J W R ~ O F CRYSTAL GROWTH

Journal of Crystal Growth 284 (2005) 47-56 www.elsevier.com/locate/jcrysgro

Microsize defects in InGaAs/GaAs (N 1 l)A/B multilayers quantum dot stacks

P.M. Lytvyna, I.V. Prokopenkoa, V.V. Strelchuka, Yu.1. ~ a z u r ~ . * , Zh.M. wangb, G. J. salamob

aLashkaryov Institute of Semiconductor Physics, NAS of Ukraiize. Prospect Nauky 45. 03028 Kyiv, Ukraine b~epar tment of Physics, University of Arkansas, Fuyetteville, Arkansas 72701, USA

Received 2 February 2005; received in revised form 4 July 2005; accepted 7 July 2005 Available online 15 August 2005 Communicated by R.M. Biefeld

Abstract

Surface morphology of microsize defects on the surface of various high-index GaAs substrates was investigated using an atomic force microscope (AFM). The surfaces investigated were the top layer of 1- and 17-period Ino,4sGaAso,ss/GaAs structures with quantum dots or buffer layer. These structures were characterized by the formation of oval defects on (1 0 0) surfaces, and microsize defects possessing the shape of multifaceted pits and hillocks on (N 1 l)A/B (N = 7,5,4,3) surfaces. The microsize defects were found to chaotically distribute on the surface and, as a rule, gathering in groups with some number of defects. Their density did not depend on the substrate orientation while the shape and orientation of the microsize defects were found to depend on the crystallographic orientation of the substrate. This dependence was determined to be the result of anisotropy of surface diffusion and surface elastic properties. The anisotropy of elastic properties of high-index surfaces was found to be the dominating factor in determining the microsize defect shape. We also report direct evidence of the fact that the effect of quantum dot lateral ordering observed on high-index (N 1 l)B surfaces is determined by the anisotropy of surface elastic properties as well as elastic interaction between adjacent quantum dots. 0 2005 Elsevier B.V. All rights reserved.

PACS: 07.79.Lh; 68.35.D~; 68.55.Jk; 68.65. +g; 81.05.Ea

Keywords: Al. Atomic force microscopy; Al . Nanostructures; Al. Oval defects; A3. Molecular beam epitaxy; B2. Semiconducting 111-V materials

1. Introduction

'Corresponding author. Tel.: + 1479 575 7476; fax: + 1479 575 4580. Molecular beam epitaxy (MBE) offers the

E-mail address: [email protected] (Yu.1. Mazur). interesting possibility to produce semiconductor

0022-02481% -see front matter 0 2005 Elsevier B.V. All rights reserved. doi: 10.1016/j.jcrysgro.2005.07.005

We have ~ltown that wvd wiM> sp.xed *a.n--ys of atep WI be p m d 1 d &rough suhlinwtion vn top uf f a b k M e d mesa stmctii~cs. A k g fa- of Lhe p m s g Is rhe irkhid spon~m~nts ,' fannation of a ndgc m)~r1d the -rap of each ~llrsa m O ~ L T to mdwx the Me chemical pmt~d ; gradiena n.mx~ated wid1 B e e&gc. T~IC p w w t &this rrdse dlt>ws khc eenw of tt# meqa m , home almost s q - h z . As tis rdgw widen hy difruskn, su'nli~wlinn cvenlually takes over und twpes the zliminafiou of *be ridge md ihe intrusivn ~i +ips from h e d g onto &c rncva 2 surfwe. Thc ~ L W ardy of steps has a r0sl3ce width itirt$mion &ttt pesdsra Rn long m 1 f : k i i t 2 t b q . &mpulu s i r n u t a ~ ~ of thc inilia1 low slop density wcs on dr mesa suggen Ba th i ~ & 1 y s p d wep m y s prodwed in the expannr?Dt tire Iormed due to 911bWo11 Ismi& kinetics. Computer simuldons dm suggcsc wt the maximum timace size pdtld by ~S "

mahod e m be wntro11ed by v ~ g thr annatfidc t c m p e m r ~ ad bq mrrodu~k a dcpsiia~t :. flux.

Y

R.-C. Cbaag 8nd 3. M. ~ i h l y . in Mat. Re$. Set. Sjtmp, Pror., f m ~ Val. 749. WJ6,$. 2. JAG& ahd %I+ J, WW, 3. Appl. Yh>%. 47+2433 {IWk. w. '@. hfdh,f' Appl. Phys. %, 333 f 3957). W. W. &tolRrns. 3. A@. k'hys. %I,?+! (1959). Ti. S. Ro&&, t,L. Rohrer. aud W. W. h&l?ins, 1, hm. C z m . Soc. 84,2099 (2001). 'Ar. %. MuIh sad G. S. Rahxer. J. Am. Ccrm. Sw. a, 214 (2000). N. Cornbc, P. lensen, and A. Pimpinclti, -5. Rev. Lott G. 1 10 (2000). W. K B I # ~ I ~ , 3. abrem aml F. C. Fmk, Philos. Trans. R. Sm. of fandon Ser. A 2113, 299 (I951'L. < ---,

E91 K-c. C h g and J. M. Bl&@ly, Surf. Sci., T h bc p u b l i ~ w C,m j). C. Mi&h ntPdO. P i r n - L ~ c ~ i ~ , RYS. &v. E53,Rd318 (1996)*

fill M. SamandM Cwaha.~~~ys. REV. 83 si,1117:! !199~),

Sh. &pdmdl*L~di, FE W a , B. M. W:mg, m d G. J. SuJ.mo &pmcnl& phy$ic~, UnivcmitJ' of Arlr;tnsas.

F-ymv&, kkimslts, 7?70i, U. S. k

wc b v d g a t e the famatim of [Tn,&)As self assemh1d quanun dmctur~s porn on &rent orientatiouv of a Ca4~ subs milt^ alotlg one side of the 3tmsgaphic ti+an& b & t ' e ~ ~ ( 1 0 ) wd ( 1 i i)A ~ v r f o m ~ . Tbz samph WSXI? grown by Mdclcular U m ISpir~xy, &to& by K @ i t i ~ n Hig7.1-Phet:gy kktron Diffra- &ing the g o \ ~ ~ h a ~ d c h i t ~ a ~ M x d by in-siw S c ~ n i n g %m,&jng ~ r o s c o ~ y aud Atomic Porn Micrwcnw. A a~!r;SenMiC m h S i f i # fmM wra dimmianal (In.GaT.kq 411antum & to QW dinm1siol1~i quantum wFfi ZV- obsenrcd as the $ubrmte W& vadod dong the side of tfie tdpnglc ruiulih 3' r n i ~ u t %in the (1001 tow&< (I 1 [}A wb& i-tia sc:~eraj &$i in& suhce5. W o pmpcw an eqliuradon for rbo role ol the c~tbmate in def:emi&ug f316 qye of t),19 ~ $ m ~ c t i u o that i3 fwid.

x-

Cmdiy, hlgh indu s e ~ q d m sahwdtas tmw the portr~lial tb 9ct s &pl@m h Ihe g m ~ & of q u m s t brrnctur~~. 'Shis p~mEialir; on the u*iymsurf~ wrrpbaIogy of ?u9;11 i W ~ur faC8s tkdt influen& & sirr, &apt: Unifm*, and lmaibu of &-its&lob oiolm\m j~.Zj, EB e x m e , dxperlwc4ts ;tlrr:ady dcmnr;trrtte tii~~t the odkbtnti011 of tke rrlbsnd~e w$.xe c m Pap mf role jn the iomatim of &iRerest tyfRs t)f (h,-%dh hetz:mqPitasialqantln mt~lupes gn,m on &I% [4.5,61-

l[n &is w~k, ae inV&tigm the m s f n d o n uf (b,#a)As nmustmctm d q O f i t adc of ;he sqw~gnphic @iadc bemen (lo) and (11 l)h, ski cqlain thc ;vie af tk whsA* based on *Jky of rhehigh index { 11 5 2 ) surface. Jnte~cthgly we obwelheYlttXkB&a df di&cnt quadalm ~ ~ h 7 m jvst by changing thc miwrimtn~on ac$lc (1@3%%?!2 23" ww3Kd the (1; I W surfw.

'$be s&fw mdidhere by S/iglxulaC &t-81~ E p i a ~ y (MRE), k-&~n% & i~ f&$T>#~ag hiicrmcopy (51M1 uul.4Em1c Po~ra mmseopy ( e l ) inclvdc the (\00j, Qt 7 5 1 3 kA,

,411 LA $gdl,311 Ik e. 1 is ttbdklaar;~ virm'od dong the [01-11 d l r e ~ t i a & m s ~ l i n g e * d of t k e 5 i ~ t - f ~ invcstig@. by tho w d m i s c u t frOul TIE (Ilm) temp-fatc!.F@ ?@.*me ccrwrye & &,13af&, s syscepa& mtiitipn &om zn, dimenimd (Ia~f.&qmum rtok (QDs) to M e dmwmh%ii qumtum w&a {QH.ih) was o b , w d a< the aw i w16 varied s b q one sidr: of the triimgkc ~5wean {SW) and (1 1 1 )A. Pmie (fU,G8l-%ab$ f m on CfaAb (1424)) f l l l ) ~ SU&W~$, ~ i f % ~ $ M d fh,G)rIs QDq fomd & Oak. {#I $)A surface ~d

QWWs .w Otw:L? @ I. i}A;wi (41 l>A sutfaqs. An expla~tioo fGr the Bifwnt coufid snucmres c ~ b ~ d fw stibfi~ratc?~ Bong me h-kk of the mgb is p-.

epitaxy. Two (In,Ga)As QD layers, separated by a thin GaAs layer form an AM. Ou calculations show that the energy level splitting in an AM is a function of the s and varies from -150 meV to a few meV. For QDs grown in our laborat cross-sectional TEM analysis shows that a spacer thickness

ubband Transitions in Ino,,Gao,As/GaAs Multiple Quantum Dots of Varying formation of an AM with a ground-state splitting energy of photon energy lies in the THz range. We investigate the interband transitions using Dot-Sizes photoluminescence spectroscopy. The experimentally measured spectra agree with theoretically calculated ones and prove the presence of symmetrically coupled QDs molecule with the appropriate ground state splitting. assmore"'. E.A. ~ e ~ u i r " ' . M.O. ~anasreh" . Zhiming

These studies provide valuable information that is directly useful in achie understanding of the physical effects involved in optical transitions in AMs, and development of new THz devices. ment of Electrical Engineering and Microelectronics-Photonics Program.

of Arkansas, Fayetteville, AR 72701, USA

REFERENCES ent of Physics, University of Arkansas, Fayetteville, AR 72701, USA

1. J. Faist, F. Capasso, D.L. Sivco, C. Sirtori, A.L. Hutchinson, and A.Y. Cho, Scien ( 1 994).

2. C. Sirtori. P. Kruck, S. Barbieri, P. Collot. J. Nagle, M. Beck, J. Faist, and U. 0 ectra of intersubband transitions in In,,,Ga,,,AslGaAs Phvs. Lett. 73, 3486 ( 1998). antum dots (MQDs) grown by molecular beam epitaxy were investigated. By

3. R. Paiella, F. Capasso, C. Grnachl, D.L. Sivco, J.N. Baillargeon. A.L. Hutchinson, number of In,3Ga,,,,As monolayers deposited, a series of samples with and H.C. Liu, Scie~rce 290. 1739 (2000). monolayers were obtained. The quantum dots

4. C. Gmachl. H.M. Ng, and A.Y. Cho. A w ~ . Phvs. Lett. 77, 334 (2000). 5. J. Ulrich, R. Zobl, W. Schrenk, G. Strasser. K. Unterrainer. and E. Gomik, App

77, 1928 (2000). pper and lower limit of In,,,Ga,,,,As quantum dots for infrared detectors. A 6. C. Sirtori, J. Faist, F. Capasso, D.L. Sivco, A.L. Hutchinson, and A.Y. Cho, IEE

pm is achieved for structures grown with the above Technol. Lett. 9, 294 (1997). ape of the intersubband transition absorption 7. J. Faist, F. Capasso, C. Sirtori, D.L. Sivco. J.N. Baillargeon. A.L. Hutchinson. an ements. From the lineshape, it was deduced Appl. Phys. Lett. 68,3680 (1996).

n is present in thick quantum dots and bound-to-bound 8. R.A. Suris. in "Future Trends in Microelectrot~ics. ReJectiotzs on the Road to n is present in thinly grown quantum dots. Nutzotechnology," edited by S. Luryi, J. Xu. A. Zaslavsky, Kluwer Acad. Pub

(1996). 9. N. Wingreen and C.A. Stafford. IEEE J. of Quurzt~im Electrot~ics 33, 1170 (1997) 10. C.-F. Hsu, J.-S. 0. P. Zory, and D. Botez, IEEE d. of Selected Topics in Quantum

6,49 1 (2000). 1 systems for infrared imaging is the 11-VI

11. W. Fang, J.Y. Xu, A. Yamilov, H. Cao, Y. Ma, S.T. Ho and G.S. Solomon, Opt. this material possesses a very high quantum 948 (2002). 12. I.N. Stranski. L. Krastanow, Sirz~i~rgsberichte d . Akad. d. Wissenschuften in Wien,

Band 146,797 (1937). 13. V.G. Stoleru, D. Pal. and E. Towe, Phys. E: Low-r1imen.s.

as and allow for reliable reproduction [2,31. 14. D. Pal, V.G. Stoleru, E. Towe, and D. Firsov, Jptr. J. App 15.0. Stier, M . Grundmann, and D. Bimberg, Plr-vs. Rev. B 59,5688 (1999). ny researchers to seek out an alternative solution 16. R. Heitz. A. Kalburge. Q. Xie, M. Grundmann, P. Chen, A. Hoffmann, A. Madhu the quantum well infrared photodetector (QWIP)

Bimberg, Phvs. Rev. B 57.9050 (1998). infrared photodetectors are designed from wide bandgap (III-V) 17. P. Boucaud. J.B. Williams. K.S. Gill, and M.S. Sherwin. Appl. Phys. Letr. 77,43 18. E. Batke, G. Weimann, and W. Schlapp, Phvs. Rev. B 39, 1 1 171 (1989).

ared sellsing to the far infrared. The material growth and based devices are more attainable as compared to these . the QWlP has several limitations such as the

15

Available online at www.sciencedirect.com ..

Journal of Crystal Growth 277 (2005) 72-77

mRwoFCRYSTAL GROWTH

RHEED study of GaAs(3 3 l)B surface

V.R. Yazdanpanah*, Zh.M. Wang, Sh. Seydmohamadi, G.J. Salamo Depurfmenf of Physics, Universify of Arkansas, Fu~vefteville, AR 72701, U S A

Received 24 October 2004; accepted 6 January 2005 Available online 25 February 2005

Communicated by R.M. Biefeld

Abstract

In this paper, the growth dynamics and the surface morphology of the GaAs(3 3 l)B surface have been studied by both in situ reflection high-energy election diffraction (RHEED) and in situ scanning tunneling microscopy (STM). For the first time, a RHEED oscillation is reported on high index GaAs(3 3 l)B faceted surface with (1 10) and (1 1 l)B facets. The RHEED oscillation was observed only along the [i 161 direction. Absence of any RHEED oscillations along [l 161, [i 101, and [I TO] indicates a possible growth model in which the GaAs(3 3 l)B surface is moving frontward through fractional growth of its (1 1 l)B facets. These results help us to better understand the nature of RHEED oscillation on high index GaAs. 0 2005 Elsevier B.V. All rights reserved.

PACS: 81.05.Ea; 81.15.Hi

Keywords: Al. Crystal morphology; Al. Crystal model; A l . Reflection high-energy electron diffraction; Al. Surface structure; A3. Molecular beam epitaxy

1. Introduction

Reflection high-energy electron diffraction (RHEED) is one of the most important in situ probes for molecular beam epitaxy (MBE). For example, RHEED oscillations have been intensively studied for the control of growth rates and

*Corresponding author. 1915 Maple Avenue $601-1, Evan- ston, IL 60201, USA. Tel.: + 14792836759; fax: + 1 479 575 4580.

E-mail address: [email protected] (V.R. Yazdanpanah).

growth conditions [I-31. The oscillation is generally observed under growth conditions that lead to layer-by-layer two-dimensional (2D) growth. Dur- ing the 2D growth, the surface morphology apparently varies periodically due to the nucleation and coalescence of 2D islands in the growing front. While it is well established through comparison with other methods that the period of oscillation corresponds to the growth of one monolayer (ML) of GaAs, the origin of the RHEED oscillations remains controversial. Two of the most popular explanations are based on the

0022-0248/$-see front matter 0 2005 Elsevier B.V. All rights reserved. doi: 10.1016/j.jcrysgro.2005.01.063

phys. stat. sol. (a) 202, No. 8, R85-R87 (2005) 1 DO1 10.10021pssa.200510031

Self-assembly of GaAs holed nanostructures by droplet epitaxy

Zhiming M. Wang*, Kyland Holmes, John L. Shultz, and Gregory J. Salamo

Department of Physics, University of Arkansas, Fayetteville, Arkansas 72701, USA

Received 5 April 2005, revised 21 April 2005, accepted 9 May 2005 Published online 11 May 2005

PACS 61.46.+w, 68.37.Ps, 68.55.A~ 81.15.Hi, 81.16.Dn

' Corresponding author: e-mail [email protected]

1 Introduction Semiconductors show dramatic quan- tization effects when carriers are confined by potential barriers to nano-scale regions of space in one, two, or three dimensions. For example, the optoelectronic properties of semiconductor nanostructures can be modified by structurally engineering the size and shape of nanostructures. Fab- rication of active semiconductor nanostructures, however, is extremely challenging due to technology limitations at these dimensions. Self-assembly of semiconductor nanostructures, however, provides a rich spectrum of opportunity. The most common approach to the self-assembly of nanostructures has been the Stranski-Krastanow (SK) method. In the SK approach to the formation of nanostructures, a thin semiconductor film is deposited onto a different semiconductor substrate that presents a high mismatch in lattice constant between the substrate and the thin film material being deposited [ I , 21. The lattice mismatch then self- assembles quantum dots or quantum wires having relatively simple geometries [l-61. Even more complicated configurations are possible, however, such as quantum rings and quantum dot chains [7- 101 which are currently drawing much attention. Compared to colloidal nanocrys- tals, which have been observed to be rod-, arrow-, tear- drop- and tetrapod-shaped [ l I], the shapes of the semiconductor nanostructures fabricated by epitaxy at first seem less attractive. Yet, epitaxial nanostructures with more interesting shapes will likely have novel optical and electrical properties based on the effects of shape.

0 2005 WILEY-VCH Verlag GmbH & Co. KGaA. Weinheim

While the SK growth approach has had dramatic success with lattice-mismatched systems a different approach is needed for lattice matched systems, such as GaAsI AIGaAs. Lattice matched material systems can have as much, if not more, technological importance. Although not studied with as much fever, droplet epitaxy offers the possibility of formation of quantum dots without the need for a lattice mismatch. In droplet epitaxy of GaAsIAlGaAs, Ga is first deposited to create liquid Ga droplets on an AlGaAs surface and subsequently exposured to a high arsenic flux that transforms the droplets into GaAs nano-crystals [12, 131. In this letter, we extend the capability of droplet epitaxy to the growth of more interesting GaAs nanostructure geometrical shapes never explored before. GaAs lighted nano-candles and square-holed round nano-coins are observed under different growth conditions.

2 Experimental All the samples investigated here were grown on epi-ready GaAs(100) substrates by molecular beam epitaxy (MBE) which is equipped with a reflection high-energy electron diffraction (RHEED) system and a highly accurate (f 2 "C) optical transmission thermometry system for substrate temperature determination. A valve- controlled As source was used, which enables instantaneous changes in the As flux depending on the valve position Following the growth of about 300 nm of a GaAs buffer layer at 600 O C , an A1,,Ga0,As layer of 200 monolayers (MLs) was deposited to be a barrier for the further growth

0 2005 WILEY-VCH Verlag GmbH & Co KGaA. We~nheim

APPLIED PHYSICS LETTERS 87, 21 3 105 (2005)

Photoluminescence linewidths from multiple layers of laterally self-ordered InGaAs quantum dots

Zh. M. an^,^) Y. I. Mazur, Sh. Seydmohamadi, and G. J. Salamo Department of Physrcs, Universiry qf Arkarlsas. Fayetreville. Arknr~sa.~ 72701

H. Kissel Ferdirrand-Braurr-I~tsritut fur Hocl~s~requenztecl?rlik, Gustai~-K~r.chhof-St~'ns~re 4, 12389 Berlin. Gerrrlarly

(Received 17 August 2005; accepted 20 September 2005; published online 14 November 2005)

Laterally ordered multilayered m a y s of InGaAs quantum dots are investigated by photolunlinescencc as a function of high index GaAs substrates. Different lascr wavcle~lgths are used to investigate the photoluminescence from quantum dots layer-by-layer. High optical quality is demonstrated for laterally ordered quantum dot arrays. GaAs(51 l)B is identified as the optiinum high index substrate for growth of InGaAsIGaAs multilayered quantum dots, demonstrating strong photolun~inescence with a narrow full width at half maximum linewidth of 23 meV in spite of the potential for misfit dislocations. O 2005 American Instilute nJ Physics. [DOI: 10.106311.213 I 1981

Devices based on semiconductor quantum dots (QDs) are under development to utilize the unique optical and electronic properties that results from three-dimensional (3D) quantum confinement. In many cases, ordered QD arrays are part of the design of such devices. As a result, there has been a growing intercst in lateral control of the QD position as evidenced by the report of a variety of clever organizalional techniques developeti most recently.'-' One interesting example among them is the observation of larernlly sey- orderirrg during the growth of InGaAsIGaAs QD r n ~ l t i l a ~ e r s . ~ ~ ~ Remarkably, the ordered pattern can be adjusted by selecting different high index substrates. For example, one-tlimensional (ID) ordering. (QD chains up to several microns in length), has been achieved on GaAs(100) (Ref. 4) while two-dimensional (2D) ordering (a QD check- erboard) was observed on GaAs high index s u r f a c c s . ~ l - though there have been many experimental and theoretical studies on the behaviors of QDs on high index GaAs surface^,^'^-" the choice of the optinluln indexed substrate for QD ordering, size unifomlity, and importantly, the corresponding optical properties of the QDs, still remains an in- tercsting question. Indeed, since the formation of InGaAs QDs on GaAs surfaces is strain driven, multiple QD layers may introduce misfit defects which deteriorate their optical properties. For this reason, a systematical investigation of the optical behavior oP ordered QD structures as a function of high index can be of value for optimum size uniformity and optical performance oP QD multilayers. In this letter, we demonstrate strong photoluminescence (PL) emission with a narrow linewidth from nlultiple QD layers having a high degree of lateral ordering. In particular, we systematically demonstrate GaAs(51 l)B to be the optimum growth substrate orientation for the growth of lateral ordered, high optical quality, InGaAs QD multilayers.

Samples for this study were grown by molecular-beam epitaxy (MBE) on GaAs(100) and GaAs(n.1 l)B (where n is 9, 7, 5. 4. and 3). Details of the MBE system and the growth coilditions have been reported e l sewl~ere .~~%~l oP the substrates with different indexes were indium soldered side by side on a molybdenum block. The block was rotated during growth to ensure a uniform distribution of materials. The

")~lectronic mail: [email protected]

grown layered structure is shown in Fig. 1. The thickness of the GaAs spacer between QD layers is 120 monolayers. nearly doublc that used in previous reports.' This thickness was intentionally selected to insure that vertical tunneling of carriers between IilGaAs QD layers is negligible1' and that we arc able to study the optical propcrtics of a QD n~ultilay- ered stack as optically independent layers.

Figure 2 shows atomic force microscopy (AFM) images of the InGaAs QDs in thc last exposed layer. As expected, "long chains" oP QDs and QD "checkerboards" are observed on G'aAs(100) and GaAs high index surfaces. The autocor- relation plot, which are insets in Figs. 2(a) and 2(d), givc quantitative measure to the nalure of the 1D and 2D lateral ordering. For example, the 2D ordering achieved on high index surfaces in th-is experiment is much better than noted in previous reports."

The PL measurement was performed at a temperature of 10 K using the 325 nnl line of an HeCd laser, 514.5 nin line of an Ar+ laser, 632 nm line of a HeNe laser. and 797 nm line of a Ti-sapphire laser. as the excitation wavelengths. By using different excitation wavelengths, the optical beam pen- etrated into the sample at different corresponding depths due

GaAs substrates indexed

FIG. 1 . Schematic of the grown layel. sequence. The difference between QDs in the first layer and In the last layer is illustrated.

0003-6951/2005/87 21)/213105/3/$22.50 87, 213105-1 O 2005 American Institute of Physlcs DOWlllOad@d W J a r 2008 10 130.114.231.6. Redistribution subject to AlP license or copyright; see http:i/apl.aip~rglapIIcopyright.j~p

Available online at www.sciencedirect.com

scleNcE@olnecT*


JouRwoFCRYSTAL GROWTH

Morphological instability of GaAs (7 1 1)A: A transition between (1 0 0) and (5 1 1) terraces

V.R. Yazdanpanah, Zh.M. ~ a n g * , G.J. Salamo Department of Physics, University of Arkansas, Fayetteville, AR 72701, USA

Received 12 February 2005; accepted 28 February 2005 Available online 18 April 2005

Communicated by R. Bhat

Abstract

We report on the use of reflection high-energy electron diffraction (RHEED) and scanning tunneling microscopy (STM) study that indicates that the GaAs (7 1 l)A is right at the transition between vicinal GaAs (1 0 0) and vicinal GaAs (5 1 l)A surfaces and that a variation of the As overpressure switches the surface morphology between the two vicinal surfaces. The steps on the vicinal(10 0) surface have a width of 1.5 nm creating a staircase surface with excellent possibilities for growth of quantum wells. As-rich conditions can be described by vicinal (5 1 l)A surfaces with a width of 3.5 nm. This surface could find applications as a template for quantum wire growth. The observation suggests that the transition between these two morphologies is understandable based on the increase in surface energy of a vicinal (1 0 0) surface as the step separation approaches the dimer reconstructed separation. 0 2005 Elsevier B.V. All rights reserved.

PACS: 81.05.Ea; 81.15.Hi

Keywords: Al. Reflection high-energy electron diffraction; Al. Surface structure; A3. Molecular beam epitaxy

1. Introduction

Epitaxial growth on GaAs (1 00) has been extensively investigated for its applications in optical and electronic devices. One common difficulty, however, has been the formation of micron-size three-dimensional structures during

*Corresponding author. Tel.: + 1 479 575 4217; fax: + 1 479 575 4580.

E-mail address: [email protected] (2h.M. Wang).

growth on singular GaAs (1 00) [l-31. Fortu- nately, a misorientation from GaAs (1 00) has been observed to result in a smooth growth front [1,2]. For example, GaAs (1 0 0) surfaces that are misoriented by 2"*, called vicinal GaAs (1 OO), have been widely used in practical device structures, such as lasers, giving improved performance. Vicinal GaAs (1 0 0) surfaces are characterized by a staircase array of steps with (1 0 0) terraces [4,5] and a step density that increases with the degree of misorientation. The resulting smaller terraces limit

0022-02481%-see front matter 0 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2005.02.059

JOURNAL OF APPLIED PHYSICS 98,094503 (2005)

Doped-channel micro-Hall devices: Size and geometry effects Vas. P. ~ u n e t s , ~ ) Yu. I. Mazur, and G. J. Salamo Deparrmertt of' Physics, Universiry of' Ar,kansns, Fayetteville. Arkrrnsas 72701

0. Bierwagen and W. T. Masselink Department of Physics, Humboldt-Universitat ~ I I Berlirt, Newroristrnsse 15, 12489 Berlin, Germarz)~

(Received 13 April 2005; accepted 27 September 2005; published online 1 1 November 2005)

Using a doped-channel Alo,3G~,7As-GaAs-In0,2Ga,,8As Greek-cross micro-Hall device as model structure, the size, and geomctry effects on the absolute magnetic sensitivity and noise were investigated. The data show that although the absolute magnetic sensitivity is higher for devices with larger active areas, the self-heating effects at moderate and high electric fields are more detrimental for larger devices than for smaller ones. For this reason, the scaled absolute sensitivity is greater for smaller devices. The geomeky is further optimized using finite element method simulations. Rounding the. corners within the Greek-cross geometry results in lower peak electric fields and therefore a higher signal-to-noise sensitivity. O 2005 Anlerican Institute of Physics. [DOI: 10.1063/1.2128472]

I. INTRODUCTION

Magnetic sensors based on the Hall effect in 111-V semiconductor materials are widely used for studies of magnetic phenomena in physical and material sciences. Because of their high magnetic sensitivity and low noise over a wide temperature rangc and because their lincar operation allows quantitative magnetic field imaging, miniaturized Hall devices offer potcntial advantages over othcr magnetic ficld imaging techniques. Using miniaturized Hall devices, the vorlices in superconductor materials.'-highly inhomogeneous magnetic fields produced by ferromagnetic nanoparlicles;-"ild the stray field of nanomagnet discs7 have been studied.

Micro-Hall devices have also made important contributions in the biological and medical science^.^-'^ Recently, using SQUID device as the tip in magnetic force inicroscope (MFM), the magnetic moment of an individual bacterial cell was measured to be about 10-l3 emu. Although the SQUID has an extremely high magnetic field detectivity, a drawback is that its operational temperature is 4.2 K, limiting its prac- ticality for ~ncdical diagnostics. Meanwhile, Hall devices with micron or sub-micron dimensions, operated at room tcniperature are an attractive alternative. To move in this direction, we must investigate factors limiting the ultin~ate detectivity of the Hall elemcnt; these are thc absolute magnctic sensitivity and the background noise. Both quantities depend not only on the choice of matcrial and on the design of heterostructure, but also on the device geometry and the device dimensions. The last two factors are crucially important when the device size reaches the sub-micron range. At thesc small sizes required for high spatial resolution ineasure- ments, Hall devices suffer from a high inhomogeneous electric lield within the device active area. The high electric field spikes will result in higher background noise and lower absolute magnetic sensitivity due to the self-heating effects as well as hot electron effects degrading device performance.

")~.lectronic mail: [email protected]

This work addresses the Greek-cross geometry optimiza- tion of micro-Hall devices with respect to device noise and magnetic sensitivity characteristics using finite element method (FEM). In addition, the size effects associated with scaling down device active area are considered. We show that signal-to-noise sensitivity and detection limit of magnetic sensor. respectively, strongly depend on the device size. Thc impact of sclf-hcating effects on the absolutc magnctic sensitivity is discussed.

II. EXPERIMENTAL DETAILS

Using a Ribcr 32-P gas-source molecular-beam cpitaxy system, Alo,3Gq,,7As-GaAs-Ino,2Ga,,8As QWs were grown on semi-insulating (100)-oricntcd GaAs wafers. A 1000 A GaAs buffer layer was initially grown on the substrate to ensure high material and interface quality of all subsequent layers. The buffer layer was followed by a 100 A, Be-doped (N,=1.5XlO'%n1-~) GaAs layer. a 3 5 0 A undoped Alo,3Ga,,7As barrier, a 103 A uniformly Si-doped (ND = 2 X 10'' GaAs, a 144 A Si-Gdoped In,12Gao.8hs quantum well, a 1.03 A uniformly Si-doped (N,=2 X 10" cm-7 GaAs channel, a 350 A undoped Alo3Gq,,7As barrier and a 70 Si-doped GaAs (ND=2 X lo i8 cap layer. The Sdoping position is ccntered wilhin strained InGalZs quantum well. The growth temperatures were 575 "C. 565 "C, and 630 " C for GaAs layers, InGaAs quantum well and AlGaAs barriers, respectively. The surface quality was monitored by a RHEED pattern both during the oxide desorption and subsequent growth.

Using standard optical photolithography, four-terminal Greek-cross micro-Hall devices, with square shaped active area of 20 X 20 pm, 10 X 10 ptn, and 5 X 5 p m , were fabricated. Chemical wet-etching was used for mesa device iso- lation. Ohmic contacts were formed by alloying an evapo- rated Au:Ge:Ni metal film. Ohmic contact resistance was optimized (0.05 Inmm) using a fast thermal ramp up to 420 "C followed by a 2 min anneal.

0021 -8979/2005/98(9)109450315/$22.50 98, 094503-1 O 2005 American Institute of Physics

Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:lljap.aip.orgljaplcopyright.jsp

IN?TIT~ .TF OF PHYSICS PLIRL.ISHLNG S M ~ R T M4TtRIAI.F i\Nl> STRIJCT~RFS

Smart Mater. Stl.uct. 14 (2005) 963-970 dni. I 0 108SIOOt~l-1'72hi lllj/01~1

Characterization of ultra-low-load (pN) nanoindents in GaAs(100) using a cube corner tip Curtis R Taylor ' , A j a y P MalsheL." G r e g o r y Salamo2, Robin N Prince', Laura Riester3 and Seong Oh Cho2

I Department of Mechanical Engineering, Univwsity of Arkansas, Mechanical Engineering Building, Fayetteville. AR 72701, USA

Dcpatmcnt of Physics, University of Arkansas, Physics Building, Faycttcvillc, AR 72701. USA

High Temperature Materials Labol-atory, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

E-mail: crtaylorCi;vcu.edu, aprn2@cngr.~1ark.edu, sala~noG>uark.cdu, [email protected], ~.ieste~lCi~ornl.gov and [email protected]

Received 13 April 2004, in final form 17 December 2004 Published 2 September 2005 Onlinc at stacks.iop.org/ShlS/14/963

Abstract In this study, nanoindentations are produced and characterized for the future patterning of nanostructures. Nanoindentation is performed on Si-doped (n-type) vertical gradient freeze (VGF) GaAs(100) and epitaxial GaAs(lO0) using a diamond cube corner indenter. Unlike previous research, the uniqueness of this work is in nanoindentation of GaAs at ultra-low loads (<400 pN) . Indentations of less than 200 nm in width are produced, and the mechanical properties of the two materials including hardness and elastic modulus are determined. The smallest indentations achieved are less than 60 nm in width and less than 7, nrn deep. The width, depth, shape, and volume of the indents are determined as a function of applied load using atomic force nlicroscopy (AFM). Also, the ratio of pile-up volunle to indent volume is determined. The experimental findings are discussed in relation to existing theories of indentation and for the directed self-asscmbly of nanostructures.

(Some figures in this article are in colour only in the clcctronic version)

1. Introduction

The dcvelopment of techniques for the sclectivc growth of nanostructures is essential for the realization of future nano- mechanical, electronic, and optical devices. Specifically, there exist lew techniques forthe highly selective growth or precision patterning of quantum dots.

Recent studies have shown that quantum dots can be patterned by the creation of mesa, trench, and hole features in GaAs by wet 11-31 and electron beam lithography [4]. Mesa and trench features have been shown to provide controlled growth of multiple linear positioned quantum dots, while hole

' Author to who111 any correspondellce should he addressed

features [4] were shown to provide highly selcctivc sites for positioning fewer dots and in some cases the growth of a single quantum do' at each hole site.

In addition, scanning probe tips have been uscd to produce trench and hole features in GaAs [5 , 61.

One such technique utilizes an oscillating atomic force microscope (AFM) tip 161 to make 70 nm wide x 6 nm deep trenches and 60 nm diameter x 10 nm deep hole-like indents. Multiplc dots were obscrved insidc the indents and most of them grew on the sidewalls of the indent. The results of these studies suggest thal hole features on the surface of GaAs provide for highly site-specific growth and patterning of quantum dots. However, the mechanism for growth at the hole sites is not well understood. It has been speculated that

0964-1726/05/050963+08$30.00 0 2005 10P Publishing Ltd Printed in Ihe UK 963


JouRwoFCRYSTAL GROWTH

Journal of Crystal Growth 275 (2005) 410-414 www.elsevier.com/locate/jcrysgro

Orientation dependence behavior of self-assembled (In,Ga)As quantum structures on GaAs surface

Sh. Seydrnoharnadi*, Zh.M. Wang, G.J. Salarno Department of Physics, University o f Arkunsus, Fuyetteville, AR, 72701, USA

Received 27 August 2004; accepted 5 December 2004 Communicated by D. W. Shaw

Available online 19 January 2005

Abstract

We report on the formation of (In,Ga)As self-assembled quantum structures grown on different orientations of GaAs along one side of the stereographic triangle between (1 0 0) and (1 1 1)A. The samples were characterized by atomic force microscopy. A systematic transition from zero-dimensional (In,Ga)As quantum dots to one-dimensional quantum wires was observed as the substrate was varied along the side of the triangle between (1 0 0) and (1 1 1)A. An explanation for the role of the substrate in determining the size of the nanostructure is proposed. 0 2004 Elsevier B.V. All rights reserved.

PACS: 79.60.J~; 85.30.V~; 81.05.Ea; 81.15.Hi

Keywords: Al . Low-dimensional structures; Al. Nanostructures; Al. Surface structure; A3. Molecular beam epitaxy

1. Introduction

The orientation of the substrate surface can play an important role in the formation of (In,Ga)As heteroepitaxial quantum structures grown on GaAs [1,2]. Growth of self-assembled InAs or (In,Ga)As quantum dots (QDs) on GaAs (1 0 0) [1,3], GaAs (3 1 l)A and B [2,4-9], GaAs (4 1 l)A [lo-121, as well as quantum wires (QWRs) on

'Corresponding author. Tel.: + 1 479 575 7660; fax: + 1 479 575 4580.

E-mail address: [email protected] (Sh. Seydmohamadi).

GaAs (3 1 l)A [6,9] and (4 1 l)A [12] are typical examples of the strong potential of high-index surfaces to act as templates for the growth of self- assembled QWRs and QDs.

In this paper, we investigate the role of the substrate by studying the formation of (In,Ga)As self-assembled quantum structures grown on different orientations of GaAs along one side of the stereographic triangle between (1 00) and (1 1 l)A. The surfaces studied by molecular beam epitaxy (MBE) and atomic force microscopy (AFM) include the (1 0 O), (7 1 1)A, (5 1 1)A, (41 l)A and (3 1 l)A. Fig. 1 is a schematic,

0022-02481s -see front matter 2004 Elsevier B.V. All rights reserved. doi:l0.1016/j.jcrysgro.2004.12.011

JOURNAL OF APPLIED PHYSICS 98. 0537 I I (2005)

Strong optical nonli~iearity in strain-induced laterally ordered Ino.4Gao.6As quantum wires on GaAs (31 l )A substrate

Yu. I. ~ a z u r , ~ ) Zh. M. Wang, G. G. ~arasov ,~) H. Wen, V. ~t re lchuk,~) D. Guzun, M. Xiao, and G. J. Salamo Physics Depnrtn7eirt. Universiry of'drkansas. Fnvetreville, Arkarrsiis 72701

T. D. Mishima, Guoda D. Lian, and M. B. Johnson Center f i r Sernicondurror Pitysics in Nanostructrrre~, University of' Oklahoma. Norrirun. Oklahorn~l 73019

(Received 1 February 2005; acccpted 29 July 2005; published online 13 September 2005)

Straminduced laterally ordered In,, ,G% &s on (3 1 l)A GaAs template quantum wires have been fabricated and identified with cross-section transmission electron microscopy technique to be of average length -1 p m , and on average width and height of 23 and 2 nm, respectively, under InGaAs coverage of six monolayers. The photoluminescence spectrum of a sample demonstrates unusually strong optical nonlinearity even at moderate excitation densities. he excitonic peak energy blueshif 5 by -25 meV without essential contribution of the quantum wire excited states at elevating excitation density. Strong decrease of the polarization anisotropy and incrcase of the energy of excitonic photoluminescence are attributed to a combined action of the phase-space filling cffccts and thc screening of thc ~nternal p~ezoclcctric field by frce carrlcrs. O 2005 Arnericai~ Institute qf Ph~~sics. [DOI: 10.106311 .2039009]

I. INTRODUCTION Enhancement of the optical nonlinearity has been rcvcalcd

Recently, self-assembled quantum wires (QWRs) of various shapes, sizes, densities, optical and electronic properties have been fabricated1-' promising further advances in optoelectronic devices (e.g., lasers6 and modulators7). The two-dimensional (2D) confinement inherent to QWRs produces a singularity in the density of states (DOS).' The narrowing of thc DOS leads to a lowcr excitation threshold for phase-space filling in Q W R S . " ~ ~ ~ enhancing nonlinear optical cffccts important for applications in optical communications.

One of the interesting optical featurcs associated with strain-driven self-assembled QWRs is an intrinsic photolu- ~nincsccnce (PL) polarization a~~isotropy duc to the complicated structure of the valence band at the center of the Bril-

'0 louin zone.' Thc con~plex, nontctragonal strain deformation, developed due to the lattice mismatch, splits the degenerated valence band into a co~nplicated subbanci s t r ~ c t u r e . " ~ ' ~ resulting in both the polarization anisotropyl' and lateral piezoelectric fields." These fields could be important for optical nonlinearities due to screening of the field as well as the range of the field away from the strained layers. Indeed, it has been demonstrated in a series of (110) InAslGaAs QWR structures that in high excitation density PL experiments produce a pronounced (up to 22 meV) blueshift of the PL lines, simultaneously with a clear reduction of the linewidth.I4 In these experiments, the observed blueshift and linewidth re-

also in quasiplanar sidewall QWRs on GaAs (311)A substrate by means of continuous-wave PL."

While screening would seem to explain the optical behavior of QWRs undcr high-density optical illumination. this is not the case due to the fact that there exist numerous contradictory observations of the PL peak energy behavior under high optical excitation density for strained QWR systems: from a sizable rcdshift -10 mcv.I6 virtually no

to a sizable blueshift [-25 meV (Ref. 19); 17 ~ n e V (Ref. 20)]. It is clear that different mechanisms can contribute to this behavior at high excitation density such as band filling. excitonic correlations, band-gap renormalization, and disorder. In fact, the nature of the PL peak shift and the polarization anisotropy properties in QWR systems are not yet well understood in spite of good potential for applications in low-threshold lasers and exotic light modulators.

In this study, we examine the nonlinear optical properties of strain-induced laterally ordered (SILO) In0,4G%oAs QWRs on the GaAs (3 I 1 )A substrate. The QWRs have been unambiguously identified with transmission electron microscopy (TEM) and PL mcasurcments carricd out at the moderate excitation optical densities. We demorlstrate the existence of interesting cxcitation and tempcrature-dependcnt polarization anisotropy and a strong blueshift in the PL spectrum coming from the SILO QWRs that give evidence for strong nonlinear optical effects.

duction were attributed to the screening of the internal piezoelectric fields by photogcnerated carriers. In addition, the 11. SAMPLES AND EXPERIMENTAL DETAILS photoexcited carriers could also result in the screening and eventually bleaching of thc exciton binding energy in QWRs. Thc QWR structures for this study were grown by

molecular-beam epitaxy on a GaAs(3 1 1 )A substrate. The . .

s) Electronic mail: [email protected] GaAs wafer was iirst covered with a buffcr layer of h ) ~ n leave from Institute of Semiconductor Phvsics. National Academv of 0.5-pln thickness grown at 600 "C. For the (In,Ga),As d e ~ o -

, , ~~ - , ~ ~

Sciences of Ukraine, Prospect Nauki 45, 03028 Kiev, Ukraine. sition, thc substrate tclnperature was rcduccd to 540 "C, the

O 2005 American Institute of Physics

Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:lljap.aip.org/jap/copyright.jsp

JOURNAL OF APPLIED PHYSICS 98. 053515 (2005)

Tailoring of high-temperature photoluminesce~ice in InAsIGaAs bilayer quantum dot structures

Yu. I. ~ a z u r , ~ ) Zh. M. Wang, G. G. ~a rasov ,~ ) Vas. P. Kunets, and G. J. Salarno Deparlment of Physics, Universizy of Arkansas, Favelleville, Arknr~.scr~ 72701

Z. Ya. Zhuchenko Institure of Semiconductor Pi~ysics, National Academy of Srlences, Prospect N~lukr 41, 0.3028 Kiev, Ukrlrine

H. Kissel Ferdincairi-Br~zrtr1-lt~stitut ,Fir HochsffrequerrztecImik. Gustov-Kirchbqff-Strrrs.re 4, 12489 Berliri, Germally

(Received 14 February 2005; accepted 27 July 2005; published online 8 September 2005)

Temperature-dcpcndent photolu~nincscence is investigated in bilayer InAsIGaAs quantum dot structures with constant TnAs deposition O1 in the seed layer, but variable deposition O2 in both the sccond laycr and the GaAs spaccr laycr. It is shown that intcrlaycr coupling, lcading to thc formation of asymmetric quantum dot pairs, strengthens the high-temperature photoluminescence and strongly influenccs carrier relaxation channels. Wc rcport that radiativc rccombination and carrier capture efficiency by the quantuni dots in the second layer can be tailored us~ng the deposition O2 and the GaAs spaccr thickness. 0 2005 Anzerrcan Itzstit~ite of Physics. [DOI: 10.106311.303927 11

I. INTRODUCTION

The high-temperature behavior of quantum dot (QD) systems is important to understand due to the fact that a quenching of photoluminescence (PL) and a rapid redshift of the emission energy are observed with increasing temperature, both of which are harmful to QD devices.'-4 For example, at high temperatures PL quenching in the InAsIGaAs QD system is caused by the carrier escape from QDs followed by subsequent nonradiativc decay in barriers or wetting layers (WLs). In addition, while for low-density QD arrays the probability of carrier tunneling is small, in dense QD systems, carrier escape can take place due to lateral coupling between neighboring QDS.'-'' Tclnperature elevation can also activate carriers into the barrier or WL due to a step-by-step transition wherc they arc retrappcd by larger QDs or captured by nonradiative centers in the vicinity of the barrier or WL.' Together thesc processes lead to a quenching of the PL at high temperatures resulting in a decrease of the PL intensity as well as a peak energy shift to lower energy.

In order to minimize high-temperature effects, QD systems with higher barriers, such as InAs/AIxGal-,As, have been proposed to prevent carrier escape."-" Such systems provide stronger confinement due to a larger band-gap difference at the r point. For example, between InAs and AlAs (-1.7 eV at low temperature) in comparison with InAs and GaAs (- 1.1 eV at low temperature).9

Temperature effects are particularly interesting in multilayered QD structures. For example, in bilayer InAsIGaAs QD systems, PL measurements have revealed a blueshifted emission of the second QD layer with respect to the seed

14 layer. This is attributed to an enhanced intermixing during the capping stage of the sccond-layer QDs. The blueshift is

"~llccrronic mail: [email protected] h ' ~ n leave fro~n Institute of Semiconductor Physics, National Academy of

Sciences of Ukraine, Prospect Nauh 45, 03028 Kiev, Ulwaine.

somewhat reduced for annealed spacer layers due to a reduction in the strain in the spacer laycr. Thercforc. intcrcstingly, the PL ernission spectrum from the small quantum dots (SQDs) of the sccd layer and the large quantum dots (LQDs) of the second layer can be made more coincident by increasing the amount of InAs deposited in the sccond laycr or by annealing the spacer layer. Thus, there exist various factors that can affecl the PL in bilayer structures. In this paper, we show that interlayer coupling is a crucial factor for high- temperature stabiliz.ation of the PL response in such a system.

II. SAMPLES AND EXPERIMENTAL DETAILS

Varying InAs coverage in the second layer 02. GaAs spacer layer (dsp) between the seed and second layers, growth temperatures, and annealing times for each QD layer, different vertically correlated pairs of unequal sized QDs (bilayer QD structures) have been grown. Such a system is called an "asymmetric QD pair"'5 (AQDP) and is ideally suited both for the exploration of multilayered QD structures and for effective control over the uniformity of size, shape, and density of the QDs in the second layer. This latter property increases the potential of AQDPs for applications. The AQDP samples explored here were grown using a solid- source molecular-beam epitaxy chamber coupled to an ultrahigh-vacuum scanning tunneling ~niroscope (STM). The structures were grown on a GaAs (001) substrate, followed by a 0.5-pm GaAs buffer layer, 28-nm AI0,3G%,7As layer, 57-111n GaAs layer, 1.8-ML (monolayer) InAs secd layer, variable GaAs spacer layer, variable InAs QDs layer covered by 57-nm GaAs laycr, and 28-nm Alo,3G%,7As layer. The total structure was capped by a 20-nm GaAs layer. Before tieposition of a 1.8-ML lnAs seed laycr. a 10-nun annealing at substrate temperature of 580 "C was used to provide a nearly dcfect-free atonlically flat surface. The sced QD layer

0021 -8979/2005/98(5)/053515/6/$22.50 98, 05351 5-1 O 2005 American Institute of Physics


APPLIED PHYSICS LETTERS 87, 073 108 (2005)

Nanoscale dislocation patterning by wltralow load indentation Curtis R. ~ a ~ l o r ~ ) Departmellt of Phvsics, University of'Ar.kansas, Fayetteville. Arkansas 72701

Eric A. Stach School of'Materials Engineering, Purdue Universit)! West L.ufi7yetre. Indiana 47907 and Natiortul Center f o ~ Electron Microscopy, Lawrence Berkeley Nariorrctl Lahorcltor)~, Berkele)! California 94720

Gregory Salamo Depnrrrnenr of' Physics, Universit); of Arknnsas, Faverteville, A~.kansas 72701

Ajay P. ~ a l s h e ~ ) Departmer?r of Mechanical Erlgineering, Uiliversit); oj'Arknlisns, Fa)~etreiille. Arkansas 72701

(Received 24 February 2005; accepted 5 July 2005: published online 10 August 2005)

The use of nanoindentation as a dislocation patterning technique for self-assembled nanostructures is investigated. In this context wc havc studied the behavior of GaAs undcr ultralow load indentation conditions. It is shown that periodic dislocation arrays are formed and can be well controlled by nanoindentation. Transmission electron n~icroscopy (TEM) reveals that the crystal deforms solely by dislocation activity with no evidence of stacking faults, twinning, fracture, or phase transformation. The resulting strain field is highly localized, indicating that ultralow load nanoindentation may provide an excellent means to mechanically bias nanostructure nucleation and patterning during subsequent crystal growth. O 200.5 American I~zstitute of Physics. [DOI: 10.1063/1.2009825]

The ability to pattern self-assembled low-dimensional semiconductor ilanostructurcs is cxtremely important for thc realization of novel electronic and photonic device t e c h n o ~ o ~ i e s . ' , ~ Self-assembly via the Stanski-Krastanow growth modc3 results in spontaneously nuclcatcd structures, such as quantum dots (QDs) or wires, that are randomly distributed on the epitaxial surface. Also, the nanostructure size and morphology are highly dispersive. For device applications, precise control of quantum structure size. location. and arrangement is critical and needs an effective bias to tailor self-assembly.

Recent work has shown that guided sclf-assembly and spat~al patterning are achieved by the formation of subsurface misfit disl~cations.~-' The localized elastic strain fields of dislocations provide strain-relaxed surface sites that enhance surface adatom migration resulting in specific patterns. Elasticity calculations and transmission electron microscopy (TEM) show that the linear QD chains are aligned with the underneath dislocations. However, precise spatial control of misfit dislocations is not feasible due to the arbitrary formation of dislocation sources at the heterointerface. Thus the patterns formed are irregular. If the placement and formation of subsurface dislocations can be controlled, then it would be a highly effective patterning mechanism.

In this letter, unlike the traditional use of nanoindentation for hardness measurement, we investigate the use of nanoindentation as a tool for the injection of dislocations at precise positions in semiconductors for the creation of periodic dislocation arrays. The novelty of using nanoindentation is the ability to mechanically perturb the crystal lattice by accurate control of the applied stress. rate of deformation, and volume of dislocated regions. However, despite the potential of nanoindentation, this technique is dependant upon

"'~lectronic mail: [email protected] h ' ~ ~ t l , o l . to who111 correspondence should be addressed; alectmnic mail:

[email protected]

understanding material deformation mechanisms at ultralow loads. It has been shown that nanoindentation of semiconductors produces a range of defects in the crystal lattice including dislocation^.^'^ phase transforlnations.' and subsurface f r a ~ t u r e . ~ ~ " The suppression of these defects and phases is critical to enable nanoindentation as a dislocation patterning technique.

Towards this goal, we seek to understand the material deformation of GaAs at extremely low loads (<0.2 niN) that approach the threshold of plasticity. GaAs is a material of intense interest in the electronics industry because its direct band gap is suitable for ultrafast optical devices. Defonna- tion of GaAs has been studied extensively in the past using high load (50-200 mN) (Refs. 8 and 12) as well as low load (0.2-8 mN) (Refs. 9, 13. and 14) indentation combincd with TEM characterization of the deformation mechanisms. How- ever, relatively less is known for loads in the lower regime of <0.2 mN, which is of interest for dislocation-assisted patterning. At such low loads the presence of only dislocations is expected, however due to the increased applied stress from the sharpness of the indenter (<40 nm) needed to produce nanoindents, it is highly likely that othcr defects are nuclc- ated.

Our approach is to investigate thc deforination of GaAs(100) near nanoindented regions by cross-sectional TEM (XTEM). The type of defects, their density, preferred deformation mechanisms. and distribution pattern are analyzed as a function of load and in relation to their use for subsequent crystal growth and potential for QD patterning.

A ~ r i b o ~ n d e n t e r ~ (Hysitron Tnc.) was utilized for ultra- low-load nanoindentation. Nanoindentations were performed at room temperature with a diamond 90" NorthStarTM cube corner (tip radius <40 nm) indenter. Indents were made on epitaxial GaAs(100), which was prcpared by growing a 500 nni buffcr layer of GaAs by molecular bcam cpitaxy (MBE) on an epiready Si-doped GaAs(100) wafer, followed

0003-6951 /2005/87(7)/073108/3/$22.50 87, 0731 08-1 O 2005 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcopyright.jsp

IEEE ELECTRON DEVICE L E P E R S , VOL. 26. NO. 9. SEPTEMBER 2005

Broad-Band Photoresponse From InAs Quantum Dots Embedded Into InGaAs Graded Well

Jie Liang, Ying Chao Chua, M. 0. Manasreh, Euclydes Marega, Jr., and G. J. Salamo

Abstract-A broad-band photoresponse is obtained from undoped InAs multiple quantum dots grown by the molecular beam epitaxy (MBE) technique on a (100) GaAs substrate. The quantum dots were embedded in an In, Gal-, As graded well where the In mole fraction is chosen in the range of 0.3 2 x 2 0.0. The photoresponse of the reversed biased device, obtained at 80.5 K and using the normal incident configuration, was found to span the spectral range of 3.5 - 9.5 pm. While the photoresponse is significantly high under the influence of the reverse bias voltage, the forward bias photoresponse is found to be negligible.

Index Terms-Indium compounds, infrared detectors, infrared measurements, quantum dots.

D UE to the absence of energy-momentum dispersion relation in 111-V semiconductor quantum dots [I], the in-

tersubband transition selection rules that govern the electron- photon interaction are irrelevant. Thus, unlike the selection rules associated with the intersubband transitions in n-type multiple quantum wells [2]-[4], the photons can be absorbed at any light incident angles by the electrons that undergo intersubband transitions in multiple-quantum dots [5]. This feature is one of the main attractions of the multiple-quantum dots as infrared detectors. since a metallic grading layer [6] is not required, which leads to a simpler device fabrication.

Another attractive feature of the quantum dots is that doping is not required. For an equilibrium steady-state condition, such as optical absorption measurements, the Fermi energy should be above the ground bound state in the quantum dots. Doping in this case is required in order to observe the intersubband transition [7]. On the other hand, doping of quantum dots is not required for the nonequilibrium conditions, such as photoconductivity measurements under the influence of a bias voltage. For the nonequilibrium measurements, quantum dots are populated by the charge carriers that are generated from the contact layers and swept under the influence of the bias voltage.

Recently, broad-band long wavelength infrared detection based on multiple-quantum wells were investigated by several

Manuscript received May 19, 2005; revised June 20, 2005. This work was partially supported by the Air Force Office of Scientific Research. E. Maraga, Jr.. was supported by Fundacao de Pesquisa do Estado de Sao Paulo and CNPQ, Brazil. The review of this letter was arranged by Editor P. Yu.

J. Liang, Y. C. Chua, and M. 0. Manasreh are with the Department of Electrical Engineering, University of Arkansas, Fayetteville, AR 72701 USA (e-mail: [email protected]).

E. Marega, Jr. is with the Department of Physics, University of Arkansas, Fayetteville, AR 72701, and also with the Intituto de Fisica de Sao Carlos, Uni- versidade de Sao Paulo, Sao PauIo 13560-590, Brazil.

G. J. Salamo is with the Department of Physics, University of Arkansas, Fayetteville, AR 72701 USA.

Digital Object Identifier 10.1109LED.2005.854392

groups (see for example [8]-[lo]). Preliminary results on broad-band InAs quantum dot matrixed into undoped GaAs barrier detectors were reported by Phillips et al. [I I]. The latter report shows a normalized response with AX/X -- 0.6, where AX is the full width at half maximum and X is the peak position wavelength. While the results of Phillips et al. are encouraging, these results could not be reproduced by us using the same structures. A simple InAsIGaAs or InGaAsIGaAs multiple-quantum dot structure produces narrower band photoresponse [5] and narrow absorbance spectra [7].

In this letter, we report on a new broad-band photoresponse result obtained from InAs quantum dots embedded into an In- GaAs graded well long wavelength infrared photodetector. The broad-band behavior is an important feature for certain applications, such as hyperspectral and ultra-hyperspectral long wavelength infrared sensors.

The sample grown for the present investigation is an n-i-n structure, which is deposited on a (100) semi-insulating GaAs substrate. A buffer layer made of n-type 0.5 pm GaAs:Si with [Si] = 1.0 x 10'' cm-3 and 0.3 ym GaAs:Si with [Si] = 1.0 x 1017 cm-3 is grown at the surface of the substrate. Then 2.2 monolayers (ML) of InAs quantum dots were grown using the Stranski-Krastanov mode followed by the growth of a 20 ML graded In, Gal-, As well, where x is varied between 0.3 down to zero. A 180-ML GaAs layer is then grown on top of the graded well to form the barrier. The InAsI In, Gal-,As/GaAs structure is grown at 520 "C and repeated ten times to increase the optical path length. The cap layer is made of 0.4 pm GaAs:Si, where [Si] = 1.0 x 10" cmP3 is followed by 0.2 pm GaAs:Si where [Si] = 1.0 x 1018 cmp3. The buffer layer was grown at 580 "C while the cap layer was grown at 520 "C. The transformation from two-dimensional to three-dimensional growth mode was monitored by using in situ reflection high-energy electron diffraction. Both the buffer layer and cap layers were used as contact layers. The quantum dot density was estimated from scanning tunneling microscopy images to be on the order of 3.0 x 10" cm-'. The average planar area of the quantum dots is - 30 x 30 nm and the average height of the dots is - 13.5 nm. Mesa structures on the order of 400 x 400 ym were fabricated using wet-etching and photolithography techniques. Gold-germanium alloy met- allization pads were deposited on the mesas for wire bonding. For a comparison reason, two InAs quantum dot samples where grown. The structures of these two samples are exactly the same as the sample described above except one sample is embedded in 20 ML Ino.15 Gao,85As quantum well and the second sample was grown without any quantum wells. For simplicity, the samples are referred to as an InAs dot-graded well, an InAs

1.00 O 2005 IEEE

Control on self-organization of lnGaAs/GaAs(100) quantum-dot chains Zh. M. an^,^' Yu. I. Mazur, K. Holmes, and G. J. Salamo Departmerrt of Ph.y.ric.r, U~ziver.riry of Al.kan.ros, Fayertcville, Arkor~sas 72701

(Received 23 January 7005; accepted 19 March 2005; published 25 July 2005)

The spontaneous formation of long chains of quantunl dots during the growth of Ii~C;aAs/CiaAs multiple layers has been reported recently. The effects of In content and spacer on the evolution of dotchains are investigated in the present work. By reducing the In content in the InGaAs layer, the quantum dots in chains are more connected and finally arrays of quantum wires would folm. By changing the GaAs spacer layer thickness, the vertical and also lateral spacing between dotchains can bc continually tuned. Thc capability to insert a thick layer of AlGaAs as part of the spacer laycr enables us to fabricate InGaAs quantum-dot chains without vertical electronic coupling. The achieved control of self-asscmbly of organized InGaAs quantuin dots inay be advantageous for novel oploelectronic applications. 0 2005 American Vcrcuurn Society. [DOI: 10.1 11611 . I 9425091

I. INTRODUCTION II. EXPERIMENT

Over the last decade, strain-driven sclf-asscinbly has been developed into a proinising method for the fabrication of InGaAs quantum dots (QDs) in a GaAs matrix.'" Corre- sponding interest in an ensemble of randomly distributed InGaAs QDs is driven by potential applications for optoelectronic devices, such as low-threshold lasers and normal- incident intersubband infrared detector^.^ However, some applications, such as QD-based optical memory devices, will require precise control of the QD position for information addressing. The lateral spatial ordering of QDs has been hin- tlered by the stochastic nature of self-assembly, which raises a question of whether this process can by itself achieve the desired lateral control of QD positioning or whether outside help, such as lithography, is needed. While hybrid methods made up from self-assembly and lithography have demonstrated successful lateral control of the QD positio~~,4.5 our recent results 011 QD lateral ordering prove there is still plenty of opportunity to play with self-organization aloi~e.~-" During the growth of TnGaAs/GaAs multiple layers, QDs not only vertically align by themselves as the result of the strain interaction through spacer ~ a ~ e r s ' ~ , ' ~ but they also laterally line up to form ordered In particular long chains of uniform QDs over 5 pnl in length have been achieved,' and they are unyielding to a moderate deviation of surface orientation from ~ a ~ s ( 1 0 0 ) . ' The dot-chain structures are dislocation-free as evidenced by transmission electron nlicroscopy (TEM), and photolu~ninescence and photoconductivity measurement reveal their high optical qua~ity.6.1'

To achieve a routine procedure for the fabrication of In- GaAs dot chains and to gain a better understanding on their formation mechanism, we have investigated the evolution of the dot chains with varying In content, spacer layer, and substrate temperature. The results demonstrate our control on the vertical and lateral spacing between dot chains and between the QDs in the chains.

")~.lectronic mail: [email protected].

All samples are grown by a solid-source molecular beam cpitaxy which is connected to a scanning tunneling microscopy (STM) through an ultra-high vacuum module. For STM lnorphology measurements, epitaxial ready N-type GaAs(100) substrates are used. For atomic force microscopy (AFM) measurements, scmi-insulating GaAs(100) substrates are used. Before the growth of InGaAs multiple layers, a 0.5-pm GaAs buffer was grown at 580 "C to achieve a flar surface for every sample. If not specially mentioned, all the InGaAs nlultiple layers were grown at the substrate temperature of 540 "C. The growth of InGaAs layers is monitored in sit11 by refection high-energy electron diffrac~ion. Seven~een layers of InGaAs are spaced by GaAs layers or with AlGaAs added as shown in Fig. 1. While the layered slructure for every sample will be separately described in this article. the detail growth conditions can be found elsewhere.'-"

Ill. RESLILTS AND DISCUSSION

AFM images in Fig. 2 explore the evolution of QD chains as a function of In content. Of course. growth with a differ- cnt In content requires a different amount of material, i.c., critical thickness, deposited to form QDs. The appearance of

I GaAs(100) substrate ( /

FIG. 1. Sketch of one typical lnyrred structure for samples imaged by AFM or STM in this work

1732 J. Vac. Sci. Technol. B 23(4), JulIAug 2005 0734-211W2005/23(4)/1732/4/$22.00 02005 American Vacuum Society 1732

HTML AESTRHCT * LlnKS

JOURNAL OF APPLIED PHYSICS 98. 014506 (2005)

Highly sensitive micro-Hall devices based on Alo.121no,88Sb/lnSb heterostructures

Vas. P. ~ u n e t s , ~ ) W. T. Black, Yu. I. Mazur, D. Guzun, and G. J. Salamo 13epc1rrment oJ Pl~ysics. Universily of Arkansas. Fiz,vetieville, Arkar~,rtrs 72701

N. Goel, T. D. Mishirna, D. A. Deen, S. Q. Murphy, and M. B. Santos Dcpartmerlt of Phy.rics and Astrotroti~y, Ut~iversity qf Oklahoma, Norman. Oklaliorna 73019

(Received l l February 2005; accepted 20 May 2005; published online 13 July 2005)

Micro-Hall devices based on modulation-doped Alo,,,Ino,88Sb/InSb heterostructures are labricated and studied in terms of sensitivity and noise. Extremely high supply-current-related magnetic sensitivities of 1800 V A-' T-' at 77 K and 1220 V A-I T-I at 300 K are reported and observed lo be independent of the bias current. The detection limit of the devices studied at low and room temperature are at nanotesla values throughout the broad frequency range from 20 Hz to 20 kHz. The low detcction limit of 28 nT at 300 K and 18 nT at 77 K wcre found at high frequencies where the Johnson noise is dominant. A measured detection limit per unit device width of 630 pT lnm Hz-'" is reported indicating the potential for picotesla detectivity. O 2005 Atnericon Itlstitrcte of P h ~ s i c . ~ . [DOI: lO.l063/1 ,19548671

I. INTRODUCTION

Magnetic sensors based on the Hall effect in 111-V semiconductor materials have found wide application in both physical and material sciences. For example, sensors with micron-range lateral sizes and an absolute magnetic detectivity in the nanotesla range. have improved scanning techniques for magnetic field imaging, such as scanning Hall probe microscopy"2 and magnetic force microscopy.' Even more recently, a method for the dctcction of nuclear magnetic resonance and electron-spin resonance has been proposed using advanced micro-Hall device^.^

In addition to these applications, techniqucs using micro- Hall devices are of particular interest for biology and medi- cine where nondestructive and noncontact techniques. as well as room-tenlpcrature operation are required. To datc, the most sensitive magnetic sensor is the superconducting quantum interference device (SQUID). Using SQUID devices, magnetic fields ranging from several femtotesla up to 9 T can be detected. Such high detectivity is attractive to medi- cine since the neuromagnetic field of the human brain is only a few tenths of a f ~ m t o t e s l a . ~ Rcccntly, using a SQUID as the tip in a magnetic force microscope, the magnetic moment of an individual bacterial cell was ~ n e a s u r e d ~ - ~ to bc about 10-l3 emu. While the high sensitivity of the SQUID can open incredible medical opportunities, the drawback is that the operational temperature of the SQUID is 4.3 K, limiting its potential for medical investigations.

When room-temperature operation or portability is required we may consider developing micro-Hall devices as a replacement for SQUIDS for biomedical applications. To move in this direction, we must investigate which factors determine the absolute detection limit of the magnetic sensor. The two most important factors are the absolute inagiletic sensitivity and the background noise. Absolute magnetic sen-

" ~ e c t r o n i c mail: [email protected]

sitivity scales with the electron drift velocity, device sizc, and depends on the geometry. At a fixed device geometry and sizc, the absolute magnetic sensitivity will bc deter~nined by the electron drift velocity. Thus, materials with the highest electron drift vclocity are the most attractive. With appropriate doping, good material choices having a high electron velocity arc bulk InAs (Rcf. 9) and InSb (Rcf. 10) In addition, with selective doping and a reasonable doping profile, similar high mobility ~natcrials can bc fabricated as hetcro- structures with low noise due to a nanosize conductive channel.

Here, we report on the studies of micro-Hall effect devices fabricated lrom modulation-doped Alo,121~o~88Sb/InSb heterostructures. The detection limit of 7-8 nT at 300 K and 18 nT at 77 K is found as a result of the high absolute magnetic sensitivity and moderately low-noise level. We project that further transpol.1 and noise stuciies of these materials in co~ijunction with further molecular-beam epitaxy (MBE) growth will result in picotesla-detectivity micro-Hall devices.

II. EXPERIMENTAL DETAILS

The layer sequence lor the A1,,Inl-,SblInSb quanlum- well structures in this study is shown in Fig. 1. All growths were performed in an Intevac Gen I1 n~olecular-bean1 epitaxy system. A I - p m AlSb nucleation layer, which has a lattice constailt about midway between GaAs and InSb. was grown on a semi-insulating GaAs (001) substrate. The buffer layers that follow (1 p m of Alo~17_Ino,s8Sb, 0.2 p m of A10,241no,,6Sb, and 2 p m of A10,121no,88Sb) have been shown to substantially reduce the microtwin and dislocation densities in the quantum-well layer." Thc substrate temperalure of -450 " C used during buffer layer growth was lowered to 355 " C prior to the deposition of the first Si Sdoped layer in ordcr to minimize Si dopant compensation."

A Si Sdopcd layer, with a iict donor dcnsity of 4

0021 -8979/2005/98(1)/014506/6/$22.50 98, 014506-1 O 2005 American Institute of Physics


PHYSICAL REVIEW B 71, 235313 (2005)

Nonresonant tunneling carrier transfer in bilayer asymmetric InAsIGaAs quantum dots

Yu. I. Mazur,* Zh. M. Wang, G. G. Tarasov: and G. J. Salamo Departmeift of Physics. Univer.sit)? 01' Arkanscu, Eiryetteville. Arkun.~us 7-7701, USA

J. W. Tomnl and V. Talalaev Max-Born-Instittrt ,fur Nichtlineart' Optik ~rnd K~rrzzeitspektroskopie, Ma-Born-Srrusse 2A, 12489 Berlin, Gerrncln!

H. Kisscl Fer-dii~und-Bruun-Inslim jiir HGchstfrequenztechnik, G'ustc113-Kirchhoff-Sti.asse 4, 12489 Berlin. Germany

(Rcccivcd 15 October 2004; revised manuscript rcceivcd 1 March 2005; published 13 June 2005)

Camer hansfer in InAslGaAs asymmetric quantum dot pairs has been studied by means of continuous-wave and time-resolvcd photolumincsce~~cc in a bilayer InAsIGaAs quantum dots systcm. Thc dependcncc of the tunneling time on the thickness of the separation layer is dctem~ined and the tunneling time is found to span the rangc from 250 to 2500 ps. A n~icroscopic nod el of carrier tra~~sfcr, including nonrcsonant clcctro~~ tunncling from a direct into a cross excitoil state, with subsequent generation of two direct excitons in adjacent quantum dot laycrs, is proposed.

DOI: 10.1 103lPhysKcvH.71.235313 PACS numbcds): 78.55.Cr, 78.47.+p, 78.66.Fd

I. INTRODUCTION

Many proposed electronic or optoelectronic applicatioils of quantum dots (QDs) require uniformity in performance and consequently in both QD size and shape.' One promising approach pursued to achicve the rcquircd homogeneity is the growth of nlultiple layers of QDs2 In this approach, layers of thrce-dimensional (3D) islands are each scparated by a laycr of a material that acts as both a barrier and a spacer layer. Control over the spaccr laycr thickness has beell shown to influence the degree of strain transmitted from the first QD laycr into the subsequent separation layer, so that nucleation of the second QD layer is vertically ordered above the first. In fact, a significant degree of both vertical and lateral ordering has been achieved when multilayers of QDs are prepared in this

Given some early success with this approach it has become interesting to investigatc thc behavior of orjianized arrays of QDs. Developing an understanding of such struch~res is not necessarily casy, howevcr, sincc each dot layer must be close enough to influence another one, making tunneling between dots more complicated. In order to cxplore the behavior of 3D arrays, some recent investigations have focused on the simpler structure consisting of only two QD layers separated by one spacer For example, a recent paper12 has rcported 011 an investigation of bilayers of InAs QDs with GaAs as the barrier, giving photoluminescence (PL) up to = 1.4 p m at room temperature for a second layer with a reasonable density of 2 X 10" cin-' and a remarkably narrow PL linewidth of as low as 14 meV.

For a bilayer QD structure, the two layers can either be grown identically or under different conditions. Using different InAs deposition values, growth temperature, or annealing time for each QD layer, a vertically stacked pair of unequal sized QDs can be realized. Such a system has been called an "asymmetric QD pair" (AQDP) in analogy to the extensive work on asymmetric double quanhun-well (ADQW)

Not only call studies 011 the AQDP lead to a bctter

undcrstanding of multilaycrcd QD stluch~res and hence more control over uniformity of size and shape, but they may also prove to be useful for applications. For exanlple, AQDPs sharing a single electron in each pair have attracted great interest in connection with the realizatioil of quantum bits and quantum computation. '

Of course a good understanding of the bilaycr AQDP requires exploration of many phenomena. One important characteristic of such a system is thc tunncling time of carriers between pair dots, which can be investigated by either continuous-wave (cw) or transient PL. While both approaches can give valuable insight, transient studies provide clues on thc inner dynanlics and give accuratc values for the tunneling times of electrons and holes. In this paper, we report on detailed cw PL as well as time-rcsolved PL studics on carrier hmneling transfer among vertically aligned double-stacked InAsJGaAs QD layers.

11. SAMPLE CHARACTERIZATION

The AQDP samples explored here were grown using a solid-source molecular bcain epitaxy chamber coupled to an ultrahigh vacuum scanning tunneling microscope (STM). The structures consist of two InAs layers. Each sample was grown on a GaAs (001) substrate, with a 0.5 p m GaAs buffer layer and 10 min annealing at 580 " C to provide a nearly defect-free ato~nically flat surface. The seed QD layer is tbcn growl1 by dcpositing 1.8 monolayers (ML) of InAs under an As4 partial pressure of 8 X Torr at a substrate temperature of 500 "C. This is followed with either 30, 40, 50, or 60 ML of GaAs spacer (d,) deposited on top of the seed QD layer. The second QD layer is then added through deposition of 2.4 ML InAs. Each sample for optical shtdies was finally capped with a 150 ML GaAs layer. The growth rates of InAs and GaAs for all layers were 0.1 and 1.0 ML,/s, respectively.

The san~ples were structurally characterized by plan-view STM and cross-sectional transmission electron microscopy

02005 The American Physical Society

Diffraction management in 2D waveguide arrays

Aqiang Guoa, Yongan Tanga, Will Blacka and Gregory J. Salamoe, Mordechai segevb 'Department of Physics, University of Arkansas, Fayetteville, Arkansas 72701

b~hysics Department, Technion - Israel Institute of technology, Haifa 32000,Israel

ABSTRACT

We demonstrate experimentally a fixed two-dimensional (2D) periodic waveguide array by using plane-wave interference in a photorefractive (PR) ferroelectric crystal, and then report on the experimental observation of diffraction management in the 2D "fixed" periodic waveguide arrays.

Keywords: Photorefractive nonlinear optics, Waveguide arrays, Discrete dieaction.

1. INTRODUCTION

Thus far, optical waveguide arrays have provided a fertile ground for nonlinear optical applications [I-121. Waveguide arrays are a set of coupled identical waveguides. During propagation, light energy can be transferred among the waveguides through linear coupling. The broadening of light propagating in such an array can lead to discrete difbaction [I]. The diffraction can be engineered [ I ] by varying the waveguide spacing andlor the transverse wave vector (initial phase tilt). When the Bloch wave vector is nearly half-way to the edge of the Brillouin zone in the reciprocal space, dieaction is arrested. Near the edge of Brillouin zone diffraction becomes anomalous ('negative') [I].

1D waveguide array can be fabricated by using semiconductor technique, however, the fabrication of 2D periodic structure presents many technological challenges. PR ferroelectric crystals have drawn considerable attention in the nonlinear optical area and device applications because of their electro-optical properties and for the feature of ferroelectric domain reversal which substantially modifL the material's tensor properties at a desirable dimensionality and configuration. In one of the most commonly used PR ferroelectric crystals, strontium barium niobate Sr,Bal-,NbzO6 (SBN:x), SBN:75 has the smallest ferroelectric domain size [I 51. Therefore, either the PR space-charge field or an applied external electric filed can modify the orientation of ferroeletric domains. Transforming a "real-time" screening soliton into one or multiple permanent or "fixed" waveguides by means of ferroelectric domain reversal in SBN:75 has already been demonstrated [13].

In what follows we will demonstrate the formation of 2D optical lattices. Also inspired by the diffraction management in 1D arrays [I], we demonstrate experimentally that similar diffraction management also exists in 2D waveguide arrays.

2. EXPERIMENTS AND RESULTS

2.1. Forming 2D waveguide arrays in PR ferroelectric crystal The "fixed" 2D waveguide arrays are formed by combining our earlier work [13] for fixing oneltwo waveguides with the so-called holograghic lithography technique [14]. In our experimental, three Argon laser plane waves with ordinary-polarization at 514nm interfere to form an optical lattice in a Icm cube, SBN:75 crystal. The three interference beam's wave vectors are chosen such that the differences ki-kj (i=l, j=2,3), which define the reciprocal lattice, give the 2D desired lattice. Two typical geometry of the three beam's wave vectors is arranged as shown in Fig. l(a) and (c) which gives a triangular 2D lattice shown in Fig. I(b) and square 2D lattice shown in Fig. I(d). This optical lattice, under a 3 KVIcm electric field, leads to a strong space charge field compared with coercive field in the crystal. It turned out the domain structure of the crystal is modified and a "fixed" 2D waveguide array pattern formed.

Nonlinear Optical Phenomena and Applications, edited by Qihuang Gong, Yiping Cui, Roger A. Lessard, Proceedings of SPIE Vol. 5646 (SPIE, Bellingham, WA, 2005) ,0277-786)(1051$15 . dol: 10.11 17112.575567

PHYSICAL REVIEW B 71, 16531 5 (2005)

Surface dynamics during molecular-beam epitaxy of (In,Ga)As on GaAs(331)B: Formation of quantum wires with low In content

Zh. M. Wang,* Sh. Seydmohamadi, V. R. Yazdanpanah, and G J. Salalno Department ojPhysics, Univer-sit?, o f Arkansas. Fnyetteville, Arkarlsns 72701, US.4

(Rcccivcd 24 Novcmber 2004; published 18 April 2005)

Ino 2Gao,8As molecular-bcam hctcrocpitaxial growth on GaAs(331)B was invcstigatcd by scanning tunncling microscopy. While GaAs(33 l)B homoepitaxial growth always leads to a faceted ridgelike surface, ln,l,2Cia,l,8As growth over this surfacc can lcad to cithcr wirclikc cormgations or a flat surfacc dcpending on growth parameters. The transition between the phases of wirelike corrugations and a flat surface is reversible, indicating that both phascs arc thermodyna~nic favored at diffcrcnt temperatures. Based on this obscrvation, we also demonstrate a novel approach for the fabrication of (ln,Ga)As quantum wires in the GaAs matrix with low In contcnts. The carricrs are confined to onc-dirncnsional quantum wircs in thc Ino,2Gao,aAs laycr boundcd by a lower corrugated Ino,2Ga,l,sAs-on-CTaAs interface and an upper flat t i a A s - o n - l ~ ~ ~ , ~ G a ~ ~ , ~ A s intcrface.

DOI: 10.1103/PhysRevB.71.1653 15 PACS numberis): 81.15.Hi, 68.35.Bs, 81.16.---c. 71.55.Eq

To date, strain-driven self-organization has been extensively studied in order to produce coherent semiconductor nanostructures over large area^.'^ In principle quanhlm dots (QDs) and quantum wires (QWRs) are each achievable by this strain-relief nlechanisn~. An interesting question, however, for self-asscmbled growth, has been whether a QD or a QWR will fornl. For example, strain driven islanding of mismatched (In,Ga)As on GaAs(100) produces only QDs. Thc controlled fabrication of QWRs in a GaAs matrix remains a challenge. Several approaches have bcen pursued, such as the use of a lithographically patterned V groove, ridge, or sidewall substrate for g ro~th5-7 or by the use of stcp- bunching on a high index or vicinal low index surface.7-" Anlong these approaches, self-organization based on the instability of a high index surface is a natural and simple process that shows promise for fabrication of QWRs with high uniformity and high density. Utilizing a phase change of the sul.facc co~~ugat ion , GaAslAIAs(3 I l ) A QWRs arc naturally tor~ned at GaAs-thicker regions;' as indicated in Fig. ltal However, the surface corrugation does not exist at the initial stages of (In,Ga)As strained growth,13 although the amplitude of the com~gation is still controversially d i s c u s ~ e d . ' ~ , ~ ~ (In,Ga)As QWRs are indeed realized on GaA~(221)A,~ as schen~atically illustrated in Fig. l(b). The corrugated GaAs- on-(ln,Ga)As upper interface is thernlodynamic~~lly favored, while the flat (In,Ga)As-on-GaAs lower interface is kinetic limited.8,'6 In order to obtain the surfacc corrugation, the (In,Ga)As(221)A layer needs to be grown at a rather high substrate temperature ( a 5 7 0 OC). Such a high ternpcrature introduces significant In-desorption, which makes it very difficult to control the In content. In this letter, we demonstrate a novel approach to fabricate (In,Ga)As QWRs in a GaAs matrix by n~olecular-bean1 epitaxy (MBE) as illustrated in Fig. l(c). While GaAs(331)B is unstable to the fonnation of a faceted ridgelike surface, Ino,zG%,8As growth over this surface can lead to a smooth surface at the substrate temperature of 540 "C, the highest temperature at which In desorption is negligible.

The present experiments were perfo~med in an MBE growth chamber equipped with reflection high-energy elec-

FIG. I . Schematic illustration of the C)WR struchlrus on GaAs high index surfaces. (a) QWRs with a phase change of interface corrugations as proposed in Ref. 12. (b) QWKs with a top intcrface co~lugation as investigated in Ref. 8. (c) QWRs with a lower interface corrugation as dcmonstratcd in this work.

tron diffraction (RHEED) and connected to a scanning tunneling microscopy (STM) chamber via an ultrahigh vacuum transfer module. Epiready n-type GaAs(331)B substrates were t ransfe~~ed into the growth chamber where the oxide was thermally desorbed at 580 OC under As flux from a valve controlled source. After f u ~ ~ h e r annealing at 610 "C for 10 min, a 0.5-pm-thick GaAs buffer layer was grown directly at this temperahue with an As beam equivalent pressure of 1 X 10 Torr. The resulting surface was quenched by decreasing the substrate temperature and As background pressure in

1098-0 12 112005171(16)11653 15~4)1$23.00 165315-1 (02005 The American Physical Society

HTrkIL HBSTRHCT LlllKS

APPLIED PHYSlCS LETTERS 86, 132904 (2005)

Phase diagrams of epitaxial BaTi03 ultrathin films from first principles Bo-Kuai ~ a i , ~ ' lgor A. Kornev, L. Bellaiche, and G. J. Salamo Pli.ysics Depnrhnenl, Universiq of Arkansas, Fn.vetteville, Arkansas 72701

(Received 12 November 2004; accepted 5 Febmary 2005; published online 22 March 2005)

Using a first-principles-based scheme, we dete~mine the qualitative and quantitative effects of surfacelinterface, thickness and electrical boundary conditions on the temperature-misfit strain phase diagrams of epitaxial (001) BaTi03 ultrathin films. The microscopic reasons leading to such ei-fects are also revealed. 6 2005 American bzstitute of 'Pl~~,sics. [DOI: 10.1063il. 18904801

The potential of ferroelectric thin films for applications such as dynamic random access memories,' nonvolatile random access m e m ~ r i e s , ~ and integrated deviccs3 has attracted a lot of research attention. Interestingly, when going from bulk to thin films, many factors (c.g., strain, thickness, elec- trode, surface termination, interface roughness, and charge transfer at the free surface and interface)"" can dramatically affect material properties. The experimental evaluation of the effect of each factor on physical properties is difficult because the properties of real thin films are a combined result of these factors. As a result, theoretical study is important since it can untangle these factors and provide fi~ndamental insights into thc behavior of thin films.

For examplc, Pertsev et a/. ' I predicted the tcmperah~re- "misfit strain" phase diagrams for epitaxial and single- donlain (001) PbTi03 and BaTi03 thin filnls using a phe- ~loinenological method. Five crystallographic phases were obtained: the p phase (Px= Py= Pz=O) at high temperatures; the c phase (Px=Py=O, Pz + 0) at high con~pressive strains: the a a phase (Px=Py+ O,Pz=O) at high tensile strains; and the uc (Px= PZ # 0 , Py=O) and r phases (P.x= Py + 0 , Pz + 0) at low temperatures and low strains. Here, Px, Py and P z are the Cartesian conlponents of the spontaneous polarization along the [loo], [OI 01, and [OOl] pseudocubic directions, respectively. The same method was also used to study periodic two-domain stmchlres, but with a different set of paranietcrs.12 Recently, Dieguez et a1.I3 pointed out that the different sets of parameters used in Refs. 11 and 12 actually yield different low temperature phase behaviors, and thus decided to use ab initio approaches to avoid such depen- dency on experimentally deduced parameters. Their resulting temperature-misfit strain phase diagram for epitaxial (001) BaTi0, films n~ostly differs from the one of Ref. 11 by two features: the ac phasc is now absent. and the phase diagram is symmetric with respect to zero misfit strain.

These two pioneering studies of Refs. 1 1 and 13 have led to a bctter understanding of epitaxial thin films. Howcvcr, their use of periodic bulks, with mechailical constraints, to minlic epitaxial films has prevented them from investigating the effcct of two important factors on physical properties of thin films: namely the cxistence of a surfhc:e/inte~face and the film tliickness. Similarly, both works assume ideal short- circuit (SC) electrical conditions (that is, a vanishing total internal electric field), whereas real thin films likely exhibit a nonvanishing depolarizing field-even when sandwiched between metallic e~ectrodes. '~ The aim of this letter is to reveal

"Electronic mail: [email protected]

the role of the surfacelinterface. thickness, and electr~cal boundary conditions on the temperature-nl~sfit strain phase diagram of (001) epitaxial BaTiO, ultrathin films, by using a first-principles-derived method.

Similar to the approach used in Refs. 11 and 13, our BaTiO, thin filn~s are simulated to be grown along the (001) direction (z axis) and assumed to be Ba-0 terminated at all surfaces/interfaces. They are modeled by (large and z-elongated) 10 X 10 X 40 periodic supercells that contain a few number of (001) B layers (to be denoted by n?) sandwiched by nonpolar systems (representing, e.g., air, nonfer- roelectric substrates, metallic electrodes, etc.. . .). Here, we typically choose m = 5 or 7, implying that these nonpolar regions are quite large. As a result, we obtain well-converged results for the film's properties, and the film thickness is automatically accounted for in our simulations. Following Ref. 8, the total energy of such a supercell is used in Monte Carlo siillulatio~ls and is written as

where EHCn is the (effective Hamiltonian) intrinsic energy of the ferroelectric film. Its expression and first-principles- derived parameters are those given in Refs. 14 and 15, respectively, for bulk B~T~o~. '"u;} are the local fe~~oelectr ic distortions from paraelectric symmetry in unit cells i of the film-which are directly proportional to the electrical polarization and whose conlponents along the z axis are denoted as ~ i ~ , ~ . {v;) are the inhomogeneous strain-related variables inside thcsc films, whercas q is thc homogcncous strain tensor. Epituxiul (001) films are associated with the freezing of three of the six q components (in Voigt notation), LC., v6 =O and q1 = q2= 8, with S being the value forcing the film to adopt an in-plane lattice constant cqual to the onc of the chosen in Ref. 13. {u,} and {v,} are forced to vanish outside the filnls whcreas the sccond tcrin of Eq. (1) ~ l~ i iu ics the effects on properties of partial or full screening of polarization-induced charges at surfaces. It is directly proportional to a p parameter that characterizes the strength of the Ed total electric field insidc the film. P=O corresponds to ideal open-circuit conditions with Ed adopting its rnaxinl~~m magnitudc, whereas an incrcasc in 0 lowcrs this magnitude. The value of P corresponding to ideal SC conditions (to be denotcd by Psc) is 0.69 and 0.732 for thin films with rrr-7 and 5, respectively, for 10 X 10 X 40 periodic s ~ ~ e r c e l l s . ~ This second tenn is also dependent on the Born effective charge (Z'), the lattice constant ( a ) , the electronic dielectric constant of BaTiO? (E,) , as well as the average of the z

0003-6951/2005/86(13~/132904/3/$22.50 86, 132904-1 O 2005 American Institute of Physics Downloaded 09 ~ a r ' 2 0 0 8 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapl/copyright.jsp

APPLILD PHYSlCS LETTERS 86. 143106 (2005)

Selective etching of lnGaAslGaAs(100) multilayers of quantum-dot chains Zh. M. an^,^) L. Zhang, K. Holmes, and G. J. Salamo Departmerrt o f Pl~~vsics. U~zrversig ofArkorrsns, Fayetteville. Arkansas 72701

(Received 30 November 2004; accepted 21 Febniary 2005; published online 29 March 2005)

We report selective chemical etching as a promising procedure to study the buried quantum dots in multiple InGaAsIGaAs layers. The dot layer-by-dot layer etching is demonstrated using a mixed solution of NH40H:H2O2:H20. Regular plan-view atomic force microscopy reveals that all of the exposed InGaAs layers have a chain-like lateral ordering despite the potential of significant 111-Ga intermixing during capping. The vertical self-coirelation of quantum dots in the chains is observed. 0 2005 Amerirar~ Institute of Physics. [DOI: 10.1063/1.1898425]

Selective chemical etching is an important step in processing compound semiconductor heterostructures for both research and applications in which one material is etched rapidly while the other is etched very slowly or not etched at all. As one example, strained InGaAsIGaAs nanotubes' can be created by preferential AlAs etching over GaAs and InGaAs with a HF solution. The etching rate of GaAs is higher than InGaAs in a solution of mixed NH40H:H202: H 2 0 . Bascd on this selectivity, InAs fiee-standing mcm- brancs and nanoscale cantilcvcrs wcrc processed for their potential applications in micro1nanoelectroincchanical systeins.'.' In this letter, we use the selective etching capability of NH40H:H202:H20 on an InGaAsIGaAs multilayered structure, one-period-at-a-time, to examine the nano- strucuture evolution during the growth.

Due to the strain-field interaction in InGaAsIGaAs multilayered structures, vertical self-correlation4p5 and lateral se~f -order ing~-~ of quantum dots (QDs) have been demonstrated. Recently, self-assembled QD chains of sevcral mi- crometers in length can be achicved by m u l t i ~ a ~ c r i n g , ~ providing the possibility to engineer some interesting structures to study carrier interaction among Q D S . ~ " ~

Growth of long QD chains has been performed at relatively high temperatures, often with growth interruptions during GaAs capping.7."" Under these conditions, In-Ga intelmixing is expected to be significant. The anisotropy of the In-Ca intermixing during GaAs cappirig is one possibility to account for thc QD alignnient." It has bccn shown in scveral works that InGaAs QDs capped with GaAs lcavc behind a nlound structure that is oriented along the [Ol-11 d i r e ~ t i o n . ' ~ , ' ~ An overlap of such neighboring mound struc- - - tures would yield In-rich wire structures, consistent with the observed optical behavior of one-dimensional s t r ~ c t u r e s . ' ~ ~ ~ In another direction, it has been suggested that buried InGaAs QDs would elongate and connect into quantum wires during annealingl%r perhaps even begin, at the first dot laycr, as quantum wires. As a result of thc divcrsity of opinion, it would be interesting to structurally study the buried InGaAs QDs, dot layer-by-dot layer. The widely employed instrunlent to investigate the lateral ordering of QD chains is the plan-view atomic force ~nicroscopy (AFM) which normally cannot access the buried QDs. Cross- sectional transmission electron microscopy (TEM) and cross-sectional scanning tunneling microscopy (STM) are

able to characterize buried structures but their drawback is the inability to provide the needed clear p~cture that includes latcral ordering. In this Ictter, wc utillzc sclcctivc ctchlng together with the capability of thc plan-vicw AFM to providc a more complete image of the buried QDs. With this approach, the buried QD chains are exposed, dot layer-by-dot layer, displaying both their vertical self-alignment and lateral self-ordering.

The sample used for this investigation was grown in a solid-source inolecular beam epltaxy (MBE) on selnl- insulating GaAs(100) After a half-mlcron thlck GaAs bufer laycr grown at 580 "C, thc substratc tcmpcraturc was low- crcd to 540 "C for the dcposition of thc InGaAslGaAs multilayered structure. The grown structure is depicted schematically in Fig. 1 and consists of 16 periods of 5.2 non no layer (ML) Tno,55Gao.4sAs and 50.0 ML GaAs and one layer of 2.0 ML InAs capped with 8.0 ML GaAs. Due to In-Ga intermixing, both InAs and Ino,5sGrq,45As can be written as InGaAs in general. InAs and Ino,55Gao,45As arc uscd when it is necessary to make the layer subsequence clear.

Figure 2(a) shows the resulting surface morphology which is characterized with wire-like cormnations. The wires - are running along [Ol-11 with little but observable inotlula- tion in height within the wires. The question investigated here is what do the buried IilGaAs QDs look like? Do they transform into wires as they are buried or are they InGaAs QD chains as they were before being buried?

The AFM image in Fig. 2(b) illustrates the surface structural evolution after one sccond etching with NH,0H:H202: H 2 0 (1: 1: 18). The specific mixing concentration is chosen to

After 8.0 MI, A q ~ ~ a n t u m - d o t chains

CiaAs(1CX)) substrate

FIG. 1. Schematic of the grown layer sequence. The surface morpholog~ss of the InAs QD layer and after 8.0 ML CraAs capping are illusrrated. ,

0003-6951 12005186(14~43106131$22.50 86, 143106-1 0 2005 American Institute of Physics Downloaded 09 Mar ZOO8 te 130.184.237.6. Redistribution subject t e AIP license or copyright; see http:llapl.aip.orglaplleopyright.~sp

APPLIEI) PHYSICS LETTERS 86, 1 13 106 (2005)

Evolution of elongated (ln,Ga)As-GaAs(l00) islands with low indium content

S. 0. Cho, Zh. M. an^,^' and G. J. Salamo Phj~sics Departtnenr. University of ilrkansas, Fayeneville. Arktrnsas 72701

(Received 2 August 2004; accepted 26 January 2005; published online 8 March 2005)

Nucleation and growth of (In,Ga)As-GaAs(100) islands with low In content by molecular-beail1 epitaxy is investigated by scanning tunneling microscopy. The islands tend to nucleate at upper convex edges of surface steps clue to elastic strain relaxation. They are elongated along [Ol-I] with a flat top (100) facct. The growth of thc islands, nlainly through uphill transport of thc (1n.Ga)As material, is characterized by shrinking of the top (1 00) facet but the ratio of island elongation keeps constant. Q 2005 American Institute of Physics. [DOl: 10.1063/1.1883709]

The study of strained (In,Ga)As layers for the fabrication of GaAs-based quanhlm structures remains a challenge. When the indium content is lower than 20%, the growth of (In,Ga)As on GaAs(100) proceeds in a layer-by-layer fash- ion, which is generally referred to as two-dimensional (2D) growth. However, when thc In contcnt is largcr than 20%, a transition from the initial 2D growth to three-dimensional (3D) growth is observed to take place at a critical (In,Ga)As thickness. The critical thickness depends on several growth parameters including the In content. The 3D islands, formed on top of a wetting layer and recognized as self-assernbled quantum dots (QDs), have been studied extensively over the last d e ~ a d e . ' . ~ Despite this intense effort, the nature of the 7D + 3D transition is still of interest and even the source of soinc debatc. One inodcl, proposes 2D platclcts at equilibrium with one inonolaycr (ML) height as precursors for the formation of 3D i s l a n d s . ~ c c o r d i n g to this mechanism, when the 2D platelets reach a critical thickness, the 2D -3D transition occurs spontaneously and con~pletely without additional growth involved. The inodel is compelling but, to our knowledge, there is no real-time, real-space monitor- ing of the growth of (In,Ga)As islands on GaAs(100) to test this idea (although there have been similar investigations for Ge-Si(100) islands"). Howcver, a qucnch-and-look study, using an ultrahigh vacuum (UHV) scanning tunileling microscopy (STM), does reveal the existence of "quasi-3D islands" (2-4 ML high) which are intermediate during the 2D - 3D transition and act as precursors for the fonnation of 3D InAs i s ~ a n d s . ~

For the Ids -GaAs (LOO) system, the lattice mismatch is about 7% and the 2D + 3D transition is often quoted as taking place at the critical thickness of 1.6 MLs, which includes a 1 M L InAs wetting layer. In fact, the quench-and-look approach reveals that a "quasi-3D islands" appears bcforc thc colnpletion of the deposition of the first InAs M L . ~ This observation suggests the possibility that perhaps the step structure on the bare GaAs (100) surface may well play an important role in determining the 2D+ 3D transition.

We have investigated the 2D+3D transition of an (In,Ga)As alloy with low In content (0.25) by lnolecular beam epitaxy (MBE). In this case, the 1D-3D transition occurs after the deposition of 12 MLs of Ino,25Gao,75As due

to the much smaller lattice mismatch of only 1.8%. In this case, the wetting layer is morc than 10 ML, which is morc than enough to buffer the potential influence of the GaAs surface structure on thc 2D+3D transition. As a result, wc have investigated the 2D - 3D transition on (In,Ga)As providing a different perspcctive of the role of the supporting surface. We observe that the 2D+ 3D transition starts from thc nucleation of 2D islands at thc uppcr convcx stcp cdgcs. With continued deposition, the islands grow in height while the top flat facet of the QD continually shrinks.

Thc experiment was perforined in an MBE-STM UHV combined system. After deoxidization of the GaAs (100) substrate, a 0.5 pnl GaAs buffer was grown at 600 O C to achicvc a smooth starting basc. Thc substrate tcmpcraturc was lowered to 540 "C for Ino,,5Gao.ssAs growth. The growth was monitored by in situ reflection high-energy electron diffraction (RHEED). The RHEED patterns transfoinled from streaky to spotty at 12 MLs deposition, indicating the occurrence of the 2D+ 3D transition. After that the sample was cooled down immediately to quench the surface morphology. The sainple was then transfcrrcd to thc STM chain- ber through an UHV module. Constant current STM images were obtained using a tunneling current of 0.1 nA and a sample bias of -3.0 V.

Thc fonnation encrgy of surface steps on CiaAs (100) is small6 and generally MBE growth on it leads to ML steps with ragged edges as illustrated in Fig. 1. Due to the bonding energy at steps, under conditions of 2D growth, surface ada- toms tend to incorporate into the lower step edge, especially the lower concave region [Fig. I (a)]. The upper convex region [Fig. 1 (b)] is the least favorite of surface adatoins under a 2D growth mode. However, Fig. l(b) is thc most favorcd site for strain relaxation. Whenever strain-drive 7,D-k3D

a ' ~ u t l ~ o r to whom call-espondence should be addressed; electronic mail: z~~iwang(Ljuark.ed~~

FIG. I . A ledged surface is schematically shown with a lower coilcave edge denoted as (a) and an upper convex edge denoted as (b).

0003-695112005186(11)111310613/$22.50 86, 113106-1 O 2005 American Institute of Physics Downloaded 09 Mar 2008 to 130,184,237.6, Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcopyright.lsp

APPLlhD PHYSICS LETTERS 86. 063 102 (2005)

lnterdot carrier transfer in asymmetric bilayer InAsIGaAs quantum dot structures

Yu. I. ~ a z u r , ~ ' Zh. M. Wang, G. G. ~ a r a s o v , ~ ' Min Xiao, and G. J . Salarno Department of Physrcs. Universiv of Arkansas, Fuyetteville, Arkrm.sns 72701

J. W. Tornrn and V. Talalaev Max-Born-Institrit fur Nichtlineare Optik irnd Kz~rzzeitspektroskopie, A4ax-Born-Strasse 2,4, 12489 Berlin, Gertnrzny

(Received 17 September 2004; accepted 7 December 2004; published online 1 February 2005)

Transient photolumi~~escence froin a series of asyminetric lnAs quantum-dot bilayers with a CiaAs barrier layer thickness varying from 30 to 60 monolayers between the quantum-dot planes is investigated. The interdot carrier transfer process is analyzed. In the kamework of a three-level system, interdot carrier transfer times between 200 and 2500 ps are derived and compared with similar data from the literature. Within the sen~iclassical Wentzel-Krainers-Brillouin approximation, the observed "transfer time-barrier thickness-relation" supports nonresonant tunneling as the microscopic cairier transfer mechanism. 0 2005 Anzerican Institute o f Physics. [DOI: 10.1063/1.1861980]

Seiniconductor quantum dots (QDs) are a promising material for various optoelectronic applications.' For many practical purposes, such as semiconductor lasers and infrared detectors, three-dimensional QD arrays are needed in order to achieve significant interaction between the optical field and carriers confined within the QDS.' As a result, both lateral transport of nonequilibrium cairicrs within one laycr of QDs as well as vertical transport between QD layers is expected to play a significant role in device perfo~mance. In prcvious studics, we llavc addrcsscd late?-a1 carricr transfcr, i.e., the transport of carriers within one QD plane. For this purpose we studied QD structures with a bimodal size distribution and rnonitorcd their transient population, in particular the population of larger-sized dots by carriers from smaller- sized QDs, by means of time-resolved photoluininescence ( P L ) . ~ . ~

The present study is devoted to the vertical transfer of carriers, i.e., the transport of carriers between two neighboring QD layers along the growth direction. That is, in analogy to our previous work on lateral transport, we report on the transient population of a structure consisting of two QD layers separated by a spacer layer, similar to the structure used in Refs. 5 and 6. By using different InAs deposition rates, growth temperatures, or annealing times for the specific layers, vertically stacked QD layers with differently sized dots, but uniform size distribution in each layer, have been

In this case, the spacer thickness between the QD layers represents a very important parameter. In this letter, the population of larger-sized dots (LQD) affected by carriers from smaller-sized dots (SQD) is monitored and transient PL is used as the main tool for the visualization of this inter-QD carrier transfer. In particular, the investigation of a series of asymnletric QD bilayers with barriers of different thicknesses separating the QD layers allows the iden-

"Electronic mall: [email protected] h'(.)ll leave from Institute of Semlconductor Physics. National Acad. of Sci.

of Ukraine, prospect Nauki 45, 03028 Kiev. Uhaine.

tification of the nature of the interdot carrier transfer process as nonresonant tunneling.

Thc samples studicd hcrc are grown using a solid-sourcc molecular beam epitaxy system couplcd to an ultrahigh vacuum scanning tunneling microscopc (STM). Thc s tn~c- tures consist of two InAs layers grown on a GaAs(001) substrate, followed by a 0.5 p m GaAs buffer layer. 10 inin of annealing at 580 O C provides a nearly defect-free aton~ically flat surface. The seed, or first QD layer, is fabricated by depositing 1.8 monolayers (ML) of InAs with a growth rate of 0.1 ML/s under an As4 partial pressure of 8 X 10-"orr at a substrate temperature of 500 OC:. GaAs spacer layer (d,,) of 30, 40, 50, or 60 ML was depositcd 011 top of tllc sced QD layer for cach sample, respectivcly. Thc second QD layer is then added by deposition of 2.4 ML InAs followed by a 150 ML GaAs cap layer. Structural characterizations are accomplished by plan-view STM and cross-sectional transmission electron microscopy (XTEM). In the seed layer we find a size distribution of ( 4 1 1.5) nm for the height, (2013) nni for the width, and a dot density of about 4.5 X 10" cnlC'. The QD density in the second layer changes over the range of (2.5-4) X 10" cin dcpcnding on thc value of d,,. Thc QDs in thc sccond laycr are found to havc nearly twice thc volume of the seed QDs (for dsp=30 ML). This is due to additional deposition as well as to the influence of the strain field from the seed ~ a ~ e r . ~ . ~ ~ ' ' ' Thus, we obtain two vertically correlated QD layers with different sized dots in each layer. Such a system has been called an "asymmetric QD pair" (AQDP) in analogy with extensive work on an a.v,vmmetric double quantum-well (ADQW) system.' Naturally, the number of QDs in the seed layer participating in the creation of AQDPs must be strongly dependent on the dsp. Our XTEM statistical analysis proves this dcpendence of the AQDP fraction (the correlation degree a) on thc d,, value: ~ 0 . 9 5 , 0.70, 0.50, and 0.10 for the dsp=30, 40, 50, and 60 ML, rcspcc- tively. A single layer 1.8 ML QD sample is fabricated as well and analyzed as a reference.

0003-6951 12005/86(6)1063102/3/$22.50 86, 063102-1 O 2005 American Institute of Physics Downloaded 99 Mar 2998 to '13Q8184.237.6, Redistribution subject to AIP license or copyright; see http~Napl~a~v~orglavIlcopyr~ght.~sp


Growth and characterization of InAs epitaxial layer on GaAs(ll1)B

H. Wcn,* Zh. M. Wang, J. L. Shultz, B. L. Liang, and G. J. Salanio Depurtmerzt of physic.^, Univer.sity of Arkansus. E'ayetteville. Ar1cnti.s~~ 72701, IJS'4

(Received 7 May 2004; ~.evised manuscript received 24 August 2004; published 5 November 2004)

The behavior of InAs deposition on GaAs(l l l)B substrates and the correspondi~ig routes toward strain relaxation havc been invcstigated. InAs growth was for depositions ranging from 2 monolaycrs to 30 monolayers. Over this deposition range, different routes for strain relaxation caused by the lattice ~nismatch werc obscrvcd. Thc strain rclaxcd through ragged stcp edgc fonnation and Ga-In intcnnixing for low InAs dcposition and through the I'ormation of step bunching and dislocations for thicker depositions.

DOI: 10.1 103lPhysRevB.70.205307 PACS number(s): 81.15.Hi, 68.35.Bs, 71.55.Eq, 81.16.-c

- - I. INTKODUCrrION kept as (419 X i191R23.4" reconstruction in ordcr to sup-

Many investigations have shown that InAs growth on GaAs(100) leads to beautiful three-dimensional (3D) islands under noinial growth condition^.'-^ More recently, similar nanostl-uctures have been obseived through InAs deposition on GaAs high-index surface^.^-^ In the early 1990s however. it was found that InAs epitaxial growth on GaAs(ll0) and GaAs(1ll)A could grow as a two-dimensional (2D) ~ u r f a c e . ~ . ~ These results attracted interest in non-(100) surfaces for cpitaxial growth. This was especially true for GaAs(1 I I )A substrates because of a potentially useful strain- induced piezoelectric field on [ l I 11 and an incrcase in the optical inatrix elenlent that arises from a heavier hole mass.1° Although CiaAs(l1 l )B has the same piezoelectric properties as GaAs(l1 l)A, only a few studies of InAs growth behavior on GaAs(l1 l )B have been reported and the ~.esults may be viewed as still controversial. For example, both 3D island formation and smooth 2D growth have been reported for InAs deposition on GaAs(l1 l ) ~ . " - ' ~

In this work, InAs films with different thickness were deposited on GaAs(l1 l)B vicinal substrates (cut at 2" towards [2-1-11) by molecular beam epitaxy (MBE). Reflection high- energy electron diffraction (RHEED) pattern observation, scanning tunneling nlicroscopy (STM) images, and photoluminescence (PL) nieasureinents were used to study the deposition behavior. Different stain relaxation routes and the corresponding surface morphology at different deposition stages were explored.

11. EXPEKIMEN'T

The experiments reported here were perfoimed in a combined MBE-STM ultrahigh vacuum (UHV) system, which is equipped with an optical systein that monitors the substrate band edge to give accurate growth temperatures. MBE growth under various growth conditions was monitored in situ by RHEED. Growths were carried out on GaAs(l1 l)B substrates cut at 2" towards [2-1-11, Before starting the growth, the native oxide of the GaAs surface was desorbed at 580 " C under 3.7X 10-"err As4 flux. This was followed by deposition of a 0.5 p m GaAs buffer layer at a temperature of 600 "C, a growth rate of 0.75 monolayer (ML)ls, and a VIIII flux ratio of 7.5. During growth, the RHEED pattern was

press the growth of pyrainidlike defects and achieve a slnooth GaAs ~urface. '~ After deposition of the buffcr laycr, the substrate temperature was cooled to 500 "C for the growth of a 111.4s epitaxial layer at a growth rate of 0.09 MLIs and a VIIII flux ratio of over 40. After InAs being deposited, the temperature was reduced simultaneously under As pressure. The samples were subsequently transferred through an UHV transfer chambcr into thc STM clianiber. Constant current STM images were obtained using a tunneling current of around 0.1 nA and a saniplc bias of -3.0 V.

111. KESUL'T AND DISCUSSION

The growth of a sinooth buffer layer surface on a conventional GaAs(l1 l )B substrate is difficult to obtain due to the formation of defects, such as, pyramid structures and micro- twins. It has been reported, however, that nearly atomically flat films can be produced within an extreme& narrow window of growth conditions between (\. '19X \!19)R23" reconstruction and high temperature ( I X 1) recon~truction."~'~ The restrictcd growth conditions are in fact thc likely reason for the low interest in GaAs(1 1 l)B as a substrate for epitaxial growth. On the other hand, the use of a GaAs(ll1 )B vicinal substrate with a 2" tilt toward [2-1-11 has been demonstrated to be a more favorable choice for smooth liomoepi- taxial growth over a broad range of growth condition^.^^.'^ Figure 1 presents STM images of the GaAs buffer layer surface grown on a vicinal GaAs(l l l)B substrate. The large area STM image shown in Fig. I (a) indicates that the surface is sinooth and without pyramid defccts. Most of steps on thc surface are n~onolayer steps and only a few are two or three steps bunched together. The higher rcsolution STM iinage shown on Fig. I(b) indicates the step edges are all straight.

After the buffer laycr growth, 2 ML of lnAs was dcpos- ited. Figure 2 shows the corresponding STM surface images. The large arca STM iinage shown in Fig. 2(a) indicates a perfect two-dimensional mirrorlike surface. The higher resolution image shown in Fig. 2(b) indicates the surfacc is consistent with monolayer steps toward the [2-1-11 direction. However, the edge of the steps are now all "saw-cdgc" shaped. The formation of such ragged step edges shown in Fig. 2(b), as opposed to thc relatively straight stcp cdge

1098-1312 1/2004/70(20)/205307(4)/$22.f(O 70 205307-1 02004 The American Physical Society

JOURNAL OF APPLIED PHYSICS VOLUME 96. NUMBER 11 1 DECEMBER 2004

High anisotropy of lateral alignment in multilayered (ln,Ga)AslGaAs(IOO) quantum dot structures

Zh. M. an^,^) H. Churchill, C. E. George, and G. J. Salamo Phj:rics I)epurltnenl, Ui~iversiry ofA~.k~z/~sus, Fu.yerteville, Ark(m.~us 72701

(Received 26 April 2004; accepted 21 September 2004)

A formation process for long chains of quantum dots during the molecular-beam epitaxial growth of (In,Ga)As/GaAs(lOO) multilayers is presented. The morphology evolution monitored by atomic force microscopy for a series of (In,Ga)As layers demonstrates that the highly anisotropic lateral alignment of dots is gradually developed as the result of the strain field interaction mediated by the GaAs spacer coupled with the anisotropic surface kinetics that occurs during capping the dots. The dot-chain structure, providing unique properties of its own, is demonstrated to serve as a template for the spatially controlled growth of strained quantum dots in general. 0 2004 Ainericnn Institute of Phy.~ics. [DOI: 10.1063/1.1815382]

1. INTRODUCTION contain single, 2, 7, and 12 layers of (In,Ga)As QDs, each

Strain-driven self-organization of semiconductor 1 .Y- tun1 dots (QDs) has recently attracted much attention due

to the potential advantages of QD-based electronic and optoelectronic devices. For example, low-threshold lasers and normal-incident intersubband infrared detector^'^^ have both been ciemonstrated theoretically and experimentally. In spite of this success, it remains technically challenging to achieve QD assembles with sufficient dot density and narrow size distribution for practical devices. However, recent results have shown that the growth of QD multilayers is a promising route for both increasing the dot density and improving the size ~ n i f o r m i t y . ~ - ~ ~ One of the most interesting observations for stacked QDs is the vertical self-alignment that results

with an In content of 0.5. Each QD layer was fornled by depositing 9.0 monolayer (ML) of InoSG%,,As at the substrate temperature of 540 OC. The first layer of QDs was observed to fornl after 4.0 ML of Ino,5G% ,As, whereas the following layers of QDs were each observed to form after only 3.4 ML of lno,,Gao,,As. The observatlons are based on the transition of the reflection high-energy electron diffraction (RHEED) pattern from the two-dimensional (?D) streak pattern to thc threc-dimensional spot pattcm. Exccpt for the outermost Ino,,Gan,,As QD layer exposed for atomic force microscopy (AFM) measurements, each of the other QD layers is buried by depositing 60-ML GaAs, the first 30 ML at 540 "C and the second 30 ML at 580 OC.

- from the strain field transmitted through the spacer ~ a ~ e r . ~ - ' ~ In fact, thc transmitted strain has lcd to lateral ordering and Ill. RESULTS AND DISCUSSION

improved size homogeneity, and even to three-dimensional isotropic QD crystal-likc s t r u c t u r c ~ . ~ - ~ ~ Of further interest is the observation of a highly anisotropic lateral alignment of QDs that devclops during the growth of (In,Ga)As QD multilayers on G a ~ s ( 1 0 0 ) . ' ~ - ~ ~ The lateral alignment can best be described as long QD chains running along the [Ol-11 direction. While the growth of QD chains shows the possibility to cnginccr some intcrcsting device structures, the mechanism for the high anisotropy of the lateral QD alignment remains unclear. 111 this paper, we reveal the sequence of events that occul. during the growth of (In,Ga)As QD multilayers. The physical origin of the dot-chain structures is discussed in telnls of the strain field interaction and anisotropic surface kinetics.

II. EXPERIMENT

All samples used in this investigation were grown by solid-source molecular-beam epitaxy on semi-insulating GaAs(100) substrates. A 0.5-pm GaAs buffer was grown at 580 OC, followed by 10-min annealing, to provide a nearly defect-free atomically flat surface. The first series of samples

")~lccrronic mail: [email protected]

The AFM images shown in this papcr have not bcen processed except for background subtraction. A typical AFM image of the first QD layer is given in Fig. I(a). Thc surface is almost fillly covered by closely packed Tq sGao,sAs islands with a density of about 430 pin '. They are slightly elongated with the long axis parallel to the [Ol -I] direction, a conllnon fcaturc of (1n.Ga)As islands observed on G a ~ s ( 1 0 0 ) . ~ ~ i ~ u r e I(b) shows the surface topology of the sample with two laycrs of III~,,G%.~AS QDs. As mcntioncd above, the observation of the RHEED pattern indicates that the fonnation of QDs in the second layer happens earlier than in the first layer. The strain field induced by the buried islands in the first layer directs the material in the second layer to the local minimum of lattice mismatch. which gives rise to the 0.6-ML earlier formation for the second layer of Q D S . ~ ~ Meanwhile, the surface density of Ino,,Gao,5As islands shown in Fig. I(b) actually reduces to 380 pm-'. Cor- respondingly, the islands become wider and taller than that in the first layer shown in Fig. I(a). In addition, the islands tend to lineup in the [Ol-11 direction, an observation that becomes clear after the growth of seven layers of I I I ~ , - G ~ ~ , ~ A S QDs, as shown in Fig. 1 (c). The average island size continues to grow bigger and has a surface density of 350 pm-' by the seventh layer. The most surprising characteristic of this surface, how-

0021 -8979/2004/96(11)/6908/4/$22.00 6908 O 2004 American Institute of Physics

Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:lljap.aip.orggaplcopyright.jsp

APPLIED PHYSICS LETTERS VOLUME 85. NUMBER 17 25 0C.I 013ER 2004

Anisotropic photoconductivity of InGaAs quantum dot chains measured '.

by terahertz pulse spectroscopy D. G. cookea) and F. A. Hegmann Department of Physics, University of Alberta, Edmonton, Alberta T6G 251, Canada

Yu. I . Mazur, W. Q. Ma, X. Wang, Z. M. Wang, G. J. Salamo, and M. Xiao Department of Physics, University of Arkansas, Fayetteville, Arkansas 72701

T. D. Mishima and M. 6. Johnson Deparfrnent of Physics and Astronomy, University of Oklahoma, Norman, Oklahoma 73019

(Received 14 May 2004; accepted 2 1 August 2004)

We report results of time-resolved terahertz (THz) pulse spectroscopy experiments on laterally ordered chains of self-assembled InGaAs quantum dots photoexcited with 400 nm, 100 fs laser pulses. A large anisotropy in the transient photoconductive response is observed depending on the polarization of the THz probe pulse with respect to the orientation of the dot chains. Fast (3.5-5 ps) and eficient carrier capture into the dots and one-dimensional wetting layers underneath the dot chains is observed below 90 K. At higher temperatures, thermionic emission into the two-dimensional wetting layers and barriers becomes significant and the anisotropy i n thy photoconductive signal is reduced. O 2004 American Institute of Physics. [DOI: 10.1063/1.1807959]

Carrier capture in self-assembled quantum dots (QDs) has been an extremely active area of research in recent years, in part due to the potential applications for tunable, efficient QD laser structures1 and photodetectors.2 Often, carriers are injected into the barrier or wetting layers (WLs) and must be captured by the QDs before they can radiatively recombine. These capture mechanisms are therefore intimately linked to the operating parameters of QD lasers and other QD photonic devices.

Recently, a high degree of linear ordering of QDs was achieved in an Ino,36Gq,64As/GaAs superlattice by controlling strain in the Stranski-Krastanow growth process.334 Re- searchers were able to fabricate aligned dot chains with an average length of 0.9 ,urn, as shown in Fig. l(a), with dot densities of -1.8 X 10" per layer. This structure leads to very different potential profiles parallel (11) and perpendicular (I) to the dot chains. This letter demonstrates that polarized, subpicosecond, far-infrared light pulses can be used to probe anisotropic carrier transport resulting from this ordering.

The sample was grown on a semi-insulating (Sl) GaAs [001] substrate with a 150 nm GaAs buffer layer deposited by molecular beam epitaxy at 580°C. A 15-layer structure of Ino,36Gao,64As/GaAs was then grown at 540 "C, each layer containing densely packed chains of QDs with an average diameter and height of 45 nm and 5 nm, respectively. The transmission electron microscopy (TEM) plan- view image of Fig. I(a) shows the dot chains running in the - [110] direction. TEM images also reveal the existence of a one-dimensional (ID) WL with an estimated height of 1.5-2 nm directly underneath each dot chain, all sitting on top of a -0.7 nm thick two-dimensional (2D) WL. A schematic of this structure is shown in Fig. I(b), and further details on growth and characterization of this sample can be found in Ref. 3.

a Electronic mail: [email protected]

We use time-resolved terahertz spectroscopy (TRTS) to ... probe the ultrafast carrier dynamics in this QD str~lcture as a "; function of temperature, excitation density, and direction either parallel or perpendicular to the dot chains. TRTS has been used previously to investigate transient photoconductivity (PC) and carrier dynamics with subpicoseconcl time resolution in a variety of samples including bulk an<! thin tilm

7 8 semiconductors,5~6 insulators, organic crystals, and semiconductor nanostructures such as InP nanoparti~le arrays, self-assembled InAslGaAs quantum dots.q-" and GaAsIAlGaAs multiple quantum wells.12 Specitically, the terahertz (THz) pulse used in this technique is sensitive to the product of free carrier density and carrier mobility. Once carriers become captured by traps or localized $3 states. they are no longer mobile and therefore do not contribute to attenuation of the THz pulse transmitted tiirutj: !: l i ' p ~ a m p i a . TRTS is therefore an ideal probe of carrier capturT: d:.rl,rrni\:,5

in QD systems, complementing traditional techniques such *

as photoluminescence rise time measurements. . .

While the details of the experimental setup have b d n described elsewhere,6.I3 the basic technique is 2~ follt5ws, The output from a -0.7 mJlpulse, 800 nm, I kH7. arnplitied

-..

FIG. I . (a) TEM image of QD sample surface morphology. 1'1 1 ; polar~zation .. parallel (11) and perpendicular (I) to the dot chains IS i . . ?..,.1 f i i ) Sche- matic diagram o f the sample (not to scale) showing arrangemrlli c!' Ids .):I

ID WLs. . .

0063-695112004185 17)13839/3/$22.00 Id 3839 63 2004 American Instilute of Physic* Downloaded 27 ar 2008 to 130.184.237.6. Redistributlon subject to AIP license or copyright; see hnp:llapl.aip.orglapIlcapyright.jsp .

P H Y S I C A L R E V I E W L E T T E R S wc::, e n d ~ n g VOLUME 92, NUMBER 16 23 .APRIL 2004

Experimental Observation of Discrete Modulational Instability

J. ~ e i e r , ' G. I. Stegeman,' D. N. ~hristodoulides,' Y. ~ilberber~, ' R. or an dot ti,^ H. yang4 G. S a l a n ~ o , ~ M. or el,' and J. S. itch is on^

1 School of Optics/CREOL, University of Central Florida, Orlando, Florida 32816, USA 2 Department of Physics of Complex Systems, The Weizmann Institute of Science, 76100 Rehovot, l s r ~ r l . -

31nstitut national de la recherche' scientijique, Universiti du Quibec. Varennes, Quibec J3X IS2. Canado 4 ~ h y s i c s Department, University of Arkansas, Fayetteville, Arkansas 72701. USA

' ~ e ~ a r t m e n t of Electrical and Electronic Engineering, University of Glasgow, Glasgow GI2 8QQ, Scotlan~l 6 ~ e p a r t m e n t of Electrical and Computer Engineering. University of Toronto. Toronto, Ontario M5S 3G4, Canotbr . . .

(Received 13 November 2003; published 23 April 2004)

We report the first experimental observation of modulation instability in a discrete optical nonlinear array. : .

. .

DOI: 10.1103/PhysRevLe~t92.163902 PACS numbers: 42.82.Et, 42.65.Sf. 42.65.Tg

Modulational instability (MI) by which a plane wave breaks up into filaments at high intensities is a ubiquitous process that occurs in many branches of physics. Over the years, MI has been observed in various physical settings, including hydrodynamics[l,2], plasma physics [3], nonlinear optics [4,5], and quite recently in Bose-Einstein condensates [6]. MI is the outcome of the interplay between nonlinearity and dispersiveldiffraction effects. It is a symmetry-breaking instability so that a small pertur- bation on top of a constant amplitude background expe- riences exponential growth, and this leads to beam breakup in either space or time. Since this disintegration typically occurs in the same parameter region where bright solitons are observed, MI is considered, to some extent, a precursor to soliton formation [7]. In nonlinear optics, MI has been experimentally demonstrated in both the temporal and spatial domain. In particular, temporal MI has been observed in optical fibers [5] as well as its spatial counterpart in nonlinear Kerr [8,9], quadratic [lo], and biased photorefractive [ l l ] media with both coherent and partially coherent beams.

In recent years, the behavior of nonlinear discrete systems has received considerable attention in areas such as biology [12], optics [13], solid state physics [14], and Bose-Einstein condensates [15]. The linear properties of this class of systems are strongly modified and as a result their nonlinear response is known to exhibit features that are otherwise impossible in the bulklcontinuous regime [16]. In optics [13], arrays of nearest-neighbor coupled nonlinear waveguides have provided a fertile ground where such discrete nonlinear interactions can be experimentally observed and investigated [17-191. Thus far, discrete spatial solitons (nonlinear eigenstates) have been successfully demonstrated in such arrays in both one-dimensional systems [17,20] as well as in two- dimensional geometries [21].

A fundamental process that is possible in such an array system is that of discrete MI. Modulational instability in

discrete nonlinear Schrodinger-like lattices was first predicted at the base of the Brillouin zone [13j, and, sub-

'

sequently, this result was generalized to describe this process within the entire first band [22]. The interplay between spatial and temporal effects on the de; elopment - of MI has also been investigated in detail [23]. In view of. the fact that discrete MI can occur in many other physical systems in nature, its experimental obsl..~ . d ;(,. i- nf importance.

The existence of MI depends on thc relative signs associated with diffraction and nonlinearity. ;In spatially homogeneous media where the diffraction 8 = -a2k,/ak: is always positive (normal), MI occurs o d y for self-focusing nonlinearities. However, in defoc~lsi~~g homogeneous media, beams are stable against symmetry- breaking perturbations. Because diffraction in a. discrete ID array of waveguides can be either positive or negatee . .- depending on the excitation angle, both M1 and stable propagation are possible in the same array with a noli? linearity of either sign. This is a unique feature of discre!;

. . systems. In this Letter we report the first experimental observ: . .

tion of discrete MI in any physical system. Using ,an AlGaAs waveguide array with a self- focusing Kerr lion- linearity, we found that discrete MI occurred ns long as the spatial Bloch momentum vector of the initjal exc,ic~- tion was within the normal diffractloll , _,'-r of rhe Brillouin zone. The growth rate of this instagilit) ~ a c also experimentally determined by providinp an additional weak seed wave to produce a modulation on t'he high intensity beam. On the other hand, in the onomalotls '

diffraction regime, modulational instability was found to be totally absent even at very high power le\:cls. Our observations were found to be in good agrec,ilent with ..

theoretical predictions. In discrete optical nonlinear waveguide arrays, the'

wave dynamics of the electric field amplitudr ,it wave- -: guide site n is known to obey the followi~lg discrete .

163902-1 0031-9007/04/92(16)/163902(4)$22.50 O 2004 The American Physical Society 163902-4


ecIE.cC


muRwoFCRYSTAL GROWTH

Self assembled (In,Ga)As quantum structures on GaAs (4 1 l)A

Sh. Seydmohamadi*, Zh.M. Wang, G. J. Salamo Department of Physics, University of Arkansas, Fayetteville, AR 72701, USA

Received 31 March 2004; accepted 17 May 2004

Available online 20 July 2004

Communicated by Dr. D.W. Shaw

Abstract

We demonstrate the use of high index GaAs (4 1 l)A substrates as templates for the growth of (In,Ga)As quantum wires and quantum dots by molecular beam epitaxy. Scanning tunneling microscopy is used to characterize the facets of these nanostructures. For a deposition of six monolayers of (In,Ga)As on GaAs (4 1 1)A, quantum wire structures were observed to form along the [- 1221 direction with side facets indexed to (1 1 5 2). By increasing the (In,Ga)As deposition to twelve monolayers, three-dimensional islands were observed to form above the wires bounded by similar (1 1 5 2) facets as well as steeper (1 10) and (1 1 l)A facets and a convex curved region composed of (1 00) facets. 0 2004 Elsevier B.V. All rights reserved.

PACS: 79.60.J~; 85.30.V~; 81.05.Ea; 81.15.Hi

Ke.ywords: Al. Low dimensional structures; Al. Nanostructures; Al. Surface structure; A3. Molecular beam epitaxy

1. Introduction

High index semiconductor substrates have the potent ial to a c t a s templates f o r t h e g rowth o f q u a n t u m wires (QWRs) a n d q u a n t u m d o t s (QDs). T h i s potent ial is based o n the un ique surface morphology o f high index surfaces t h a t c a n influence t h e size, shape, uniformity a n d position

*Corresponding author. Tel.: + 1-479-5757660; fax: + 1-479- 5754-580.

E-mail addresses: [email protected] (S. Seydmohamadi), [email protected] (G.J. Salamo).

of self-assembled nanostructures [ 1 4 ] . F o r example, improved uniformity o f the size distribution o f q u a n t u m dots , a s indicated by atomic force microscope ( A F M ) images, a nar row photoluminescence (PL) linewidth, a n d a high peak P L intensity, h a s recently been reported f o r InAs Q D s o n a (4 1 l )A substrate [5 ] .

I n this letter, t h e morphology of self assembled (In,Ga)As nanostructures o n a G a A s (4 1 l ) A template is reported for t h e g rowth o f confined structures by molecular beam epitaxy (MBE). Depending o n the thickness o f the deposition o f (In,Ga)As, o r t h e number o f monolayers (MLs)

0022-02481%- see front matter 0 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2004.05.069

HTML fiESTRHCT + LlnKS

APPLIED PHYSICS LETTERS VOLUME 85, NUMBER 21 22 NOVEMBER 2001

Surface ordering of (ln,Ga)As quantum dots controlled by GaAs substrate indexes

Zh. M. an^,^) Sh. Seydmohamadi, J. H. Lee, and G. J. Salamo Department (fPh)~sics, University (?/'Arkansas, Frr)'etteville, A ~ ~ ~ I I ~ S L I S 72701

(Received 7 June 2004; accepted 21 September 2004)

Self-organized surface ordering of (In,Ga)As quantum dots in a GaAs matrix was investigated using stacked multiple quantum dot layers prepared by molecular-bean1 epitaxy. While one-dimensional chain-like ordering is formed on singular and slightly misorientated GaAs(100) surfaces, we report on two-dimensional squarc-like orderiiig that appcars on GaAs(n1 l)B, whcre n is 7, 5. 4, and 3. Using a technique to control surface diffusion, the different ordering patterns are found to result from the coinpctition betwccn anisotropic surface diffusion and anisotropic elastic matrix, a similar mechanism suggested before by Soloinon [Appl. Phys. Lett. 84, 2073 (2004)l. O 2004 Americc~n Oisfitute uf Physics. [DOI: 10.1063/1.1823590]

Self-assembly of strained (In,Ga)As quantum dots (QDs) in a GaAs matrix has been extensively investigated because of its potential application in nano-optoelectronics. However, without the aid of lithographic control, achieving the self- assembled QD lateral anangements needed to organize them into novel device structures remains a challenge.'-' This challenge is complicated by the fact that many factors influ- cnce self-assembled QD formation. For cxan~ple, there are two competing processes responsible for self-organization of (In,Ga)As QD on G ~ A S ( I O O ) . ~

First, anisotropic surfacc diffusion resulting from surface reconstruction, QD shape, and surface steps plays an important role in the fonnation of chain-like QD structures along the [Ol- 11 d i r e ~ t i o n . ~ - ' ~ For examplc, the tendency for QD nucleation to take placc at step ed cs has been well docu- mented in single layers of ODs.& Likewise, during the stacking of multiple QD layers, surface diffusion on an upper layer can bc affected by thc forination of asymmetric QD shapes on the layer just below. The interplay of such mecha- nisnis is responsible for the fact that 1011 chains of QDs up to several microns can be fabricatcd.'1114 Consequcntly, thc QD chains are interesting structures that can shed light on the studies of one-dimensional ( ID) carrier interaction among QDS."

Second, the elastic anisotropic properties of the GaAs inatrix can also play an important role in the nucleation of QDs and the resulting lateral ordering. For example, the elastic properties could give rise to a prefel~ed arrangement of (Tn,Ga)As QDs on ~ a ~ s ( 1 0 0 ) . ' ~ However, the added coin- plexity of the relationship of strain and nucleation to that of diffusion points to the value of investigations of lateral ordering.

By controlling the growth parameters to reduce the anisotropy of surfacc diffusion, a weakly defined rcctangular QD lattice can be optimized as a result of balancing between the anisotropic surface processes and the anisotropic elastic illatrix properties.5 For GaAs(100), the natural surface anisotropy is mainly dete~mined by its corrugated (2 X 4) rccon- stmction with dimmer rows which run along [OI-11 as illustrated by straight lines in Fig. l . However, this anisotropy can be tuned by an intended miscut to introduce different types of steps with dit'ferent densities. As shown in Fig. 1, a

"'Electronic mail: [email protected]

miscut towards [Ol I] and [Ol- 11 rcsults in type-A steps running along [Ol- 11 and type-B steps iunning along [Ol I], r e ~ ~ e c t i v e l ~ . " ~ ' ~ As one would expect, the density of steps depends on the degree of nlisorientation. For a 2' misorientation away from GaAs(1 OO), the density of monolayer (ML) steps is 0.125 nn7I. Generally, when the misorientation is less than lo0, the resulting orientation is called a vicinal GaAs (100) surface. When the niisorientation is larger than lo0, the resulting surface is called a high index surface, such as GaAs (7 1 l )B with 11.4' of misorientation toward [01 - I]. The ML step density for GaAs (71 1)B is 0.714 nni-', comparable to thc density of up or down ML steps within thc (2 X4) reconstruction (0.625 ni~i- ' l . The miscut toward [01 -11 introduces steps perpendicular to the corrugation of the (2 X 4) reconstruction. As a result, the anisotropic diffusion induced by reconstruction can be compensated by an appropriate step density. In this letter we examine this possibility in order to control lateral ordering.

Morc specifically, our investigation is focuscd on multilayered (Tn,Ga)As QD structures that are stacked on CiaAs(100) and a set of GaAs substrates with different cle- grees of misorientation toward [Ol- I]. While QDs are lined up in chains along [Ol- 11 on singular and 2' misoricntatcd vicinal GaAs(100) surfaces, two-dimensional (2D) lateral QD ordering is present on GaAs high index surfaces with ML step densities in the range from 0.714 to 1.663 nm-I.

The samples used in this investigation were grown by molecular beam epitaxy on singular GaAs(100), and 7"

FIG. 1. Schcnlatic illustration of GaAs(100) surL1ce structures, cmphasiz~ng the evolution of surface anisotropy induced by suiface reco~istmetion and steps due to misorientation.

0003-6951 /2004/85(21)/5031/3/$22.00 5031 O 2004 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to Alp license or copyright; see http://apl.aip.orglapl/copyright.jsp

Session 1432

Microelectronics-Photonics Interdisciplinary ScienceIEngineering Graduate Program Startup - Lessons

Learned at the Five Year Point

Ken Vickers, Ron Foster, Greg Salarno University of Arkansas

Background:

The University of Arkansas defined in 1998 an experimental interdisciplinary technology graduate program in Microelectronics-Photonics (microEP). While the rnicroEP Graduate Program is an interdisciplinary degree-granting entity reporting directly to the Graduate School, its academic program elements are reviewed and approved through the normal academic processes of both the Fulbright College of Arts and Sciences and the College of Engineering. Faculty and students enter the program primarily with Physics, Chemistry, and Electrical, Chemical, and Mechanical Engineering backgrounds, but may enter from any rigorous science or engineering degree program. The first students entered the program in the fall 1998 semester, with the MS and PhD microEP degrees fully approved in July 1999 and July 2000 respectively.

The traditional research and educational focus of this program is electronically and photnnically active materials, the devices that can be made from those materials, and the high performance solid systems that can be made from the combination of materials and devices. The nop- traditional educational focus is in the management of the systems and human resources that move these technologies from the laboratory into full commercialization for the benefit of society. Specifically, the microEP graduate program strives to emulate an industrial work group in an academic environment, an environment that is based in assessing performance through evaluation of individual projects and knowledge rather than in meeting group objectives.

The microEP program also stresses the concepts of civic responsibility through the concept of the "citizen technologist". All microEP students are trained in their responsibilities to lead their communities after graduation to repay the large investment that society has placed into their graduate education. Inherent in this is the need to support the K-16 educational pipeline that will produce the next generation of their professional colleagues. It is important to lead through example, so the microEP faculty and administration have pursued resources to actively participate in all of these activities. The microEP program has received NSF IGERT, REU, RET, GK-12, and MRSEC awards; and a Department of Education FIPSE award to implement the microEP educational concepts in the traditional Physics Graduate Program.

The history of the microEP program formation, along with the details of its approach to graduate education, have been fully described in a paper presented by the authors at the 2002 ASEE Annual convention1. The program has now produced over twenty-five MS graduates and three

Proceedings of the 2004 American Society for Engineering Education Animal Conference & Exposition Copyright 02004, American Socielyfor Engineering Education

APPLIED P1IYSIC:S LETTERS VOLUME 85. NUMBER h I) AUOUST 2004

Tuning Ino.3Gao.7AslGaAs multiple quantum dots for long-wavelength infrared detectors

Ying Chao Chua, E. A. Decuir, Jr., B. S. Passmore, K. H. Sharif, and M. 0. ~ a n a s r e h ~ ) Department of Electrical Engineering, 3127 Bell Engineer-ing Center; U~liversity of Arkur1~a.r. Fuj~eIleville, Ar.kun,srrs 72701

Z. M. Wang and G. J. Salamo Deportment of Phj~sic.~. University q/'.4rkunsm. F'oyetfeville, Arkn~sus 72701

(Received 24 November 2003; accepted 7 June 2004)

Optical absorption spectra of intersubband transitions in Ino.3G%.7AslGaAs multiple quantum dots were investigated using the optical absorption as a function of the number of Ino,3Ga,l,7As inonolaycrs deposited using the inolccular-beam cpitaxy Stranski-Krastanow technique. Thc peak position energy reached 1.3.7 p m for a sa~nple containing 50 monolayers of Ino,3Gao,7As. The lack of the observation of intersubband transitions in sinall quantunl dots, where the number of the deposited monolayer is less than 15 monolayers, is an indication of the absence of quantum confinement. On the other hand, the prescnce of high dislocations density in larger quantum dots, where the deposited number of monolayers exceeds 50, could be the reason of why the intersubband transitions arc degraded. 6 2004 American Institute of Pl?ysics. [DOl: 10.1063/1.1777522]

Long-wavelength infrared detection is one of the major applications of self-assembly semiconductor qitantum dots.' This interest steins from the fact that unlike multiple quantuin well structures2 normal incident photon-electron coupling is possible due to the lack of wave vector selection rules. The limitation of matching the dominant optical phonon energies with the range of excited electron energy introduces what is called phonon bottleneck,394 which improves the excited carrier lifetime. This effect is very helpful in collecting the photoexcited electrons before they recombine which ultimately increases the photocurrent and leads to a higher photoconductive gain. Most of the results reported for long-wavelength infrared detcctors fabricated from quantum dots have been expressed in temls of photocu~rents-9 andor p h ~ t o r e s p o n s e . ' ~ ~ ~ ~ ' ' These detectors were limited in inost cases to the peak position wavelength range of 4-9 pin. This limitation is inherent to the quantum dot size itself. The lower size limit of a quantum dot is given by the condition that at least one energy level of an electron or a hole or both exists. An electron level exits in a spherical quantum dot1 if thc critical diamcter (Dmin) exceeds thc value of Dl,;, = d l J2m * AEc, where AEc is the conduction-band discon- tinuity and rn* is the electron effective mass. There also exists an upper limit for size of the quantum dot. A thermal population of high-lying energy level is undesirable for infrared detector applications. For example, thermal population fraction of higher-lying energy level of 5% will limit the operating temperature to k T S 1 /3 (El - E,), whcrc Eo and E, are the confincd energy lcvcls in the dot.' Unless ncw idcas emerge, the wa\~clength range of quanh~m dot infrared dctec- tors will remain constrained by the dot size.

In this letter, we report on the optical absorption of intersubband transitions in 1r1,.~G%,~AslGaAs inultiple quantum dots grown by n~olecular-beam epitaxy (MBE). For an In con~position of 30%, we found that the intersubband transitions are not observable in quantum dot structures with less

than 15 monolayers of Tn,3Gao,7As. The intensity of the intersubband transitions was found to dccreasc as the quantum dot size is increased, which can be explained by the introduction of dislocations that cause degradation of the quantum dots. The wavelengths associated with the peak positions of the intersubband transition spectra were found to increase as the number of the deposited In,,G%,,As inono- layers is increased. The experimental optical absorption nlea- surcincnts of the intersubband transition is compared with theoretical results.

The quanhim dot structures were grown by MBE on scini-insulating GaAs(100) substrates. After loading thc substrate in the MBE growth chamber, the substrate was heated to 600°C in As atmosphere to allow the desorption of the oxide layer. During the MBE growth, the As beam equivalent pressure was kept constant at 1 X 10 ' Torr for all samples. Subsequently, a 0.5 p m Si-doped CiaAs buffer layer, which also serves as a contact layer, was deposited at the abovc incntioncd substrate temperature. The substrate was then cooled down to 500°C and 30 periods of Ino ,3G~,7AslGaAs were grown. The growth was monitored by an in sit11 rcflcction high-cncrgy elcctron diffraction (RHEED). A 0.5 pin Si-doped GaAs contact layer was deposited at the top of the structure after increasing the substrate temperature to 580°C. A series of samples with differ- cnt Ino,3G%.7As thicknesses ranging from 15-50 monolaycrs were grown. The transition from a layer-by-layer growth mode to three-dimensional islanding occurs at 10 monolayers of l i ~ ~ , , G % . ~ A s as judged from the point at which the RHEED patterns transfoln~ed from being streaky to spotty. The Ino,3G%,7As quantum dots were doped with [Si]- 1 X 10" cm-', while the buffer and cap layers were doped with [Si]- 1 X 1018 cnC3. The thickness of thc GaAs barrier layer was 120 monolayers for all samples. The density of the quantum dots was cstiinatcd to be on the order of 10'' cm ', which increased as the number of the deposited monolayers is increascd. The structure is schcinatically shown in Fig. 1 Tt should be pointed out that the number of deposited inono- layers should count for the sum of the qiiailtum dot height

0003-6951 12004185(6)11003/3/$20.00 1003 O 2004 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcopyright.jsp


Enhancing the in-plane spatial ordering of quantum dots

W. Q. Ma,* M. L. Hussein, J . L. Shultz, and G. J. Salamo Department of Physics, Universily of Arkansrzs, F'ayetteville. Ar1can.sas 72701, iJSA

T. D. Mishima and M. B. Johnson Dej~artment of Physics & Astronomy, UniversiO~ qf Oklahornu, Norman, Oklahonla 73019, US.4

(Received 10 Novetnber 2003; revised manuscr~pt received 10 February 2004; published 25 June 2004)

Wc report on thc usc of (In,Cia)AslGaAs multilaycr stacking at elcvatcd growth ternpcraturcs to produce enhanced in-plane spatial ordering. Cross-sectional tra~ls~nission electron n~icroscopy images reveal that the (ln,Ga)As islands arc veitically corrclatcd whilc atomic forcc microscopy imagcs demonstrate latcral ordcring - of quantum dots that are closely aligned along the [Oil] direction as chains which are then~selves positioned - periodically along thc [01 I] direction. Thc in-plane spatial ordering along thc [01 I] and [01 I] dircctions is directly seen by asylninetric (311) glancing exit x-ray diffraction with the x-ray beam along the respective direction. Growth studies as a hnction of tcmpcraturc indicatcd that the obscrvcd latcral ordcring rcsults froin enhanced surfice diffusion and the vertical transfer of corresponding anisotropic strain pattern due to the anisotropy of surfacc diffusion.

D01: 10.1 103lPhysRevB.69.233312 PACS number(s): 68.65.Hb, 68.65.Ac, 68.55.--a

The fabrication of quantum dots (QD's) based on the self- asseinbly of three-dimensional (3D) islands in lattice- mismatched growth is usually characterized by broad size and shape profiles as well as a random spatial distribution. The broadened characteristic features result in a correspond- ingly inho~nogeneously broadened optical emission spectrum that seriously limits potential optoelectronic applications. Likewise, the observed randoin spatial distribution also limits potential applications, such as the fabrication of photonic crystals. 111 fact, i~nproveinents in both of these areas are needed to make it possible to realize many proposed novel devices.'

One approach to improve the uniformity and spatial ordering of QD structures is through the vertical ordering that is achieved by stacking QD layers between barrier spacer layers.'s3 In this case, if the spacer layer thickness is relatively thin, the strain mediation of buried QD's causes subsequent Q D layers to deposit on the locations of the spacer layer where the strain field induces a local minimum of lattice mismatch. With this approach, island uniformity is observed to improve.A4

To improve on lateral ordering, one approach taken is based on selective growth on patterned substrates so that the QD's can nucleate and be positioned at defined locations. Another promising approach has been to deposit QD's on a n a t ~ ~ r a l tern plat^,^,^ such as a uniform quantum wire stnlcture7 or a periodically dislocated buffer layer.%ore recently, well-defined QD structures demonstrating distinct in- plane spatial ordering have been achieved using only vertical stacking of QD's on a flat ~ u r f a c e . ~ In this paper, we clarify the underlying physics responsible for the lateral ordering observed in these layered structures.

For the investigation reported here, two types of samples were grown by solid-source moleccilar beam epitaxy (MBE) on semi-insulating GaAs (100) substrates. The growth process for the first type of sample, denoted as sample a, is as follows. After the desorption of the native oxide at 580°C

under As4 atmosphere, a 0.5-pnl-thick GaAs buffer layer was deposited also at 580°C with a growth rate of 1.3 MLIsec. The substrate was then cooled down to 540°C for the growth of a QD superlattice s t ruct~~re. For the growth of the superlattice structure, the growth rate of GaAs was reduced to 0.1 8 MLIsec. The superlattice consists of 15 periods of an I ~ , 3 8 G ~ , 6 2 A s / G a A s multilayer. The thickness of thc (In,Ga)As and GaAs layers in thc supcrlattice was 2.4 and 22 nm, respectively, which were detem~ined by x-ray diffraction (XRD) mcasuremcnts of the sy~nmetric (400) reflection after the growth. Finally, in order to see the surface morphology, a 2.4-nm-thick (In,Ga)As laycr was grown on top of the superlattice. Because the growth temperature is relatively high, following every deposition of the (1n.Ga)As laycr, thrce rnonolaycrs of GaAs wcre grown without intcr- n~ption to suppress In segregation. After 10 sec of growth interruption, the rest of the GaAs layer growth was initiated. The As4 to Ga beam equivalent pressure (BEP) ratio during thc growth was about 15. The growth procedurc of the sccond sample, denoted as sample h, was exactly the same as sample a except that the growth temperature of thc (In,Ga)As/GaAs superlattice was 480" C instead of 540" C.

Figures l(a) and l(b) show tapping-mode atomic force microscopy (AFM) top views of samples a and b, respectively. For Fig. l(a), a distinct in-plane spatial ordering along - the [OI I ] and [Ol 11 dircctions can bc sccn. It is obscrvcd that

the QD's are closely arranged along the [ O G ] direction as chains and the chains are positioned periodically along the [0 1 I ] direction. The bottom of Fig. I (a) shows a line scan of the AFM iinage along the [Oll] direction corresponding to the position at the top of Fig. l(a) marked as a white line. The AFM iinage shows that thc diameter of the QD's is about 54 nm and the periodicity along the [OI I ] direction i s about 110 nm. As a clear comparison, Fig. l (b) docs not rcvcal in-plane ordering. The size of the QD's grown at 480°C is about 30 nm, which is smaller than that grown at 540°C. This is consistent with reported r e s ~ ~ l t s . ~ ~ ' ~

01 63-182912004169(2311233312(4)1$22.50 69 233312-1 02004 The American Physical Society

HTML HESTRHCT + LlnKs

APPLIED PHYSICS LETTERS VOLUME 84, NUMBER 23 7 JL!NF 2004

Persistence of (In,Ga)As quantum-dot chains under index deviation from GaAs(1OO)

Z. M. ~ a n g , ~ ) Yu. I. Mazur, and G. J. Salamo Deparfnlent of Phyrics, Uiirversity of Arkcltisa.r, Favefte~~ille, Arkansas 72701

P. M. Lytvin, V. V. Strelchuk, and M. Ya. Valakh Lashkaryov Institute of Serniconcfzictor Physics, NAS of Ukr~line, Prospect Nauli); 45, 03028 Kviv, Lrltraine

(Received 17 February 2004; accepted 14 April 2004; published online 19 May 2004)

Utilizing the uaturally curved surface contours provided by oval defects on a GaAs(100) surface, we demonstrate that alignnlcnt of quantum-dot chains formed during the growth of (In,Ga)As multilayers is unyielding to a modest deviation of surface orientation from (100) of about 0.7" along [0 1- 11 and 8" along [Oll]. This finding suggests that the strain-driven kinetic anisotropy responsible for the fonnation of the quantum dot chains dominates over selective island formation at steps due to surface misorientation. The robustness of the quantun~ dot chain adds to its potential for its future application. O 2004 American Institute oJ'Physics. [DOI: 10.1063/1.1760219]

Quantum-dot (QD) chains are fascinating three- dimensional island polynlers that provide a unique playing field to study ca i~ ie r and optical interactions between closely spaced islands. For example, temperature-dependent polarization of photolumincsccnce spectra was obscrvcd as a rcsult of carrier transport along (In,Ga)As QD chains.' Until recently, the dominant approach to grow QD chains has rc- lied on prcpattemed substrates. This approach is coinplicated and can easily suffer in quality due to defects introduced by the ~ubstrate. ' ,~ Most recently, a sclf-organized process for the fonnation of (In,Ga)As QD chains was dcmonstrated using (In,Ga)As/GaAs QD inultilayering on GaAs(100) sub~trates.~-%ith this approach QD chain lengths well over 5 p m have been realized. Interestingly, for each of these studies, due to the tolerance associated with a purchased vicinal (100) cut ( t0 .2") , the GaAs(100) substrates utilized have been slightly misoriented from the exact (,loo)-direction. In fact, a small miscut is often planned in order to suppress morphological instabilities that arise on cxactly oriented surface^.'^^ As a result, it is natural to explore the role of the iniscut on formation of QD chain structures in order to gain insight into the growth mecha~>ism. The expectation that misorientation~ may play an important role in the formation of the dot chain formation is warranted since misorientations are oftcn becn found to enhance the lateral ordering of sclf- assembled nanost~uctures.~ I' To investigate thc influence of a misorientation on the evolution of nanostructures, wafers with specific values of the nlisorientation can be ordered from sen~iconductor n~anufacturers,~- '~ or spherically shaped substrates can of course be polished to produce a continuous spectn~m of misorient at ion^.'^^" However, a spectrum of mi- soricntations in all azimuthal directions can also be provided in a natural way using oval shaped surface defects on a GaAs surface. Although the MBE technology for the growth of nanostructure is well established, oval defects are still a regu- larly occurring phenomena, appearing with a surface density from 10' to lo5 cin-'. l 5 Oval defects are classified into a-

"Electronic mail: zmwang(?juark.edu

and P t y p e defects, with and without a core, respectively. While one may observe many types and shapes of oval defects, they all can provide an interesting range of misorientations on an othcnvisc GaAs (100) sul.facc. In this letter, we report on a study that utilizes oval defects as a tool to investigate thc role of a n~isoriented surface as a substrate for on the growth of (In,Ga)As QD chains. It is found that the align- lncilt associated with the formation of QD chains is surprisingly robust and is observed to flow over oval defects in straight lines ignoring the different surface kinctics expected as a result of the distribution of surface steps and edges induced by the inisoricntation.

The samples used in this investigation were grown by molecular beam epitaxy and consist of an (In,Ga)As/GaAs self-assembled QD multilayer grown on a 0.5 pin GaAs buffer layer on a semi-insulting GaAs (100) substrate. For all (In,Ga)As/GaAs QD samples, the growth temperature of the strained multilayers was 540°C and the GaAs spacer layer was 60 monolayers (ML). The last (In,Ga)As QD layer was deposited without a final GaAs capping for characterization of surface morphology by atomic force microscopy (AFM).

Figure 1 shows a representative AFM imagc of the surface morphology after stacking 17 QD layers of 7.6 ML Ino,4Gao,6As. For this structure, one growth intelruption ev- ely I0 s per 3 ML GaAs growth is applied for the first 18 ML GaAs spacer growth. The growth interruption at the initial stage of the GaAs spaccr growth plays a crucial role in the QD alignment.' The fast Fourier trailsfom1 of Fig, l(a) clearly reveals the periodic nature of QD lateral ordering. The QD chains are running along [Ol-I], spaced with the separation of about 100 nm. Thc separation between QDs within chains is about 50 nm. Many of the QD chains are now over 5 p m long, and as in our previous repo~.t,"rc spatially interrupted by elongated plates along LO1 -11 instead of dotted by QDs. As shown in Fig. l(b), most of the QD chains tenninate with elongated plates. In fact. for (,In,Ga)As/GaAs mutlilayers, having an In content of 0.3, the QDs are typically totally replaced by elongated plates and quantum wires are observed instead of QD chains.'"

A P t y p e oval defect 011 a GaAs sample is shown in Fig.

0003-6951/2004/84(23)/4681/3/$22.00 4681 O 2004 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcopyright.jsp

APPLIED PllYSICS LETTERS VOLUME 84, NUMBER I1

Fabrication of (In,Ga)As quantum-dot chains on GaAs(100) 2. M. Wang,a) K. Holmes, Yu. I. Mazur, and G. J. Salamo Physics Department, Universig) of' Arkansas, Fayetieville, Arkarrsrzs 72701

(Received 10 November 2003; accepted 9 January 2004)

Nanostructure evolution during the growth of multilayers of Ino,5G~.5As/GaAs (100) by molecular-bcam cpitaxy is investigated to control the formation of lines of quantum dots called quantunl-dot chains. It is found that the dot chains can be substailtially increased in length by the introduction of growth interruptions during the initial stages of growth of the GaAs spacer layer. Quantum-dot chains that are longer than 5 p m are obtained by adjusting the Ino,5Gao.5As coverage and growth interruptions. Thc growth proccdure is also used to creatc a template to forin lnAs dots into chains with a predictable dot density. The resulting dot chains offer the possibility to engineer carrier interaction among dots for novel physical phenomena and potential devices. 9 2004 Atnericnn histitute of Pliysics. [DOT: 10.1063/1.1669064]

By reducing the three dimensions of sen~iconductor stmctures to the nanoscale, quantum dots (QDs) can be fabricated and shown to possess unique optical and electronic properties that can change the paradigm of electronic and optoelectronic devices. Among the approaches for the fabrication of colnpound semiconductor QDs, strain-driven self- assenlbly has been demonstrated to be one of the best for the generation of defect-free QDs. However, due to the stochastic nature of the nucleation of self-assembled growth, control of spatial ordering of the QDs has proved to be extremely challenging.

One approach to spatial ordering that produces vertical self-alignment occurs during QD stacking, and is based on a seeding effect resulting from the strain field transmitted from the QD layer just below.'" In fact, the strain field from thc underlying QDs, interacting through the spacer layer, not only causes vertical alignment but in some material systems also leads to lateral ordering, and even to three-dimensional QD "crystal-like" However, for 111-V seinicon- ductors, such as (In,Ga)As QDs in a GaAs(100) matrix, the achievement of control of lateral ordering has dominantly relied on lithographic t e ~ h n i ~ u e s , ' ~ ~ which when coinpared to strain induced self-assembly, requires extra processing steps and can easily introduce defects into the QD structures.

A second approach to achieve lateral ordered (In,Ga)As Q D arrays in a GaAs matrix has been realized using high index substrates, such (311) and (41 I ) . ~ , ' ~ In this case, the Q D in-plane self-alignment is believed to originate from the anisotropy of the special crystallographic arrangement of high index surfaces. Even inorc recently, an unusual and perhaps unexpected approach to lateral ordering has been reported. This approach is based on the self-assembly of long lines of QDs, which are referred to as QD chains, and which form during the growth of multilayers of (In,Ga)As QDs separated by ~ a ~ s ( 1 . 0 0 ) . ~ ' - ' ~ o m ~ a r e d to single layer (In- ,Ga)As QDs, which are randomly distributed on the GaAs(l00) surface, the long chains of QDs provide an exciting playing ficld to study carrier and optical interactions between the QD structures.

"Electronic mail: zmwang(i+irk.edu

In this letter, the effect of GaAs capping and (In,Ga)As coverage on the formation of QD chains is investigated. It is found that by introducing growth interluptions during the initial stages of the growth of thc GaAs spacer laycr, the (In,Ga)As QD chains can be substantially lengthened. Long chains of QDs, over 5 pin along the [Ol- I ] direction, are achieved for particular (In,Ga)As coverage and growth interruptions.

In the experiment, epi-ready semi-insulting GaAs (1001 substrates are first degassed for 20 lnin at 350°C and then loaded directly into the growth chamber of a solid-source molccular beam epitaxy systenl. Aftcr thc native oxide layer is desorbed by annealing the substrates at 610 OC for 10 min, a GaAs buffer layer of 0.5 pnl is grown with a growth rate of 1.0 monolayer per second (MLIs) at 600 OC, under a constant As beam equivalent pressure of I X Torr. The substrate temperature is then reduced to 540°C for the growth of 16 X (9.OML Ino,sGk5As/60 ML GaAs) quantum dot multilayers. Thc growth rate of GaAs and Ino,5Gal~,5As is 0.4 and 0.8 MLIs, respectively. Finally, the last QD laycr was typically deposited without a final GaAs capping layer so that the surface morphology can be characterized by atoinic force lnicroscopy (AFM).

Figure I shows AFM images of the surface nlorpholo- gies after stacking 17 Ino,5Gao,5As QD layers. For the structure imaged in Fig. l(a), one growth interruption of 10 s is introduced after the first 3 ML GaAs spacer growth. For the structure imaged in Fig. l(b), one growtl~ interruption every 10 s per 3 ML GaAs growth is applied for the first 18 ML GaAs spacer growth. The QD density is 210/pm2 for Fig. I (a) and 205/pm2 for Fig. I (b). The difference is negligible when considering the statistical variation. The average QD size is also the same for both cases, 6 uin in height and 43 nm in diameter. Where there is a rather remarkable difference is in the length of the QD chains. The average length of the QD chains is about 1 p m for Fig. I (a) but close to 2 pni for Fig. I(b). The average chain length is about 0.7 p n ~ without a growth intermption.16 It is easy to see that the introduction of growth interruptions, during the initial stages of the growth of the GaAs spacer, significantly improves the QD alignment along [Ol- I]. Moreover, the QD alignment

0003-6951 12004184(11 )I1 931 131$22.00 1931 0 2004 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl,aip.orglapIlcopyright.jsp

APPLIED PHYSICS LETTERS VOLUMF 84, NUMBER 10 8 MARCH 2004

Atom-resolved scanning tunneling microscopy of (In,Ga)As quantum wires on GaAs(311)A

H. Wen, Z. M. ~ a n g , ~ ) and G. J. Salamo Phvsics Department, Univecritv q/drkansa.r, Fayetteville, Arkansas 72701

(Received 2 October 2003; accepted 7 January 2004)

Generally (In,Ga)As strained growth on GaAs surfaces results in zero-dimensional quantum dots. The formation of one-dinlensional quanhlnl wires is demonstrated during (In,Ga)As n~olecular-beam-epitaxial growth on GaAs(3 1 l)A at high temperature. The wires are running along the [-2331 direction. Atomically resolved scanning tunneling microscopy images reveal that the wires are triangular-shaped in cross section and the two side bonding facets are {11,5,2}. These results are discussed in ternls of a mechanism of strain-driven facet formation. O 2004 American In.rtitute ofPhj,.sics. [DOI: 10.1063/1.1664018]

The surface roughing that occurs due to strain relief during the growth ol' highly mismatched semiconductor epilayers provides a simple and efficient way to fabricate nanostructures that spatially confine carriers in two or three dimensions. These confined structures are called quantum wires (QWRs) and quantum dots (QDs), respectively. As a typical example, a strained (In,Ga)As surface grown on a GaAs substrate is observed to host the nucleation and self- assenlbly of defect-free (In,Ga)As/GaAs QDS' with all three dimensions at the tens of nanometer scale. Thc thrcc- dimensional (3D) confinement results in an atomic-like discrete density of states, which promises a very different and potentially valuable behavior when compared to traditional quantum wells (QWs) where the density of states is a staircase-like function of energy. For example, the low- threshold-current lasers and nom~al-incident intersubband infrared detectors that are predicted based on the expected sharp density of states for QDs havc already been experi- ~nentally dem~ns t ra ted . '~~ However, the performance of these devices continues to be limited due to the inhoinogeneous broadening of the Q D size distribution and the resulting broadened density of states. Interestingly. the corresponding broadening associated with QWRs is not as severe and yet they do present a coinpromising position with a density of states that is intermediate in energy dependence between QDs and QWs. As a result, it becomes attractive to explore the trade off between broadening and density of states when coilsidering the choice between wires and dots for a particular application. Along with the density of states feature, the highly anisotropic morphological nature of QWRs is also reflected in other behavior, such as, emitted photoluininescence or electronic t r a n ~ ~ o r t . ~ , ~ These features in themselves open special oppoi-tunities for the application of QWRs adding interest to their investigation.

While strain-driven fomlation of (In,Ga)As QWRs on a GaAs surface can present such interesting differences and possibilities they are apparently inore difficult to produce as evidenced by the fact that only a few investigations have been reported,"' such as, thc elongated islands that were

" ~ u t h o r to whom correspondcncc should bc addrcsscd; elcctl.onic mail: zmwang@ji;,uark.edu

observed through a shape transition from 3D islands to QWRS."" In this letter, we show that (In,Ga)As islanding on a GaAs(3 1 l)A surface, in fact, leads to beautiful QWRs. Our investigation reveals the shape and faceting of the QWRs on an aton~ic scale. These results are consistent with the fact that GaAs(31I)A has been shown to be a good template for the growth of QDs.IZ. I' For example, (In,Ga)As QDs grown on a GaAs(31 I )A surface have shown unique structural and optical properties and have been studied in some detail."-"

Thc QWR structures that arc thc subject of this report were grown by molecular-beam epitaxy on an ??-type GaAs(3 1 l)A substrate. The GaAs wafer was first covered with a buffer layer of 0.5 p m thickness grown at 600 OC. For the (Tn,Ga)As deposition, the substrate temperature was reduced to 540 O C , the highest temperature at which In desorption can be considered insignificant. The As beam equivalent pressure was kept constant at 1 X lo-' Torr for all the structures. After coinpletion of the growth, the samples were inl- mcdiately transferred to the attached scanning tunneling microscopy (STM) chamber. Constant current STM images are obtained using a t~~nneling current of 0.1 nA and a sample bias of -3.0 V.

After 8.0 monolayers (ML) of Ino,,Gao5As deposition, 3D islands were observed, as shown in Fig. I(a), with the density of 1.6X 10" ~ m - ~ . Thc 3D islands arc typically 5.6 nnl high. with lateral dimensions of 53 nnl along [-2331 and 22 nin along [OI-I]. The insert of Fig. l(a) shows a schematic drawing of thc shape of the Ino,5G%,5As islands. The 3D islands are laterally bounded by the facets of (1 1 l)A and {I 10) and a near (100) convex sidewall and are capped by (1 1,5,2) facets [same as the rcportcd shape of InAs islands on ~ a ~ s ( 3 1 1 ) ~ ] . ' ~ But the most remarkable feature of this STM image is that the background area surrounding the islands is not the commonly observed Stranski-Rrastanow two-dimensional (2D) wetting layer but rather has an obvi- ous corrugation along [-2331. Although the corrugated wetting layer is somewhat distorted by 3D islands, one can still extract the lateral separation and the height of the corrugations, which are about 36 and 3.3 nin, respectively. The corrugated side facet has an angle to the (3 1 l)A surface of 10" measured along [Ol-11. It is important to point out that the corrugated wetting layer morphology is not intrinsic to the

0003-6951 /2004/84(10)/1756/3/$22.00 1756 O 2004 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to Alp license or copyright; see http:l/apl.aip.orglapl/copyright.jsp

JOURNAL OF APPLIED PHYSICS VOLUME 95, NUMBER 3 I FEBRUARY 2003

Polarization spectroscopy of InGaAsIGaAs quantum wires grown on (331) B GaAs templates with nanoscale fluctuations

X. Y. Wang, Z. M. Wang, V. R. Yazdanpanah, G. J. Salamo, and Min xiaoa) l)epa~.l~rienl of Physics, UniversiF of Arktm.sus. E'u!lellevrlle, Arkunsu.r 72701

(Received 5 November 2003; accepted 10 November 2003)

Using (331)B GaAs tcmplatcs with nanoscaIc fluctuations, wc have fabricated InGaAsIGaAs quantum wires (QWRs) with a density of -2.0X lo6 cin-' and the degree of polarization as high as -28%. In tllc samplcs with weak latcral confincment, wc observed theinlal delocalization of carriers from the one-dimensional QWR states to the two-dimensional quantum-well states with increasing tcinpcraturc, which is almost abscnt in QWR samples with strong latcral confincment. 0 2004 American Institute qj'Phy.sicy. [DOI: 10.106311.1 6377211

Low-dimensional quantum wells (QWs), quantum wires (QWRs), and quantum dots (QDs), are important quantum systems for both the study of fundamental physics and the potential applications in optoelectronic devices. Anloilg these nanostructures, the one-dimensional (ID) QWRs are the smallest dimensional structures that can be used for efficient transport of electrons, and also as polarization-sensitive nanoscale photodetectors that may be useful in various fields.' Compared with the well-fabricated and -studied two- diine~lsional (2D) QWs and zero-dimensional (OD) QDs, thc 1D QWRs have only become the focus of intensive investigations recently. This is partly due to the reason that fabrication of ID QWRs has always been a challenge for crystal growth technology since a higher structural ordering is required. and also the underlying mechanisms for anisotropic growth along one direction are still not conlpletely undcr~ tood .~

In thc past dccadc, many methods have bccn adoptcd to fabricate sen~iconductor QWRs, in which self-organized growth is one of the most promising approaches since it can effectively avoid the processing damage and contanlination that might be introduced by some other fabrication processes, such as ultrafine lithography and chemical etching. Specifically, molecular beam epitaxy (MBE) growth of self- organizcd InGaAsIGaAs QWRs on high-index GaAs substrates has been shown to be an easy process to fabricate QWR structures with high lateral density, high uniformity, and high optical quality.3-5 Basically, in the conventional process of growing narrow InGaAsIGaAs QWS?' the use of a high-index substrate would result in a fa t , lower interface and a corrugated upper interface, which may act as lateral potentials for the carriers, hence, thc formation of 1D QWRs. Thc realization of ordercd QWR structurcs can bc attributcd to the surface reconstruction or atomic rea~~angelnent observed on high-index surfaces with a high density of multi- atomic steps or microfacets.'

In our previous work,' we observed that GaAs (331),4 and (331)B surfaces are both faceted on a nanometer scale, containing (1 10) and (1 11) facets, and it was further pro-

"~lectronic mall: ~lu;[email protected]

0021 -8979/2004195(3)11609/31$22.00

posed that the resulting highly anisotropic ridge-like surfaces could be used in the fabrication of high-quality and high- density QWR structures. Here, we report the MBE growth of InGaAsIGaAs QWRs on high-index (33 l )B GaAs templates with nanoscale fluctuations. Photoluminescence (PL) from these QWR structures at 8 K shows thc degrcc of polarization to bc as high as -28%, which indicatcs good ID carricr confincmcnt. The dcnsity of these QWRs is estimated to be -2.OX 10' cm- ' , which mccts the requircmcnt of high optical gain for laser In the samples with weak lateral confinement, we obsemed thermal delocalization of carriers from the one-dimensional QWR states to the two- dimensionaI quantum-well states with increasing temperature, which is almost absent in the QWR samples with strong lateral confinement.

The samples used in our experiment were grown using a solid source MBE system (Riber 32) under an As beam equivalent pressure of I X Torr. The surface evolution during MBE growth was characterized by in sitzi reflection high-energy electron diffraction (WEED) and scanning tunneling microscopy (STM). A GaAs buffer layer, with a thickness of 500 nm, was first grown on thc n-typc GaAs (33 l )B substrate at a temperature of 61 0 "C. As shown in thc bottom of Fig. 1, thc resulting surfacc morphology, with straight ridgc-like corrugation bounded by thc (11 1) and (1 10) facets, was quenched to 540 "C as a template for the subsequent InGaAs overgrowth. The surface corrugation has an average lateral periodicity of -5.0 nm with a vertical amplitude of - 1.0 nm. A detailed description of preparing such (33 l ) B GaAs templates with nanoscale fluctuations can be found in Ref. 7. During the InGaAs overgrowth at 540 "C, atoinically flat surfaces were developed for all the san~ples with different deposition thickness (>1.0 nm), as monitored from the RHEED patterns and STM images. Finally, a 20 nm GaAs cap layer was grown at 540 "C for each sample to maintain the smooth InGaAs surface configmation. The QWR structures are thus formed at thick parts in the InGaAsiGaAs QW with a corrugated bottom InGaAs-on-GaAs interface and a smooth top GaAs-on-InGaAs intcrfacc, which is schcinatically shown in Fig. 1 . The structure of as-grown QWRs can be viewed as a totally "inverted" counterpart compared with

1609 O 2004 American Institute of Physics

Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:lljap.aip.org/japlcopyright.jsp

--

OSA Trends in Optics and Photonics Series, v 88, p 1031-1033,2003

CUA2 7

Collision-induced energy exchange between vector solitons Baolai Liang, Yongan Tang, Aqiang Cuo, and Gregory J. Sslamo

Deparlmmt of Physics, University of Arkansas, Fayetteville, Mansas 72701 Phone ( % I ) 575-5931; FAY: (501) 575-4380 ;

Dnail:solom~dvark.edu

Mordechai Segev, Physics Ikpnrlnienl; Technion - lerael Institute of technology. Haifa 32000.

Israel. and Electrical Engineeqg ddparlmcn4 Ru~celon Univenity, Rincetoi?'NJ 08344

Gary L. Wood U.S. r b y Research Labontory

%MSRL.SE-EO Adelphi. Mafyland 20783-1 197

Abstract: The energy exchange betwecn colliding photorefractive vector soli~ons is experimentalty investigated as a fundion of the collision angIe and the input intensity ratio of the vector soliton components. The application of tk results from lhjs investigation for computation wit11 solitons will be discussed 02002 Optical Society of America OClS coda: (1 90.5330 ) Uiotorefnctive nonlinear opticr; (230.7380) Spolial soliton

Vector solitons consist of two (or more) components that mutually self trap in a nonlinear medium, which were fist

suggested by Manakev for Kcrr materials [ I ] . Photorefractivc vector soliton are of particular intcrest since they can

be observed with any mn-instantaneous nonlinearity [2-51. Furthermore, the energyexchange interactions between

vector solitons provide a novel method to t d e r infonnat~on from one soliton to another, providing the foundation

for computation with solitons (61. In this paper, we investigate the energy exchange between colliding vector

solitons as a function of UE collision angle and the rnput intensity ratio of the vector soliton.

Fig. I Experimental sclup

Our experiments are canied out with the setup shown in Fig.]. The crystal is a Ce:SBN cystal with dimensions

13mmxl2mmxGmm. The c-axis is along Ihe 13mm direction and the soliton beams propagation along the 12mm

direction. Two mutual-incoherent argon lasers operating at a wavelength of 511.5mn provide the extraordinarily

QELS - Technical Digest Series, v 89, p QWA211-QWA2/2,2003

Transition from a quantum-dot to a quantum-wire electronic structure in InGaAsIGaAs quantum-dot chains

X. Y. W a g , Yu. I. Mazur,.W. Q. Ma,& M. Wmg, G. J. Sdamo, and M Xim L h p r t ~ ~ s n l ofPhynds, Unlwrrr& ofArhnrhr, F o y r W l I ~ , AR 72701

Phme (1 79) 575-6568. Fa: (1 79) 575-1580, uwJI1: (nxloo(ama1 uark.edu

T.D. Mishima and M. Johnson Doprrtrnsnt ofphyrics & A~trononry. Uniw8ily cfoklahama Norman, OK 73019. USA

Abatmct: We report on tk carrier hamport behaviols observed in a vertically and laterally organized quanhundot array. Photoluminescence poladzahon studies show a dual doUwire electronic sbuchue and a trans~tion bctween them as Ihe temperature is increased 82003 -a1 Soelcty of Amrnur OC18 coda: (300 6470) SpcNorcapy. rcnuconduclon, (160 6000) Semlmnduclon. inchdmg MQW

Recently, electronic and opto-elcctronic devices based on quantum wires (QWRs) and qua- dots (QDs) have been the subject of inlexst due to tk i r potential applications as Ism and detectors [I]. AS a result, there has been an extensive effort to manipulate and control tk position, size, shape, ard density of QDs, as well as to invesligatc their optical and electronic properties. In this talk, we report on the fabrication of a vertically and laterally ordered (In,Ga)As/GaAs QD army utilizing multi-lay- veItical stacking grown at elevated temperahue. Photolumiouceoce (PL) polarization studies showed a dual doVwire electronic structure and a transition from a zero-dimensional quantumdot system to a onedimensional quantum-wire system as the temperature was i d .

The sample was grown on a semi-insulating GaAs I1001 substrate using solid source molecular beam epitaxy W E ) . Figures l(a) and @) show dark-field plane-view TEM and cross-sectional TEM (XTEM) images of a typical multi-layer QD sample, respectively. The XTEM image shows a high degree of veItical island conelation (>95%), and the plane-view TEM image indicates that within each layer, all of the QDs ~IE densely packed in long lateral chains (-1-2 pm in length). The average diameter of the QDs within these lateml chains is - 45 nm while their height is about 5 nm. The average distance between QDs within a chain is about 20 nm. The lateml chains are separated from each other by about 70 to 80 mn and appear to sit on a common InGaAs base with an estimated height of about 2 nm.

A most interesting feature of these (In, Ga)AdGaAs QD chains is the enhanced polaridon in the direction along the chains with the incmsing temperatun. As seen in Fig.2, tbcpolarization anisotropy (PA), defined as VII - 11) 1 (11, + IL), where Ill and Ilknote the PL intensities with the polarization parallel and perpendicular to the chains direction, respectively, increases with temperature. This noted increase in PA with temperature is a strong evidence that caniers in our stmture exhibit both dot and wire quantum confinemenls in merent temperahue regions.

Based on mticulous strutaural anatysi. from thc TEM images and also theoretical calculations (21, in the inset of Fig. 2, we propose a conduction-band energy level model as shown in the inset of Fig. 2 to explain this enhanced poiarization from a temperature as low as 20 K. For all QDs united in one chak there exists a common InGaAs base, which can be considered as a 1D wetting layer (IDWL), analogous to thc 2D wetting layer (ZDWL) generally existing in the Stranski-Krastanov growth mode. Due to the relatively small mrgy sepamtion between the QD ground slates and this IDWL at 20-30 K, carrier thermal emission begins out of the QDs and is captured into the IDWL energy level. The captured caniers can relax over a long distance along the chain and find a lower energy minimum among the ensemble of QDs in the chain In this casc, a higher temperature should lead to evm more migration of canien along the chains. As a result, the chain of dots can be represented as a 1D wire-like structure and consequenlly one should expect the PL spectmm to demonstrate a PA when comparing PL emslon parallel and perpendicular to tk dimtion of the chains. This energy level model was further confirmed by the investigation of the PL spectra from these ordercd arrays of QDs, as a functron of both temperature and optical excitation intensity.

Reiernca: 1. D. Bimhcrg, M. ONdmuu~ n d N.N. Lcdhwv. planrumDorHer.rorrrucrun~ (Wilcy. Chichcatir, 1998). 2. P. Huririlo~ Qucmhim W d k WlresmdDors (Wiley. New Yak, 2000).

Proceedings of the SPIE, v 5646, n 1, p 1-5,28 Jan. 2005 , '

Diffraction management in 2D waveguide arrays Aqiang Guo, Yongan Tang, Will Black and Gregory J. Salamo

Lkplrmleni 0fPhylrcs. Universily 0fArkaIi~0~. Foy~tteville. Arkan303 72701 .. . Phone (501) 575-5931; FAX: (501) 575-4580 :

-.- - Emarl: [email protected]

Abstract: We report on the experimental observation of diffraction management in h a - dimensional (2D) "fixed" periodic waveguide arrays in a photorefractive (PR) fernelectric crystal. @2003 Optical Society of A n 1 6 OCIS rod-: (190.5330 ) PbotomfraQive nonlinear optics; (230.7380 ) Waveguides. channeled

Thus far, optical waveguide arrays have provided a .ferlile gmud for nonlinear optical applications [I-121. Waveguide m a y s are a set of coupled identical waveguides. During propagation, light energy can be transferred among the waveguides through linear coupling. The broadening of light propagating in such an array can lead to discrete diffraction [l]. TI= diffraction can be engineered [l] by varying the waveguide spacing and/or the transverse wave vector (initial phase tilt). When the Bloch wave vector is nearly half-way to the edge of the Brillouin zone in t l ~ reciprocal space, diiraction is amsted. Near the edge of Brillouin zone diffraction becorns lous us ('negative') [I]. Inspired by the diffraction management in ID arrays [I], we demonstrate expeninentally that similar diffraction management also exists in 2D waveguide arrays. The "fwed 2D waveguide anays are formed by combining our earlier work 1131 for fixing oneltwo waveguides with the so-called holograghic lithography technique [14]. In our experimental, three Argon laser plane waves with ordiily-polarization at 514nm interfere to fonn an optical square lattice in a Icm cube, SBN:75 crystal. The geomnetry of the three beam's wave vectors is arranged as shown in Fig. l(a) which gives a square 2D lattice shown in Fig. l(b). This optical lattice, under a 3 KVlcm electric field, leads to a strong space charge field compared with cocrcive field in the crystal. It turned out'tlle domain structure of the crystal is modified and a "fixed" 2D waveguide army pattern fomed,Fig. l(c) shows the "fixed" 2D wavepde arrays. The separation between the nearest waveguides here is llw Pig. l(d) gives the first and second Brillouin zone in the reciprocal space for such a square lattice shown in Fig. l(b). With the waveguides fixed, we investigated the evolution of the extraonhwy polarized argon probe beam, ~ b i i 1 - 9 i s focused onto the input face of the crystal with 15 p m Fig. 2 shows tbe diffraction in the homogeneous case (,m waveguide array exists in the crystal). Fig. 2 (a). (b) and (c) show lhe photograplt, horizontal and vertical beam profile of the input beam respectively and Fig 2 (d), (e) and (f) show the photograph, horizontal and veriical bew profile of the (diffracted) output beam, which is captured by a CCD camera. At normal incidence, wtuph corresponding to the case of Bloch wave vector at center of the Brillouin zone (r point) in the reciprocal space pig. 3(i)l, a diffnction distribution spanning approximately 6 waveguides is observed pig. 3(a)], which differs considerably f m ~ n the continuous diffraction without waveguide anay. In order to show similar diffraction management as 1D anays [l], we varying the angle of incidence beam, wl-qch is achieved by rotating the crystal. In the experiment, the rotating a ~ s is parallel to caxis of the crystal, which corresponding to changing the Bloch wave vector along the y-axis direction in the reciprocal space in Fig. l(d). By clunging incident beam's angle, diffraction behavior clunges. At an incident angle of 0.8g0 outside of the crystal, i.e., Bloch wave vector close to the middle point of r and M (0.94") pig. 3(ii)], diffraction is arrested. Fig. 3 (d), (e) and (f) show photograph, horizontal profile and vertical profile respectively. In conclusion, we have studied the linear diflraction of beams in 2D waveguide arrays. By varying the angle of incidence beam diffraction can be arrested.

Fig. 1 (a) ge&etric arrangement of 3 interference beam's wave vectors, @) 2D square lanice, (c) fixed ZD waveguide m y palt?rn ( 1 I m), (d) rrcipmcal space and Brilloum zu~w.

Authorized licensed use limited to: University of Arkansas. Downloaded on October 29, 2008 at 1651 from IEEE Xplore. Restrictions apply.

-

Conference on Lasers and Electro-Optics (CLEO), p 2 pp. vo1.2,2004

Forming 2D a waveguide array in PR ferroelectric crystal

Aqiang Guo, Yongan Tang, Baolai Liang and Gmgory J. Salamo Deprfmenr ofphysics. Universiw ofArkansas. FayenmeYIlle, Arkansas 72701

Phone (501) 575-5931: FAX: (501) 575-4580 : Email: [email protected]

Abstract: We delnonstrate experimentally a fixed two-dimensional (2D) periodic waveguide array by plane-wave interference in a photorefractive (PR) fernelectric crystal, and then demonstnte the localized states as a function of applied electric fded. 632003 Optical Society of ~iner ik OClS rode,: (190.5330 ) Photorefractive nonlinear optics; (20.7380 ) Waveguides, channeled; (220.3740 ) L i t h o ~ P h y

Recently, optical waveguide my's become the most promising candidates for nonliwar optical applications [I-121. However, the fabrication of 2D periodic structure presents many technological challenges. PR ferroelectric crystals have drawn considerable attention in ,& nonlinear optical area and device applications because of 1;ic-t-3-

optical properties and the for feature'of fenoelectric domain reversal which substantially modlfy the material's tensor properties at a desirable dimensionality and configuration In one of the most commonly used PR ferroelectric crystals; strontium barium niobate Sr,Bal.,NbzO6 (SBN:x), SBN:75 has the smallest femlectric domain size [13]. Therefo~, either the PR space-charge field or an applied external electric filed can modify the orientation of femeletric domains. Transforming a "real-time" s c w e g soliton into one or multiple permanent or "futed" waveguides by means of ferroelectric domain reversal in SBN:75 has a l ~ a d y been demonstrated [14]. In what follows we w i l l demonstrate, by using similar procedure as described in our earlier work [14], the formation of 2D optical lanices. Here, instead of using oneltwo soliton beams as writing beams we use three non-coplanar interfering plane wave beams, referred to as the holographic technique [IS]. The three interference beam's wave vectors are chosen such that the differences kik , (i=l, j-2, 3), wldc11 define the reciprocal lattice, give the 2D &sired lattice. Ln our experimental, an argon-ion laser with ordinary-polarization at 5 14nm is split into three plane waves wlqch afe then interfered to form an optical triangular lanice in a lcm cube SBN:75 crystal. The lattice constant (Cemer-t0- center separation behveen nearest waveguides) of the waveguide array is 16 pm [Fig. l(a)]. After formation of the m y a large electric field, typically 3 kV/cm is applied along the c-axis while the three beams incident o n the .

cvstal. The modified domain structun: inside the crystal after this procedure leads to a "f~ed" 2D waveguide array panem pig. I@)]. Fig. l(c) shows the far-field laser diffraction pattern, slmwing the fixed array has a triangular symmetry. Although the refractive index contrasts in such materials are small, the expected advantage of this approach is the option to have relatively long samples, and therefore even weak coupling can become significant. The guided property of the waveguide arrays is tested by an extra-ordinary polarized He-Ne laser probe beam that is focused on the input face of the crystal. The output face of the crystal is imaged by an imaging lens nnfn s CCD cameras. Fig. 2 (a) and,@) shows probe h m at the i q u t face (10 tun) and output face of the crystal when waveguide anays are absent. W k n t k probe beam was injected into a single central waveguide (on-site) it was mainly localized in the same waveguide and a couple of its nearest neighbor Fig. 2(c)]. Fig. 2(d) meanwhile shows that the probe beam cannot be guided when the probe beam injected at an off-site location. Fig. 3 shows a localized guiding state of an on-site shot can be either enhanced or reduced by applying a positive electric field Fig. 3(a)-(c)], or a negative field pig. 3(a), (d) and (e)]. In conclusion, we llave shown that 2D waveguide arrays can be faed into a PR ferroelectric material via plane wave interference that could provide reconfigurable photonic lattice with different p u p symmetry Since the light- induced optical lattices provide much greater control of the grating parameters than fabricated waveguide amlys, we believe that these results may open new possibilities for the study of various nonlinear effects in 2D optical lattice.

Authorized licensed use limited to: Univenity of Arkansas. Downloaded on October 29. 2008 at 16:52 from IEEE Xplore. Restrictions apply.

Li et al. Vol. 20, No. 6IJune 20031.1. Opt. Soc. An. I3 1285

High-efficiency blue-light generation by frequency doubling of picosecond

pulses in a thick KNbO, crystal

Yong-qing Li

Department of Physics, East Carolina University, Grcenvill~, 'Vorth Caroli~ra 27834-4353

Dorel Guzun, Greg Salamo, and Min Xiao

Dcpartmlent of Physics, University of Arkansas, Fayettcvillc, ~lrkansas 72701

Received August 5, 2002: revised n~anuscript received December 10, 2002

We report an experimental demonstration of highly efficient single-pass second-harmonic generation from 859 nm to 429.5 nm with picosecond pulses in a thick KNb03 crystal. Both the conversion-efficiency and quantum-noise properties of the generated blue pulses are measured at various pump intensities under a strong focusing condition. We find that the variation of the conversion efficiency of the picosecond second- harmonic generation is an oscillatory function of the input pump intensity (with a ~naxilnun~ efficiency of 56.5%) and is sensitive to the position of the input beam focus in the crystal. The quantuni noise on the hlue beam can be reduced below the shot-noise limit by 20% at low input power. O 2003 Optical Society of An~enca

OCZS codes: 190.2620, 270.6570.

1. INTRODUCTION Second-harmonic generation (SHG) is an attractive optical frequency conversion process to generate coherent light a t shorter wavelengths.' Very high conversion efficiency has been reported under extremely high pump

i.e., a n efficiency of 92% from 1064 nm to 532 nm in bulk nonlinear crystals with a typical pump intensity of a few G ~ l c m ? ' Recently. single-pass traveling-wave (TW) SHG with cw mode-locked pulses has attracted attention because of low average input power with high peak intensity, and conversion efficiency higher than 60% has been obtained both in bulk nonlinear crystals"8 (NLCs) and in quasi-phase-matched nonlinear w a v e g ~ i d e s . ~ ~ ' ~ Highly efficient TW SHG a t a low average input power is a n interesting nonlinear optical system that not only conveniently provides a coherent light source a t short wavelengths, but also allows generation and observation of amplitude-squeezed light that has less-intensity noise than the standard shot-noise In fact, 6.7% 10.3-dB) amplitude squeezing in SHG was observed with a conversion efficiency of 15% in the TW SHG experiment with a type I1 phase-matched bulk KTP crystal." In an experiment with a LiNb03 w a v e g ~ i d e , ~ 16.8% (0.8-dB) squeezing in the transmitted fundamental field and 7.7% (0.35 dB) squeezing in the generated harmonic light were observed with a conversion efficiency ap- proaching 60%. In the TW SHG with a thick potassium niobate (KNb03) crystal pumped by femtosecond pulses, a slope efficiency of 300% n ~ - ' for harmonic conversion was achieved5 and a 20% (1.0-dB) squeezing in the generated harmonic light was observed with a conversion efficiency of 60% a t -90 mW average input power.7~'2

In this paper, we study highly efficient single-pass SHG

in a thick, noncritically phase-matched KlVbO, crystal pumped by picosecond pulses and measure the quantufp- noise properties of the generated blue pulses. Wc show that the TW SHG process in KNb03 with picosecond pulses is quite different from that with fc~~~tosecond pulses. Much less peak intensity is necessary to obtain the high conversion efficiency for picosecond pulses nncler the strong focusing condition. We found that !.he vnria- tion of the coilversion efficiency of the picoseconfi SHC is a n oscillatory function of the input pump intensity (with a maximum efficiency of 56.5%) and is sensitive to the position of the input beam focus in the crystal. 'she quantum noise on the blue beam can be reduced Ijctlow the shot-noise limit by 20% a t low input power.

2. BACKGROUND KNbO, is an attractive nonlinear crystal that !;AS a large .

nonlinear coefficient and noncritical phase ills. 4 : .;ir r !-:" , spatial walk-off) under appropriate conditions ,;or 106- power optical applications.'g-'5 However, this crystal also has a large group-velocity mismatch (a = 1 .:! ps/mF for the wavelength conversion from 860 nm to 430 n m ) and a narrow phase-matching bandwidth of iv. = 0.88IaL (-0.2 nm for a crystal length of L = 10 mnl),"' which generally limit it from short-pulse, widehalid applications. In the previous TW SHG experiments with fern.- tosecond pulses,5-8 the input spectral width ( - 10 nrn for a Gaussian pulse with duration time t , = 120 fa) is rnuch broader than the phase-matching bandwidth. 'I'hus r;tif temporal walk-off length 1 = t , la (-100 bm), ovir which a fundamental pulse and a harmonic pulst: miss temporal overlap from each other completely, is much

0740-3224/2003/061285-05$15.00 O 2003 Optical Society ofhnerica

4PPLIFD PtlYSICS LFTTERS VOLUMF 82. NUMBER 4 17 IA \ l \ ICY 200:

.. '

Photoluminescence of metalorganic-chemical-vapor-deposition-grown . '

GalnNAsIGaAs single quantum wells M. 0. ~ a n a s r e h ~ ) Department (~'Elec/ricul and Computer Engineering, Universi~). v/'!Venv h f e . ~ i c ~ . Alhuqrierque. New Mexico 87/31 and Department of Electrical Engineering, Liniversity of ,4rkanrlrs, Fa.vette~~ille. .4rkunra.s 72701

D. J. Friedman ~Ytrfionol Renen*ahle Energy Laboratory, 1161 7 Cole Borrlevord~, Golden, Colorodo 80401

W. Q. Ma, C. L. Workman, C. E. George, and G. J. Sa lamo Deptrrtment of Phj~sics, University of'drkansas, Fayetteville, ,4rkansm 72701

(Received 8 October 2002; accepted 3 December 2002)

Photoluminescence (PL) spectra of interband transitions in GaInNAsIGaAs single quantum wells grown by metalorganic chcmical vapor deposition on scmi-insulating GaAs substrates wcrc measured at 77 K for several san~ples grown with different In compositions and dimethylhydrazine (DMH)/III ratios. The rcsults show that thc PL intensity increases as the hl mole fraction is increased from 0% to 25%, but the PL intensity is degraded for samples with an In mole fraction of 30% or highcr. The pcak position encrgics of the PL spectra wcrc invcstigatcd as a function of the DMH!III ratio. Thernlal annealing effect induced a blueshift in the PL spectra peak position energy in samples grown with high DMHIIII ratios. 0 2003 American Ir~stitllte of Phpsics. [DOI: 10.1063/1.1540731]

Diluted or small-band-gap nitride semiconductors, such as GaInNAs and GaNAs, are currently being investigated for their optoelectronic applications, such as 1.3- ant1 1.55-pm emitters used for optical communication. Devices based on this class of materials possess advantages over other material systems. For example, the higher temperature characteristic of GaInNAsIGaAs lasers provides an advantage over the GaInPAsAnP lasers. The InGaNAsIGaAs system has a larger conduction band offset, which provides a higher quantum confinement.' The GaInNAs-based system is grown on GaAs substrate that is more robust compared to the 1nP substrates. Bragg ref ectors are easy to fabricate for GaInNAsIGaAs vertical cavity surface-emitting lasers compared to GaInPAsIlnP Bragg reflectors.' The investigation of the GalnNAsIGaAs system was motivated by the fabrication of light emitters that can cover the entire visible spectral range based on the direct band-gap materials. Recently, emission was observed in GalnNAsIGaAs quanhim wells,3-' and quantum dot^.^-'^ The thermal annealing effect on the photoluminescence (PL) spcctra was also investigatcd by various groups.4.1'-'4 From various reports on the growth and characterization of the IllGaNAsIGaAs system, it is realized that nitrogen incorpo- ration leads to a number of properties that were found to be attractive for device applications.

In this letter, we report on the PL of GalnNAsIGaAs single quantum wells grown on scmi-insulating GaAs sub- stratcs. Thc PL spectra were investigated at 77 K as a function of In con~position, dimethylhydrazine (DMH)/III ratios, and thermal annealing. The PL intensity, full width at half maximum (FWHM), and the peak position energy were all

"Elec~ronic mail: manasreh@eece.~~nm.edu

found to strongly depend on the incorporatici:r G , I C ! 3 . ;I "! atoms.

Thc structures wcrc grown by atmospheric-prcswrc metalorganic vapor-phase epitaxy at 570 "C on semi-~nsulating GaAs oriented 2" from (100) to (1 10). Trimethylgalliuni. trimethylindium, arsine, and DMH werc used as precursors. The growth rate was 5 prnlh for the GaInNAs a c ! ~ ~ e layers. The N content was changed for the various sampies by van;- ing thc DMH source flow ratc. As described by ttlc yuiuntrty DMHAII, the ratio of DMH to group-111 flows urls injected into the reactor. The GaInNAs active layers w c ~ c 100, A thick, and were cladded on both sides by GaAs-tlol)cd n-type with silicon from a disilane precursor. After the growth of each structure was completed, but before it ~ . a p ~cnlovetl from the reactor, it was annealed under arslne foi 30 mln at 650 "C. A few salnples were annealed for LO mrn at 750 "C under arsine in the metalorganic chemical vapol. deposrtipn (MOCVD) reactor. The arsine flow rate during he anneal was 30 sccm. For comparison, thc actual growtl~ r > f all ttlc original epilayers was done with an arslne fit-.]. ,!. , 3 E--:-

Additional samplcs werc amcaled at 800 "C witl; riu arsi11.c; but with a GaAs wafer lying face down on the epilayer a$ i~ :

proximity cap. Thc InGaAsIGaAs singlc-qu~!iltlrn~-L\.$il samples were grown on semi-insulating GaAs suhs~l.ates]in 3. molecular-beam-epitaxy (MBE) chanlbcr. Thc growth tek- perature was 585 OC and the sanlples were post-growth an: nealcd at 600 "C for 10 min. The PL spectra wc1.c ircorded using BOMEM DA8 spectrometer in conjunctior; with continuous flow ciyostat.

PL spectra obtained for several samples are sl~own in Fig. 1. The 1n mole fraction was fixed for all nine .;;implcs at 7%, white the DMHIIII ratio varied, as shown i r ~ ~ h r irlsct.

The parameter S i n the inset was taken as 6= D,\IHi( 1)Xl.H + AsH3). Sample 1 does not contain any N (DMH I I1 rat~o' is

0003-695112003182(4)1514131$20.00 514 O 2003 American lnstitu1e of phyi1es Downloaded 27 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapl/copyright.jsp

Mat. Res. Soc. Symp. Proc. Vol. 776 O 2003 Materials Research Society Q11.29

Normal Incidence Intersubband Transitions in InGaAsIGaAs Quantum Dots With Non-monotonic Shift

M. L. Hussein, W. Q. Ma, and G.J. Salamo Microelectronics-Photonics, University of Arkansas, Fayetteville, Arkansas 7270 1

Abstract Multiple layers of self assembled Ino.sG~.~As quantum dots of different size were

grown on GaAs (100) using molecular beam epitaxy. Fourier-transform infrared spectroscopy shows absorption in the long-wavelength infrared region (8-10 pn) under normal incidence. The absorbance peak shift with dot size was investigated and revealed non-monotonic behavior of intersubband transitions. The optical absorption coefficient was calculated to be in order of 3.8~10' cm-'.

Introduction Quantum dots (QDs) are currently attracting much attention for their unique underlying

physical properties. One exciting application is in the area of long-wavelength infrared detectors. Due to their three-dimensional confinement, optical transition selection rules for QDs allow normal incident light to couple with electrons that undergoes intersubband transitions. This is in sharp contrast to quantum wells (QWs), where optical transitions for normal incident light are forbidden. This restriction has forced the development of optical detectors to add a grating layer to the detector in order to redirect normal incident light to propagate along the well and optically couple to the QW. Of course, this solution increases fabrication cost.

L€ QDs are to find application as infrared detectors the optical absorption of Q D structures must be well characterized to calculate and predict the quantum efficiency. The results of long wavelength infrared detectors are usually expressed by photocurrent or photoresponse [1,2]. In addition to the absorption coefficient, another important parameter is the peak absorption wavelength that is a sensitive function of dot size. For example, the infrared absorbance peak- shift in quantum dots [3,4] and in quantum wires [S] has been demonstrated as a function of nanostructure size with a predicted monotonic behavior of the shift of the absorption peak. While there are several theoretical calculations of the absorption coefficient of QDs [6,7] structures, there are few corresponding experimental measurements. In this paper, we report and explain a non-monotonic absorbance peak-shift in Si doped Ino.3Gao.7As IGaAs multiple QDs structures. In addition, we also report on measurements of the optical absorption coefficient in the investigated samples.

Experiment The InGaAslGaAs quantum dot structures were grown by a solid-source molecular beam

epitaxy system on semi-insulating GaAs (100) substrate. A typical structure consists of a 0.5 1.m thick Si doped (1018 ~ m - ~ ) GaAs buffer layer followed by 20 periods of Si doped Ino.sGao.7As dots and 30 nm GaAs barrier. The structure is then capped by a 0.5 thick Si doped (10 '~ cm- IF' -3 3, GaAs layer. The quantum dot layers are doped with [Sil- 1 . 5 ~ 1 0 cm . The growth temperature of InGaAsIGaAs multi-layers was 510 "C with $s4 to Ga beam equivalent pressure (BEP) ratio of about 20.

APPLIED PI-IYSICS LETTERS VOLUME 83, NUMBER 9 1 SEPTEMBFR 2003

Hidden resonant excitation of photoluminescence in bilayer arrays of InAsIGaAs quantum dots

Yu. I. Mazur, Z. M. Wang, G. J. Salamo, and Min xiaoa) Departnient q/ 'Pl iy~~ics , U~iiversitv ofArktmsas, Fa,vetteville, Arkansas 72701

G. G. Tarasov, Z. Ya. Zhuchenko, and W. T. Masselink Institutjur Plysik, Humboldt-Universit;if zir Berlin, Inl~alidenstrasse 110, 10115 Berlin, Gerrnc1n.v

H. Kissel Ferdinand-Braun-Institutyiir Hochstfequenztechnik, Albert-Einstein-Strasse 11. 0 -12489 Berlin, Germany

(Received 7 April 2003; accepted 2 July 2003)

Photolunlinescence (PL) of self-organized quantum dots (QDs) in bilayer InAsIGaAs structures is studicd with a fixed seed laycr and spaccr, but variablc sccond-layer coverage. Careful line shape analysis reveals modulation in the high-energy tail of the seed-layer PL spectrum. The oscillation-like behavior is reproducible with variatioils in both thc tenlperaturc and optical excitation energy. These oscillations are attributed to carrier relaxation through inelastic phonon scattering from the wetting layer to the QD excited states. O 2003 Anzerican Institute o f Physics. [DOT: 10.1063/1.1606109]

Carrier relaxation, including tunneling through potential barriers, represents one of the most fundaineiltal processes described by quantum kinetics. Recently, tuilneling between laycrs of sciniconductor quantum dots (QDs) has attracted attention in view of the development of the growth capability to realize vertically aligned or stacked QD g e ~ m e t r i e s . ' - ~ It is also of interest because a good understanding of the different excitation and decay channels of excited states in multilayer QD structures will have important consequences for their potential application as emitter or detector arrays.

For stacked QD structures, the coupling strength between QD layers is controllcd by systematic variation of the barricr thickness behveen layers and/or control over the dot size or composition froin one layer to thc ncxt. For cxainplc, if the barrier is thick enough, the electronic levels in the layers of dots will have carrier recombination in each layer separately. However, if the barrier is reduced the carrier wave function in each dot in one layer can spread beyond the barrier into an adjacent dot in another layer, and as such coupling is established between the levels of QDs in bilayer structures. In this case carriers can be photogenerated in a given well and tunneling to an adjacent vertically aligned dot can occur if the tunneling time is less than the recombination time. In similar manner, by varying the QD size or composition, the energy separation between the two tunneling QD layers call be used to control the tunneling probability.

While due attention has been directed toward the coupling betwccn closely stacked QD layers,5 thc role of the col~csponding wetting laycr (WL) in such vertically aligned structures has been relatively less explored. In ordcr to distinguish different contributions to thc relaxation of photocx- cited carriers we consider a structure composed of two layers of weakly coupled TnAsIGaAs QDs. This geonlehy allows us to discriminate between the coupling QDs in adjacent layers and the coupling between QDs in the seed layer and the WL associated with QDs of the adjacent or second layer.

"Electmnic mail: mxiao(Cijmail.uark.edu

The samples were grown using a solid-source molecular bcam cpitaxy chambcr couplcd to an ultrahigh vacuunl scanning tunneling n~icroscope (STM). The structures consist of two InAs layers. Each sample was grown on a GaAs (001) substratc, followcd by a 0.5 pnl GaAs buffcr laycr and 10 inin annealing at 580°C to provide a nearly defect-free atomically flat surface. The seed QD layer is then grown by depositing 1.8 monolayer (ML) of InAs at a growth ratc of 0.1 ML/s, As4 partial pressure of 8 X Torr. and substrate temperature of 500 "C. This is followed by 50 ML GaAs deposited on top of thc sced QD laycr. Thc sccond QD laycr was then added. The InAs deposition coverage in the second layer was varied from 1.8 to 2.7 ML for different samples used in our ineasurcmcnts. Each samplc for optical study was finally capped with a 150 ML GaAs layer. The samples were structurally characterized by plane-view STM and cross-sectional transmission elcctroil n~icroscopy (XTEM). The photoluminescence (PL) was studied in temperature range of 10-300 K using the 5 14.5 nm line of an Ar' laser for GaAs excitation, a Ti-sapphire laser for WL excitation, as well as a HeNe laser for intermediate energy excitatioil range, thus spanning excitation densities from 0.1 to 20 w/cn12.

Figure 1 (a) shows a typical XTEM image of the sainple with 1.8 and 2.4 ML depositions for the seed layer and the second layer, respectively. Statistical analyses of XTCM inl- ages show the resulting sample to be a weakly vertically correlated (-50% for 50 ML GaAs spacer thickness) double- laycr QD structure, designcd with a significantly differcilt average QD size in the seed layer compared to that in thc second layer. The QDs in the second layer are nearly twice the size of those in the seed layer due to additional dcposition, as well as to the influence of the strain field from the seed layer.','' The STM statistical analysis [Fig. l(b)] indicates a QD size distribution of 4? 1.5 nm for the height. 2 0 2 3 nm for the width, and a dot density of about 4.5 X 101° cmP2 in the seed layer. The dot density in the second layer is variable over the range of 2.5-4 X 10" cnl-' ; " depending on the TnAs coverage.

0003-6951 /2003/83(9)11866/3/$20.00 1866 O 2003 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http://apl.aip.orglapllcopyright.jsp

APPLlFD PHYSICS LETTERS VOLUME 83, NUMRFR 9 I SFPTFMBTR 2003

Morphology evolution during strained (In,Ga)As epitaxial growth on GaAs vicinal (100) surfaces

Z. M. ~ a n g , ~ ) J. L. Shultz, and G. J. Salamo physic.^ Deparlrne~it, Univer.~itv qf Arkansas, Fayelleville, Arkansas 72701

(Received 20 March 2003; accepted 7 July 2003)

Molecular-beam-epitaxy growth of strained (In,Ga)As on GaAs vicinal (100) surfaces is investigated by scanning tunneling microscopy. Surface roughing as the consequence of step bunching driven by strain is explored. By tuning the In content over the range from 0.05 to 0.2, the step bunching is observed to exhibit considerable uniformity and order. These results experimentally demonstrate that strain-driven step bunching is a viable approach to provide templates for nanostructure growth. O 2003 Arnericun Institzite qf'Pllysics. [DOI: 10.1063/1.1606891]

Step patterns formed on vicinal GaAs(100) surfaces, with a crystal orientation of several degrees off from (loo), provide attractive templates to fabricate semiconductor nanoscale structures. For example, nanostructure evolution is reported for submonolayer deposition of GaAs and (Al, Ga)As ~ u ~ e r l a t t i c e s ' . ~ and for the initial stages of Si dep~sition'.~ on stepped GaAs surfaces. However, the steps formed on GaAs vicinal (100) surfaces grown by molecular beam epitaxy (MBE) have a height of one monolayer (ML), arc often i~~cgular ly spaced, and havc rough ~ d ~ e s . ' ~ ~ Whilc this step configuration is helpful in facilitating smooth growth via step flow. it has also been translated into useful templates for subscqucnt ~lanostructurc growth by step bunching. In principle, bunched steps tend to be much more straighter than single ML steps. Moreover, the height and spacing of the bunches can be controlled depending on the number of steps in the bunches. Different mechanisms for the formation of step bunching have been proposed based on either kinetics or the~mod~nani ics .~-" For strained layers, the strain relaxation in the step edge produces a long-range attractive step-step interaction that leads to a step bunching instability." Computer simulations show that a strained layer on a vicinal surface can self-organize into a regular array of step bunches." Consistently, vertically c o ~ ~ e l a t e d interfacial roughness is observed in strained SiGe/Si12 and ( G ~ , I ~ ) A S / G ~ ( P , A S ) ' ~ superlattices. With these previous studies in mind, it is interesting to study strained (In,Ga)As growth on vicinal GaAs(100) surfaces to uncover how stcp evolution is driven by strain and if it can provide useful templates for nanostructure growth.

The interest in strained (In,Ga)As growth is also driven by the desire to realize long wavelength lasers around 1.3 p m from GaAs-based photonic devices for fiber-optic corn- munications. For example, the use of a strain reduction layer of (In,Ga)As to surround lnAs quantum dots has becomc a widely crnployed approach."," For this application, lulowl- edge of the surface morphology of strained (In,Ga)As layers is a prerequisite to understand the self-assembly of InAs quantum dots.

In this letter, MBE growth of strained (In,Ga)As layers

"Electronic mail: miwimg(~uark.edu

on vicinal GaAs(100) surfaces was studied by scanning tunncling microscopy (STM). On the bare (2 X 4 ) rcconstructed GaAs vicinal surface, single ML steps with ragged edges were observed to be irregularly spaced. The change of surface morphology of (In,Ga)As layers as a function of In content from 0.05 to 0.2 was investigated. Step bunching was observed after (In,Ga)As growth. Surprisingly, the bunches are vcry straight and unifommly spaced for a particular In content.

The experiments were carried out in a combined MBE- STM ultrahigh vacuum system equipped with a reflection high-energy electron diffraction (RHEED) system and a highly accurate ( 2 2 "C) optical transmission thenuo~netry system for substrate temperature determination. Epitaxial ready 17-type (Si doped ~ O ' ~ / C I I I ~ ) vicinal GaAs( 100) substrates misoriented, 2" towards the [ I 11]B direction, were loadcd into thc MBE growth chamber. The surfacc oxide layer was removed at 600 "C while exposing the surface to a 10 pTorr As4 beam equivalent pressure (BEP) from a solid source valved controlled cell. A 0.5-pm-thick GaAs buffer layer was grown at 580 "C using a growth rate of 1.0 ML/s and an As4 to Ga BEP ratio of 20. The resulting surface morphology was frozen to 500 "C by gradually quenching the substrate and lowering the As4 BEP while keeping a (2 X4) RHEED pattern. Subsequently, 15 nm (In,Ga)As was grown with the Ga flux corresponding to a 1.0 ML/s GaAs growth rate and the In flux adjusted by changing In source temperatures in order to vary the In content from 0.05 to 0.2 for different samples. The observed RHEED patterns revealed the resulting surfaces remain (2 X 4) reconstructed but become disordered. The samples were quickly quenched by cutting powcr to thc substrate heater and lowering the As4 BEP. After the As4 source was closed conipletely at a substrate tc~npcraturc of 350 "C, the samplc was rapidly transferred under UHV to the STM chamber for analysis. All STM scans were takcn at room tcmperaturc using tungsten tips. The tunneling current was set constant between 0.1 and 0.2 nA, and the negative sample bias was 2-3 V.

Figure I(a) shows a representative image of the bare GaAs vicinal ( 100) surface. In good agreement with previous

the surface is characterized by a high density of steps running along [011] with ragged edges. The average spacing between steps is around 8 nm, consistent with the 2"

0003-6951 12003183(9)/1749131$20.00 1749 0 2003 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to Alp license or copyright; see http:llapl.aip.orglapllcopyright.Jsp

Formation of low-density InAs/InP(001) quantum dot arrays

Yu. I. ~azur' , H ~ i s s e l ~ , and an^', G.J. ~alamo', and M. ~ i a o '

'university of Arkansas, Department of Physics, Fayetteville, AR, 72701

2 Ferdinand-Braun-Institut fuer Hoechstfiequenztechnik, Materials Characterization Division, Optical Characterization, Berlin, Germany

ABSTRACT Carrier transfer in low-density InAs~lnP dot arrays with a multi-modal dot size distribution is studied by means of steady-state photoluminescence. The transition from saturation of the inter-dot carrier transfer to the unsaturated regime is surely observed by analyzing the shape of the luminescence signal for decreasing excitation densities. We unambiguously show that larger size dots provide a competing but saturable relaxation channel for smaller quantum dot ground states.

I. INTRODUCTION Both energy relaxation and carrier transfer mechanisms in semiconductor quantum dots (QD) have been the focus of recent investigations based on their important role in the operation of QD-based devices [l-31. In these studies, an interesting problem occurs when the QD array exhibits a size distribution that shows more than one maximum, e.g. a bi- or multi-modal Q D size distribution [4-61. Such complicated systems have application potential since their optical properties could be tailored via the collaboration between modal sizes. In this paper we report on the excitation intensity and temperature dependent photoluminescence (PL) measurements of such QD ensembles. Our results help to clarify the processes of energy relaxation and canier transfer in multiple and coupled QD systems [7-121. In previous work, an unusual temperature dependence of both the PL peak energy position and corresponding linewidth has been reported [4,7]. In particular, a redshift of the PL spectrum and an anomalous decrease of bandwith with increasing temperature have both been observed. These unusual effects have been attributed to a themally enhanced carrier relaxation between QDs due to canier thermionic emission and carrier transport through the wetting layer (WL). The QDs PL intensity is also observed to quench with increasing temperature due to thermally driven carrier escape from the QD ground states to states of the matrix material, andlor nonradiative recombination centers. Although, some theoretical and experimental studies [13-151 have been done on this subject many issues are still not clear. For example, the influences of QD size and density on the luminescence energies and linewidths from the QD ensembles are still not well understood, and the mechanisms involved in carrier relaxation and PL quenching seem strongly dependent on the material system, excitation conditions, and even the method of QD formation. In this report, we present a detailed study of carrier transfer between QDs in low-density InAs/InP(001) QD arrays (density - (1-5)x10~ cm-2 ) with a multi-modal QD size distribution. The investigation is based on steady-state PL measurements made over a wide range of temperature and excitation intensity. The particular QD system under investigation consists of a grouping of a few QDs (2 to 4) at a low overall QD density with the mean separation within each group to be about or less than the lateral size of the QDs (- 20 nm) while the average distance between the groups is about 1 micrometer. Therefore, the study is carried out on isolated QD groups. We find that the PL temperature and excitation intensity dependence of these isolated QD groups depends on the carrier transfer between the QDs of different sizes within the same QD group. Carrier transfer involves transitions from the ground state of the small QDs into lower lying states of larger QDs, a relaxation channel that saturates at high excitation densities.

11. EXPERIMENTAL SETUP Our QD growths have been carried out in a solid source (arsenic and phosphorous) molecular beam epitaxy (MBE) chamber coupled to an ultra high vacuum, plane view, scanning tunneling microscope (STM). The MBE is equipped with a high-energy electron source enabling reflection high-energy electron diffraction (RHEED) to be performed during

VI Intl. Conf. on Material Science and Material Properties for Infrared Optoelectronics, F. F. Sizov, J. V. Gurnenjuk-Sichevska, S. A. Kostyukevych, Editors, Proc. of SPlE Vol. 5065 (2003) 0 2003 SPlE .0277-786XlO31$15.00

PIPPLIFD PHYSIC'S LETTFRS VOLUME 83. NUMBER 5 3 AIIGUST 1003

InGaAsIGaAs three-dimensionally-ordered array of quantum dots Yu. I. Mazur, W. Q. Ma, X. Wang, Z. M. Wang, G. J. Salamo, and M. xiaoa) Department of Physics, Universip o f Arkunsas. Fayetteville, Arh-ansus 7,7701

T. D. Mishima and M. B. Johnson Depurrmenr o f Physics and Asrroizomy, University of Oklnhon~u, Norman, Oklul~omu 73019

(Rcccivcd 6 Novembcr 2002; acccpted 30 May 2003)

We report on the first fabrication of (In,Ga)As/GaAs quantum dots with both vertical and lateral ordering fomiing a three-dimensional array. An investigation of tlie photoluminescence spectra from the ordered array of quantum dots, as a function of both temperature and optical excitation intensity, reveals both a lateral and vertical transfer of excitation. 0 2003 American Ii?stit2/te o fPhys ics . [DOI: 10.1063/1.1596712]

Recently? electronic and optoelectronic devices based on q ~ ~ a n t x ~ m wires and quantum dots (QDs) have been the subject of interest due to their potential applications as lasers, detectors, or photonic crystals.' As a result, there has been an extensive effort to manipulate and control the position, size, shape, and density of QDS.'

In this letter, we report on the fabrication of a vertically and laterally ordered (In,Ga)As/GaAs QD stack forming a tlxcc-dimensional QD array. The fabrication is achicvcd by utilizing multilaycr vertical stacking grown at clevatcd tcnipcraturc.6~" An invcstigation of the photoluniincsccncc (PL) spectra from thcsc ordcrcd arrays of QDs, as a function of both temperature and optical excitation intensity, reveals both a lateral and a vertical transfer of excitation.

The samples used in the experiments were grown on semi-insulating GaAs [001] substrates, with a miscut angle smaller than 0.05", using solid source molecular beatn epitaxy. After the native oxide was desorbed, a 150-nin-thick GaAs buffer layer was grown at 580 "C. The substrate was then cooled down to 540 "C for the growth of the rnultilayer dot structure. After a 2-iim-thick Ino,3GGao,G4As layer was grown, three monolayers of GaAs were deposited without growth interruption to suppress In segregation. Then, after an 8 s interruption, 16 nni of GaAs was grown. The period of thc superlattice was 15.

The saniplcs werc investigated using planc-view and cross-sectional transinission clcctron niicroscopy (XTEM) in order to cxaniinc vertical and latcral ordcring was pcrformcd using a JEOL JEM2000FX microscope opcratcd at 200 kV. PL studies wcrc pcrfomicd in a variablc-teniperaturc helium cryostat (4-300 K), under the excitation of the 514.5 nm line of a continuous-wave argon-ion laser. The PL signal was analyzed using a 0.5 m single-grating spectrometer, and detected using a photomultiplier tube.

Figures 1 (a) and l(b) show dark-field plan-view TEM and dark-field XTEM images of a typical multilayered QD samplc, respectively. The XTBM image [Fig. l(b)] shows an almost perfect vertical island correlation, and tlie plan-vicw TEM image [Fig. l(a)] indicates that within each layer, the QDs were slightly elongated along [GO] direction and densely packed in long lateral chains (- 1-2 p m in length). The effective 2D wetting layer (2DWL) thickness (about 0.7

"'Eectronlc mail: mxiao(iuark.edu

nm) is the same for all layers. The average diameter of the QDs is about 45 nm while the average height is about 5 nm. The average distance between QDs within a chain is about 20 nm. The separation between a QD in one chain and the nearest QD in a neighboring lateral chain is about 70-80 nin. The QDs in each chain appear to sit on a common InGaAs base with an estimated height of about 1.5-1 nm. The vertical correlation of tlie QDs is due to the vertical strain field betwccn thc buried and the subscqucnt QDs, wliilc the lateral ordering of the QDs is rclated to the strain-field-modulatcd surface along the [go] direction and tlie enhanced adatoni migration length at the elevated growth temperature.

According to the XTEM, the tunnel barrier thickness between the vertically aligned QDs (defined as the average distance between the island tip in one layer and the island base in the second InGaAs layer) is about I0 nni and is approxiinatcly two times sniallcr than tlic barrier thickness betwccn nciglibor dots within a lateral chain. Based on calculations using the Wentzel-Kraniers-Brillouiii approxinia- tion and treating the dots as thin quantum well^,'"'^ the electron tunneling time is estimated to be 7,-0.8 ns for square barriers of about 10 nm wide. In comparison, the typical radiative recombination time T, for InGaAs quantum dots is about 0.5-2 So, at low tcinperature, T, is comparable

FIG. 1 . (a) Dark-field plane-view TEM image of the sample obtained with g=200. (b) L ~ \ o ] XTEM imagc of the sample.

0003-6951 12003183(5)1987131$20.00 987 O 2003 American lnst~tute of Phys~cs Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcopyright.jsp

APPLIED PHYSICS LETTERS VOLUME 82, NUMBER l I

Strain-driven facet formation on self-assembled lnAs islands on GaAs (311)A

Z. M. ~ a n g , ~ ) H. Wen, V. R. Yazdanpanah, J. L. Shultz, and G. J. Salamo Department c?J'Pliysics, University of Arkrrtwas, Favefteville, Arkansas 72701

(Received 21 October 2002; accepted 17 January 2003)

The shape of InAs three-din~ensional islands grown on GaAs(3 1 l)A substrates by molecular-beam epitaxy was investigated by in situ scanning tunneling microscopy. The island is found to be laterally surrounded by (1 1 l)A and (1 10) facets together with a convex curved region close to the (100) facet. The top ridge of the islands is atomically resolved to be the most recently discovered high-index surface {11,5,2}. This observation points to the importance of the study of nanostructure growth on high-index surfaces and their characterization. Q 2003 American Institute oj'Physics. [DOI: 10.106311.1559945]

During the growth of highly strained epilayers, defect frce nanomctcr scale thrce-dimensional islands can nucleate as a mechanism of strain relicf. This growth mechanism is generally refei~cd to as self-assembled quantum dots (QDs). (In, Ga)As grown inside a GaAs matrix has bccoinc the dominant material system for compound semiconductor QDs. Extensive research has been recently directed toward developing a better understanding and improved control of this growth process. In spite of this intense effort, it is remarkable that the exact experimental deternlination of the island shape is still controversial.'-6 However, knowing the exact island structure is required to develop a complete understanding of the crystal growth mechanisms associated with its formation. Moreover, identification and control of the islaild shape is needed to engineer the QD electronic and optical properties for device applications.

For the inost part, investigations of self-assembled island foimation havc becn carricd out on GaAs (100) sub- stratcs. Thc growth of lnAs islands on GaAs high-indcx sub- stratcs provides additional opportunity for nanostructure engineering.798 For cxample, on GaAs(311)A, thc InAs islands display unique optical properties, such as a large optical anisotropy and the built-in piezoelectric e f f e ~ t . ~ ~ ' ~ Sur- prisingly, arrowhead-like faceted shape has been observed using scanning tunneling microscopy (STM)~ on the GaAs (31 1)A surface. The anisotropic shape is believed to result from the breakup of the GaAs(31 l)A surface into (331) facets. However, the resolution display in the STM results is not sufficient to make a different conclusion.

In this letter, the shape of InAs islands grown by inolecular beam epitaxy (MBE) on GaAs(3 11)A is uncovered using aton~ically resolved in situ ultrahigh vacuum (UHV) STM measurements. The shape is found to have (1 1 l)A and {llO}lateral bounding facets together with a curved near (100) surfacc profilc without distinguishing faccts. Mean- while, the top surface of the island is observed to be the high index facets, (1 1,5,2}.

The experiments were performed in an UHV-combined MBE-STM system. A GaAs(31 l)A substrate was first loaded

"'Electronic mall: m1wang(3uark.edu

into the transfer chamber, baked, and then transferred to the growth chamber. After loading the GaAs wafer, a 500 nni GaAs buffer layer was deposited at 580 "C followed by 9 A of InAs deposited at 500°C. The InAs growth was per- fornled using an In beam equivalent pressure (BEPI of I X lo-' Torr, and an As BEP of 1 X l o p 5 Torr. The InAs island forn~ation was confirincd by the transition of thc rc- flection high-energy electron diffraction (RHEED) pattern from a streaky character to a spotty character during InAs dcposition. After InAs deposition, no changc in thc RHEED pattern was observed as the sample was cooled down. After growth, the wafer was transferred under UHV conditions to a surface analysis chamber for room tempcraturc STM imaging. The STM imaging was performed using a 3 V gap voltage (filled state mode), and a constant tunneling current of 0.1 nA.

Before InAs deposition, the bare GaAsi3 1 l)A surface was observed to be an (8 X 1) reconstn~cted."." The surface shown in the inset of Fig. l(a) is characterized by 2 ML of zigzag As dimer rows running along the [-2331 direction with the lateral periodicity of 3.2 nm along [0 - I I ] . The inost frequent surfacc dcfcct is thc steps along [-2331. The ordering of the ( 8 x 1) reconstn~ction depends on the surface preparation. InAs deposition would suppress the ordered (8 X 1) surface corrugation at a very initial stage.13 Thc InAs wetting layer is disordered in the atomic scale. Further InAs deposition induces the transition from two- to three- dimcnsional growth. The rcsulting surface aftcr 9 of InAs deposition is shown in the STM image of Fig. l (a ) . The observed arrowhead-like islands are elongated along [-2331. Similar to earlier rcports, the luAs islands on GaAs(3 1 1 )A observed here have a broad and rnul tinlodal distribution of s i z c ~ . ~ ~ ~

To better distinguish the island facets, a gradient image of Fig. I(a) was takcn along the horizontal dircction, as shown in Fig. 2. Different facets in a gradient image display different gray levels. Island 2 and 3 with the height of about 3 nin are more rounded and difficult to identify facets. One facet is clearly observable on island 2. The bigger islands (Nos. 1 , 4, 5, 6, and 7) with the height of about 7 nm are more faceted. The facets are fonned when the dot size increases, which suggests that those facets are stabilized by the

0003-6951/2003/82(11 )/I 688/3/$20.00 1688 0 2003 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcopyright.jsp


Surface dynamics during phase transitions of GaAs(100)

Z. M. Wang* and G. 5. Salamo Phy.~ics Department, IJnivelsih, qf Arkan.~a.u, Fayetteville, ilr~konscis 72701

(Received 17 December 2002; revised manuscript received 10 Februaly 2003; published 31 March 2003)

Scanning h~nnciing microscopy is uscd to caph~rc thc initial stagc of thc transition from ~ ( 4 x 4 ) to (2 X4) reconslruction on a GaAs(100) surface. A model for the transition is proposed in which sursace atoms escape from thc c(4 X 4) reconstructed arca to fonn (2 x 4) rcconstructcd islands and pits. Thc proposcd explanation is consistent with ~ ( 4 x 4 ) nlodels having species intemlixing in the first or second layer.

DOI: 10.11031P11ysRevB.67.125324 PACS number(s): 81.15.Hi. 81.07.-b, 68.37.Ef, 61.14.Hg

The GaAs(1OO) surface has a large number of stoichiomctry-depcndent surfacc reconstruction^.^ An understanding of these reconstructed surfaces can play a significant role in the fabrication of GaAs devices. For example, the (2 X 4) and c (4 X 4 ) As-stable reconstructions are particularly important for GaAs-based optoeletronics, since growth by n~olecular beam epitaxy (MBE) and nletalorganic chemical vapor deposition generally takes place under As- rich conditions. As a results of extensive experimental and theoretical investigations, it is generally accepted that the (2 x 4) reconstruction contains two As dimcrs on thc top layer and another As dimer at the third layer,2x3 as shown in Fig. l(a). In conlparison with the (2 X 4) reconstruction, results from structural investigation~ of the c (4 x 4 ) reconstructed surfacc are less ~ertain. ' .~ ' In one c (4 X 4) model the top two As monolayers (ML's) are structured as shown in Fig. l(b). In this model, an excess 0.75 ML of As atoms are chemisorbed on an already As-terminated surface to foml [01 11-oriented As cli~ners in the c (4 X 4) symnlet~y, with onc missing dimer occurring at every four As dimers. While this modcl is acccptcd by many rescarch groups, rcsults from recent efforts to understand the transition between the ( 2 X 4) and the c (4 X 4 ) reconstruction argue the accuracy of this c (4 X4) model. For example, the transition from (2 X 4) to c (4 X 4) reconstruction is observed at a low substrate temperature of 400-500 "C. Consequently, the Ga surface atoms are ex ected to be relatively im~liobile in this temperature rangek'As a result. one must suspect that the simple addition of As atoms to the ( 2 x 4 ) surface leads to a c (4 X 4) surface with Ga and As intermixing in the third layer, as shown in Fig. l(c). Accordingly, a c ( 4 X 4) reconstructed region should stand above the (2 X 4) reconstructed region wllcn thc two phascs c o e x i s t . ~ o w e v c r , all c u ~ ~ e l l t invcstigations have demonstrated that the c (4 X 4) region is actually located below the ( 2 X 4) region.237,'0 As a result of the obscrvation of the coexistence of both reconstructed surfaces, a c ( 4 X 4 ) model with Ga and As intermixing in thc second layer has been proposed7 [Fig. 1 (d)]. This model is consistent with results from both mcdium-energy ion scattering studies1' and scanning tunneling n~icroscopy (STM) s t ~ d i e s . ~ . ~ More recently, yet another model for the c (4 X 4 ) structure with three mixed Ga-As dimcrs on top,8 as shown in Fig. l(e), is also claimed to be compatible with experimental and theoretical studies. Apparently, there is still sollle doubt as to which model is correct. While each of these

models are plausible several require that surface transport of a hugc amount of Ga atoms has to bc involvcd in tlic phasc transition between the ( 2 X 4) and the c(4 X 4 ) reconstruction. For example, it is already noted that the (2 X4) phase has to melt first in order to accommodate the c (4 X 4) configuration using thc ~nodcl with a top 1.75 ML of As. '" Ex- tremely long surface diffusion of Ga atoms IS proposed as the Ga source of thc phasc transition." Givcn thc uncertainty of a model for the c (4 X 4) reconstruction, a study focused on the surface transition between the (2 X4) and ~ ( 4 x 4 ) reconstructions can lead to better understanding of the c (4 X 4) structurc.

In this paper, we report on STM experiments that have observed the initial stages of the transition from the c (4 X 4) to ( 2 X 4) reconstructed surfaces. Based on observations in these experiments, the primary source of surface Ga atoms needed to support the transition from the c ( 4 X 4) to (2 X 4) surface reconstructions is Ga atoms jumping over a short range. In particular, the transition is observed to be characterized by an intermediate structure con~posed of (2 X 4) reconstructed islands and pits.

The experiment was carried out in a combined MBE-STM system under ultrahigh vacuum (UHV). The system is equipped witli an in sit24 optical system that nionitors the substrate band edge to give accurate growth temperatures. Epitaxial ready N-type GaAs(100) wafers were used for this experiment, After the growth of a 500-nnl GaAs buffer layer. the sample was annealed for 10 min at 580 "C under a constant As, flux of 1.0X Torr in ordcr to achievc a smooth growth front. A highly ordered (2 X 4) reconstructlon was maintained under this condition. The evolution of surface reconstructions was monitored by reflection high-energy electron diffraction (RHEED). By reducing the substrate tenipcrature to 480 "C at ramping rate of 30 "C pcr min. a well ordcred c(4 X 4 ) RHBED pattem was developed. The substrate temperature was then increased to 500 " C at ramping rate of 20 "C per minute to trigger the transit1011 from c(4 x 4) to (2 X 4 ) reconstnlction. Although the RHEED pattern was ~ ( 4 x 4 ) - l i k e still at 500 "C, it became less ordered. Dircctly after the temperature reachcd 500 "C, the resulting surface morphology was quickly quenched by turning off the substrate heater power and closing the As shutter. During this time the RHEED pattem remained constant. The sample was subsequcntly tralisfc~~cd through a UHV transfa chamber into the STM chamber. Constant current STM im-

0163-1829/2003/67(12)/125324(4)/$20.00 67 125324-1 02003 Tlie A~ne~icall Physical Society

APPLIED PHYSICS LETTERS VOLUME 82. NUMBER 15 14 APRIL ?O(l.:

Optical absorption of intersubband transitions in Ino.3Gao.7As/GaAs multiple quantum dots

B. Pattada, Jiayu Chen, Qiaoying Zhou, and M. 0. ~ a n a s r e h ~ ) lle[~cz~-tn~ent of h'leclrical Engineering, Bell Engineering Centel; Unii~ersiQ of Arkansas, Fuvetreville, Arkansas 7-7 701

M. L. Hussein, W. Ma, and G. J. Salamo Department of Physics, Universigj of'Arkansas, Fayetteville, Arkansas 73701

(Received 2 Janua~y 2003; accepted 18 February 2003)

Fourier-transform infrared spectroscopy technique was employed to investigate the optical absorption coefficient of intcrsubband transitions in Si-doped Ino,3Gao,7As/GaAs nlultiplc quantuin dot structures. Waveguides with 45' polished facets were fabricated from molecular beam epitaxy grown wafers with different quantuin dot size. The measured maximum optical absorption coefficient was found to be in the order of 1.10X lo4 cm-! The peak position energy of the intersubband transition was observed to shift toward Lower energy when the quantum dot size is increased as expected. The photoluminescence spectra were also measured for different samples with different quantum dot size. The internal qi~antun~ efficiency was estimated to be in the order of 58% for a sample with 40 periods of 6 nni dot size. Q 2003 Americatz Institute of Ph,v.~ic.v. [DOI: 10.1063/1.1567813]

Quantum dots are currently attracting much attention for their application as lasers and detectors as well as for their unique underlying physical properties. For example, one rather attractive application makes use of intersubband transitions in quantum dot structures for long wavelength infrared detection. The quantum dot llanostructure offers a unique solution to the polarization problem that is encountered when using quantun~ wells to engineer long wavelength infrarcd detectors. That is, one significant drawback to the use of multiplc quantuin wells for detectors based on intcrsubband transitions is that the absorption of nonnal incident photons is forbidden. Due to the dipole selection rules associated with the intersubband transitions, the incident photons must have a component (TM) of polarization normal to the s tn~c- ture interfaces to excite intersubband transitions and be absorbed. Typically, this requirement has been met by illumi- nating the sample at the Brewster's angle, polarizing the light with p-polarized light, or adding a grating layer at the top of the detector structure. Using intersubband transitions in quantum dots, where the selection rules allow nonnal incident light to effectively couple with intersubband transitions, can remove this complication.

The results for long wavelength infrared detectors fabricated from quantum dots are usually expressed in tenns of photocurrents (see for exa~nplc Rcfs. 1-5) and/or photore- sponsc (see for example Refs. 6-10). While there are extensive theoretical calculations (sec for exainple Refs. 11-21) on the optical absoiption cocfficicnt in quantum dots, thc reported experimental measurements are scarce and in some cases the absorbance measurements are indistinguishable from the background noise.'2 In this letter, we report on the measurements of the optical absorption coefficient of the intersubband transitions in Si-doped Ino,,Gao,,As/GaAs n~ul-


tiple quantuin dot structures. The peak position energies of the intcrsubband transitions of these samples were in thc 9-1 1 p m spectral range for quantum dot size of 4-7 nm. The Ino,3Gao,7As/GaAs quantuin dot structures wcre grown by a solid-source molecular beam epitaxy system on semi- insulating (1 00) GaAs substrate. A scanning tunneling microscope (STM) was attached to the growth system, which was used to determine the quantum dot size and thc dcnsity of the quantum dots. The quantum dot density was in the order of -8X 10" cm-2 for all sanlples. The STM was also used to determine the lateral dimensions of the quantum dots, which was found to be in the order of 40X 70 nin2. A typical s tn~c- ture consists of a 0.5-pm-thick Si-doped GaAs buffer layer followed by 40 periods of Si-doped Ino,,Gao,,As dots and 30 nm undoped GaAs barrier. The structure is then capped by a 0.5-pm-thick Si-doped GaAs layer. The quantum dot layers are usually doped with [Sil- 1.7X 1 0 ' ~ cm-'. The growth temperature of 1r1~,~Ga~,,As/GaAs multilayers was 500°C with background arsenic pressure of 1 X Torr. Several samples were grown with quantum dot height in the range of 4-7 nm as deternlined from the growth rate and the deposition time. Three sanlples were chosen for the present study with quantum dot size of 4, 6, and 7 nm. A high-resolution x-ray diffraction experiment was perfomled on the sa~nples to confirm their thicknesses and the In coinposition in the actual quantuin dot material. The optical absorption coefficient and the photoluminescence (PL) spectra were recorded using a BOMEM DA8 spectrometer in conjunction with continuous flow cryostat. The samples were cut into waveguide geometry with the facet being polished at 45' (see the inset in Fig. 1). The light bean1 was zigzagged across the width ( w ) of the san~ple, which was typically 1.5 nun. The sample thickness (d) including the substrate and the quailturn dot structure is in the order 0.35 mm. Thus, the number of passes

0003-6951 /2003/82(15)12509/3/$20.00 2509 O 2003 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to Alp license or copyright; see http:Ilapl.aip.orglapllcopyright.jsp


Piezoelectric effect in elongated (In,Ga)As islands on GaAs(1OO)

Wcnquan Ma,* Xiaoyong Wang, Zhiming Wang, Mohammad L. Hussein, John Shultz, Min Xiao, and Grcgo~y J . Salamo 1)eparltnenl of Ph.v.yics, IJniversilv of Arkun.ra.s, I+'uyelteville, .4r,kansas 72701

(Received 22 August 2002; revised manuscript received 19 November 2002; published 17 January 2003)

The piczoelcctric (PZ) cffcct is dc~nonstratcd for thc clongatcd three-dimensional (In,Ga)As islands grown 011 a GaAs (100) substrate. The photolu~ninescence (PL) spectrum is studied as a [unction oC excitation intensity. With increasing excitation intensity, a bluc sh~ft and a lincwidth reduction of thc PL pcak from thc (In,Ga)As islands are observed. The observed phenonlena are attributed to the screening of the internal atrain- induccd PZ field in thc (In,Ga)As islands.

DO]: 10.1 I03iPhysKcvH.67.0353 15 PACS nutnbcr(s): 78.55.Cr, 78.67.Lt. 7 7 . 6 5 . L ~

Sincc Smith first prcdictcd a large piezoelectric (PZ) effect in strained 111-V or 11-VI seiniconductor hctcrostructurcs,~ investigations havc focused on thc study of the PZ effect in semiconductors.'-' One reason for this interest is that the strain-induced PZ field can modify the band structure of semiconductors, and therefore, offers another design parameter for fabricating novel optical and elcchonic devices. For pseudomorphic heterolayers, the PZ polarization is determined by the symmetry of the latticc of the substrate and the orientation of the surface.' For example, for zincblende 111-V sen~iconductor materials, thc PZ polarization is along the growth direction for pseudon~orphic layers grown on the (1 1 1)-oricntcd ~ubstratc , ' ,~ while it vanishes for those grown on the (, 100)-oriented s ~ b s t r a t e . ~ For other non- (100) substratc orientations, the PZ polarization and PZ ficld usually can have both vertical (growth direction) and lateral (perpendicular to growth direction) components.2 This is probably the reason why the overwheln~ing maiority of previous invcstigations of thc PZ effcct in sen~iconductors wcrc perfonned on structures grown on the (Ill)-oriented s u r f ~ c e ~ - % n d on high index substrates in general.739,'0 How- ever, there have been some recent efforts in realizing the PZ effcct on the (100)-orientcd substrate for applications. For example, one approach to generate the PZ effect on the (100)-oriented substrate is to pattern thc as-grown structure along the [ O fl] direction8 while another is to apply surface acoustic waves." In this paper, we discuss a different approach based on the fabrication of three-dimensional (3D) structures, such as, quantum dots and quantum wires.

Quantum dot and wire struchlres have raised considerable interests because of their unique characteristics for optoelectronic and electronic device applications. Self-organized 3D islands based on the Stranski-Krastanov (SK) growth mode has proved to be a successful approach to fabricate quantum dot and wire structures. The shape of these SK islands can be controlled by changing the growth conditions'' making it possible to directly engineer quantum dot and quantum wire structures. For example, long and uniform (In,Ga)As/GaAs quantuin wires based on elongated islands have been fabricated using a su erlattice (SL) growth scheme plus an annealing process.'P Recently, the PZ effect has been observed in self-organized quantum dots grown on different high index surface^.'^'^ Howevcr, these saine studies also demonstrate that self-organized quantunl dot structures grown on the GaAs (100) substrate do not show the PZ In

this work, we demonstrate that the PZ effect is in fact observed in elongated 3D (In,Ga)As islands grown on a GaAs (100) substratc. Thc cvidencc for this obsc~liation is based on characterization of a sample with elongated islands by excitation intensity dcpendent photolumincscencc (PL) spcctros- copy. Specifically, a blue shift and a linewidth reduction of thc PL pcak from the (In,Ga)As islands arc observcd to take place with increasing optical excitation intensity. We at- tribute both the shift and thc lincwidth reduction to thc screening of the internal PZ field in the elongated islands. As a conlparison, a similar but morc syninietrical shaped quantum dot s t n ~ c h ~ r e was investigated in the saine way and a band filling effcct rathcr than thc PZ cffcct was obscrvcd.

A typical sample was grown on a semi-insulating GaAs (100) substrate by n~olecular beam epitaxy (MBE). Aftcr thc sample was introduced into the MBE growth chan~ber, the native oxide was desorped at 580°C followed by overheating the substrate to 600°C. After keeping the substrate at 600°C in Asq atinosphcrc for 10 min, thc sainplc was cooled down to 580°C anti a 150-nin-thick GaAs buffer layer was deposited. The substrate was then cooled down to 540°C and a 15-period (In,Ga)As/GaAs superlatticc was grown. Each time, immediately after the deposition of the (In,Ga)As layer, three monolayers of GaAs was grown without interruption to suppress In segregation and 20.3-nin GaAs was then added with growth intcriuption of 10 scc. Finally, a top (1n.Ga')As layer was grown in order to characterize the surface morphology. The whole growth process was moilitored by in situ reflection high-energy electron diffraction (RHEED). The RHEED pattern bccanlc spotty aftcr the growth of (1n.Ga)As layer indicating the appearance of 3D islands. The structural characterization was perfonned ex situ by x-ray diffraction (XRD) and atomic force microscopy (AFM).

Figure I (a) shows the w-2 0 scan of a double crystal XRD around the GaAs (400) reflcction ineasurcd with an open detector. Eight satellites appear within the measureinent rangc indicating thc good structural qual~ty of thc sample. The SL period of the san~ple was determined to be 23.7 nni from thc spacing bctwcen thc satellitcs. Figure l (b ) shows the siinulation by the XRD dynamic theory assuming a co- hercnt growth of the (In,Ga)As on GaAs. Good agrecnlcnt between the experimental and the simulated curves is observed. The obtained thicknesses of (In,Ga)As and GaAs layers are 2.5 and 21.2 nm, respectively, and the In con~position 0.28, which is in agreemc~lt with the values from RHEED

C2003 The Arnerican Physical Society

APPLIED PlIYSIC'S LETTERS VOLUME 82, NUMHEK l l 17 MARCII 2003

Highly anisotropic morphologies of GaAs(331) surfaces V. R. Yazdanpanah, Z. M. ~ a n g , ~ ) and G. J. Salamo Mici~oelectronics-Pizotonics, Univers iq of Arkansas, Fayetteville, Arkansas 72701

(Received 3 1 May 2002; accepted 24 January 2003)

The surface morphology of the GaAs(331) surface was investigated by in sitzl reflection high-energy clcctron diffraction and scanning tunneling microscopy. It was found, that GaAs(331) A and B surfaces are both faceted on a nanometer scale, containing (110) and (111) facets which are atomically resolvcd in real spacc. The rcsulting highly anisotropic ridge-likc surfaces can provc useful In the fabrication of quantum wire structures. O 2003 Amcric~ln h,stc.tutc of Phj~sics [DOI: lO.1063/1.1561571]

High index GaAs surfaccs have recently attracted significant attention based on thcir influence on the growth of GaAs based nanostructures.' First, they play an important role in determining the self-assembled nanostructures shape, size, and un i forn~i ty .~-~ Second, they provide unique templates for nanostructure Third, some of high index surfaces, like GaAs(33 l ) , are unstable and faceted,' providing site for nucleation of nanostructures. In fact, growth on faceted high index surface is dcmonstrated to be an alteima- tive to the well studied strain-driven growth of one and three-dimensional nanos t ruch~res .~~ '~ For example. surface faceting has found application in the realization of quantuin wire lasers with lower threshold currents than their quantuin well c o ~ n t e r ~ a r t s . ' ~ They also have potential for novel tran- sistor concepts based on lateral switching between wires." As a rcsult, efforts to i~nprovc understanding of these faceted surfaces and their role in the growth of nanostructures are needed in order to better control their size, shape, density, and uniformity. The pursuit of this understands began when GaAs(331) surfaces were observed to be faceted in reciprocal space using low-energy electron diffraetion12 and high- cnergy clectron diffraction (RHECD).~ More rcccntly, quasi- periodic inultiatomic step arrays with a lateral periodicity at the subinicroscale were revealed on the GaAs(33 1)A surface in rcal space by atomic forcc microscopy. However, a elear description of the GaAs(33 1) surface with nanometer resolution is still lacking.

In this lcttcr, the growth process 011 GaAs(33 1) by molecular bean1 epitaxy (MBE) was systemically studied by in sitzr RHEED and scanning tunneling ~nicroscopy (STM). Ridge-like surface modulation at a nanometer scale with bounding facets of (1 10) and ( 11 1) was atoinically resolved in real space. The observed surface corrugation with a nanoscale lateral periodicity gives credibility to the promise to act as a template in order to achieve high density uniform quan- tun1 wires with unique electronic properties.

Epiready n-type GaAs substrates with both sides polished were used to investigate both the (331)A and (331)B

As, flux. The GaAs epitaxial growth was performed at a growth rate of 400 nndh calibrated from RHEED oscillations for GaAs growth on the GaAs(100) surface. The beam- equivalent-pressure ratio of As, to Ga was 40 during the growth. Although, the As solid source was equipped with a valved As4 cracker, only the valve was used to control the As pressure. The growth on the GaAs substrate was monitored, in situ, by RHEED operating at a voltage of 20 kV. Observed surface morphologies deternlined from the RHEED pattern were retained by quickly quenching the sample. This was accomplished by turning off the substrate heater power and gradually decreasing the As4 background pressure by controlling thc As valvc. After completely closing the As, valvc at a substrate temperature of about 300 "C, the samples were transferred from the MBE growth chamber, through a transfer module, to an attached ultrahigh vacuum STM system. The surfaces were imaged using the constant current inode with a tunneling current of 0. I nA and sample biases between - 2 and - 3 V.

The (331)A-oriented GaAs surface is between the (1 10) and the (1 l l )A planes, 13.3" away from (1 10) and 22.0" away from (1 11)A. For GaAs growth at 500 "C, the observed typical RHEED patterns are shown in Figs. I (a) and l(bj. Similar to previous work,g the RHEED patterns in Fig. 1

"lrfaces The with a size of l o x l o n11n2, flG, 1, RHCED panclns of [Ilc G ~ s ( ~ ~ l ) A surhccs sccordcd ia) along were loaded into the solid source MBE system alld the [I - lo] and (b) along [ll-61 during CiaAs growth at 500°C. (c) KHEED face oxide layer was removed by annealing at 600°C in an pattern along [I-101 afier 5 ~n in anncaling at 580 "C. (d) RHEED paucin

taken along [- 1101 during AlAs growth at 580 "C:. The KlIEED streaks arc marked by~white lines in la) and ib). The white lines and dottcd lincs in ( C J

" ~ u t h o r to whom correspondence should be addrcsscd; electronic mail: indicate the titlcd strcaks reflccted from (110) faccts and ( 1 I I)A facets. amwang~uei-k.edu respectively.

0003-6951/2003/82(11 )I1 766/3/$20.00 1766 O 2003 American Institute of Physics Downloaded 09 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcopyright.jsp

Session 1470

Land Grant Research University Partnerships with HBCUs for Enhanced Undergraduate Research Opportunities

Ken Vickers, Willyerd Collier, Benita Douglas Wolff, Greg Salamo University of Arkansas

Background

The University of Arkansas (UA) is a Land Grant University with the stated mission of being "a nationally competitive, student-centered research university serving Arkansas and the world". Because of this mission, it is imperative that the University provides a nurturing environment for students from all portions of our society. Only then will we gain the benefit from the talents that exist throughout our population, including talents hidden in population segments that have not traditionally enrolled at this University (or sometimes in any post-secondary institution).

In the United States there is a tradition of strong Historically Black Colleges and Universities (HBCUs) that provide both a nurturing culture and strong academic preparation to students of color in our society. But many of these institutions do not support a graduate research program, instead developing relationships with graduate research institutions for post-graduate opportunities for their students.

As a research institution, the University of Arkansas has the need for strong graduate students to support its research initiatives. However, most graduate programs have not devoted the same level of attention, energy, time or resources to the recruitment of graduate students that athletic programs or enrollment services have devoted to the recruitment of undergraduate students. The result is that the student side of the search for graduate progradstudent matches has driven the graduate "recruitment" system.

Potential graduate students today have an advantage in finding good graduate matches because of the extensive search capability and knowledge contained in the World Wide Web. At the same time, these undergraduates also seek guidance from individuals in their home institutions concerning the general reputation of a university, college or department. Without recruitment efforts by graduate programs, undergraduate faculty lack knowledge of graduate programs at other institutions, which can limit prospective students' confidence in accepting academic opportunities that would well support their academic and career goals.

Even with the difficulties involved, HBCU students have found and enrolled in UA graduate programs. Upon arrival on campus, they found that there existed a lower level of interaction between research faculty and students at the UA as compared to the students' undergraduate HBCU. This change of academic operational culture, coupled with the change in workload at the graduate versus undergraduate level and the change in social factors from a black majority

Proceedings of'the 2002 American Society for Engineering Education Annual Conference & Exposition Copyright O 2002, American Society for Engineering Education

Session 1654 (2002-2251)

Launching an Innovation Incubator in a University Setting

Ron Foster, Ken Vickers, Greg Salamo, Otto Loewer, and John Ahlen University of ArkansasIArkansas Science and Technology Authority

Abstract:

A novel Innovation lncubator has been launched with the goal of enhancing both education and commercialization of technology. The lncubator supports area clients that have new ideas, but lack the resources to advance towards proof-of-concept. Graduates students are involved with the Incubator in screening clients, and working with clients to improve intellectual property position and develop initial business plans. Once a client is accepted for full Incubator support, a graduate student is assigned to the client for up to one year in order to perform on-campus research targeted at developing proof-of-concept for the client's idea. A voucher is included in order to provide for access to University facilities and equipment. Graduate students gain experience in real-world commercialization situations, and simultaneously provide benefit to the area economy.

This paper is a continuation of the paper delivered at ASEE 2001 conference entitled "University of Arkansas Innovation Incubator: Flaming the Sparks of Creativity" by Vickers, Salamo, Loewer and ~ h l e n ' . In the 2002 conference, we will discuss early implementation details of the Innovation Incubator and considerations on clients in active consideration. In addition, we will discuss strategies for managing communications, successes and failures.

A number of policies and procedures have been developed in support of the launch of the Innovation Incubator. The "rules of engagement" have been developed, including the limitation on scope of activity both geographically and technologicalIy. The applicant screening process is hndamentally linked with education goals, since graduate students participate at every stage. In addition, faculty members are involved in the critical decision-malung processes. An objective scoring method has been created in order to insure that bias is minimized, and a committee drawing from a broad knowledge and experience base has been created. Clients are rated on five factors that intend to be predictive of success in commercialization.

A major activity of the lncubator is the matching of talents, desires and skills of graduate students with a client opportunity. Ideally, the work that the graduate student completes with a client will lead naturally to a Masters-level thesis.

Proceedings ofthe 2002 Americun Societyjor Engineering Education Annual Conji?rmce & Exposition Copyright O 2002, American Society for Engineering Education

Session 3142

Graduate Student Practice of Technology Management: The Cohort Approach to Structuring Graduate Programs

Ken Vickers, Greg Salamo, Ronna Turner University of Arkansas

Background

Many conferences have been held to discuss the skills needed by engineering and technology program graduates to be successful in technology based careers. These conferences strive to understand the full spectrum ofjob requirements by typically including representatives of

1 . 2 . 3 . 4 . 5 academe, government, and industry. A common result of these conferences has been lists of desired characteristics in newly hired technologists, including first and foremost the academic competency demanded by the technology job position. But they follow this need for technical competency with a need for proficiency in operational and interpersonal skills, skills that allows technologists to apply their academic training in an efficient manner in today's high tech work environments.

In the field of technical decision-making, it was felt that technical proficiency will not be sufficient to assure that future scientists and engineers make proper decisions, or to even assure that they are successful in their personal careers. They must also be able to work effectively in areas outside of their technical expertise, as they are no longer allowed to exist in an isolated technical environment. The fact is that many products require a high level of technical sophistication to even evaluate if it is the proper product for an application. As a result, today there must be more interaction between the developers of a new technology product and the customer. The scientist or engineer is therefore forced into active participation in such areas as customer negotiations, marketing and business planning, and manufacturing support. While their need for technical competence is not being reduced to support their primary task, their need for other non-technical knowledge is being increased by the many secondary roles that they are being asked to play.

From the large industry perspective, the need for a broadened knowledge base in their scientists and engineers lies in the broad financial impact of the decisions they will make. In a survey of manufacturing engineering jobs, aso on^ reports, "The results.. .also emphasize the importance of a broad education. Engineers need to be technically proficient at their job and at the same time understand the economic and engineering implications of their decisions." The Boeing Company CEO Philip Condit has stated, "...it is important that engineering education also have breadth. Students need to know about business economics: What does it cost to build a project? What's involved in integration?"

Proceedings of'the 2002 American Sociefyjor Engineering Educa~ion Annual Conference & Exposit~on Copyright O 2002, American Society,/or Engineering Education

Universitv Research Fundine: More than ' . Support~ng the Best to D: the Best

H oward Birnbaum argues that Over the past 40 years, the US ented labor pool needed by our coun- research funding needs reform culture has changed significantly. try and will uncover new knowledcc

at both the university and funding- It no longer supports national goals to fill our spirit, drive our economy, agency levels (PHYSICS TODAY, based on authoritative arguments and improve our quality of life. Thv March 2002, page 49). He suggests without compelling logic or decisive approach that is now taking shape rather strongly that the decline in evidence. As a result, say that requires that we engage talented sci- dollar amount (in constant dollars) research at universities needs to be ence students and that we educate of a typical academic research award supported because exploration is our nation on the value of research is due to the diluting effects of such what we humans do or because it is as a means to attain a higher qual~ty

I'

things as multi-investigator awards the approach that has previously of life. Although outreach and ,. : . :L

and the requirement that research delivered so many good things, tion will continue to take time awnv proposals incorporate outreach though still true, is no longer suffi- from research, all scientists must programs. cient. We are now compelled to edu- play a role and not expect others tn

I offer an alternative explanation cate and convince our entire popula- do the job. I hold that science out- of the same data: What has led to tion of the crucial role that research reach and education programs can this decrease in funding is not new at universities will play in the con- better engage all talented individu- attitudes, new programs, and new tinued prosperity and defense of our als while developing a public apprc- requirements, but rather a great way of life. It is, therefore, simply ciation of the importance of research

between the methods used good strategy to engage every state when they are led by each of US. 1 in the recent past to secure in this endeavor. Both funding agen- believe that this effort will lead to a sional funding and those required cies and university administrators more skilled, diverse workforce t h ~ I

today. fact, the reform that ~ i ~ ~ - recognize the need and have adopted will, in turn, generate and contribute baum is suggesting is exactly what an attitude to develop and support the knowledge needed to meet the the new programs and requirements experimental programs accordingly. challenges ahead and win greater are all about, and the attitude he For example, to provide our public and congressional support for exemplifies in his article is exactly nation with the needed labor pool in university research. the reason that funding for research science and engineering, funding GREGORY SALAJIO at universities has been decreasing. agencies have begun to reach Out to fsalarnoQuark edu)

Although I agree that research at &risk youths and provide o ~ ~ o r t u - Un~versrty of Arkan uas nities for talented individuals to Fayetteutllv universities is funded to educate,

create new knowledge, and provide choose and pursue careers in science service, I disagree on how best to and engineering. In the 10% run, A ccording to Howard B~rnhaurn. accomplish these goals. How will the this approach will build a stronger the " 'marganne method' , , US produce the researcher pool national infrastructure and a more spreading research funds equally needed now and in the future? H~~ competitive nation than would be thin among all possible recipients 1s can we best nurture and harvest ere- true with an approach that supports a waste of resources." ative &as and talent? How do we the best to do the best. Funding Quite the contrary. Despite the maintain our strong contri- agency outreach programs, like all insulting sound, "margarine fund- bution to science and a robust econo- experiments, must be critically eval- ing" is the best way to encourage my? ~~d very important, and per- uated. However, a crucial part of the serendipity, creativity, and origi- haps more to his point, how do we evaluation of NSF outreach pro- nality in research. All university ensure that we will have the dollars grams is the independent assess- professors are expected to be effi- to make these goals possible? In ment that the research team must cient teachers and researchers. The answer to these questions, reform is provide to As a panel highly competitive system of faculty now taking place via the very pro- reviewer, I have seen proposals appointments assures that, with

grams questioned in his article. rejected due to an inadequate assess- rare exceptions, all university proiea- ment plan or poor track record sors have the ability and training for

Lettea a d O P & O ~ ate encouraged based on assessment of previous both of those roles. Although equal and sfiould he sent E ~ B I PHYSICS work. Although i t has taken some grants for all are indeed impractical.

American Center forPh~i=, time, accountability for outreach there are viable and fiscally respon- One physics fKpse, College Park, is now firmly built into the peer sible alternatives to the present all- MD 20740-3842 or by 6-mail to review system. or-nothing funding model. ~rkm@&~*o ( k g YQw s u m Of course, as Birnbaum says, all If we keep in mind the known *Sub@ . PL inJude pwram- of this effort takes valuable time rule of economics that the first dol-

iationl m A g a d Apt;me away fmm research. However, we lars are the most cost-eflici-+ thp J?h~ae mmbeL wereserve the tight to have an obligation to find a research funding model under which all ~ L I ~ s 13 edir lerrers. approach that will produce the tal- university researchers receive a

APPLIED PFIYSICS LElTERS VOLUME X I . NUMHFR 6

Vertically stacking self-assembled quantum wires Xiaodong Mu a n d Yujie J. Dnga) Department of Electrical and Co~nplrter Eng~neering, Lehigh Universily, Betl?lehetrt, Per~nsylvanra 18015

Haeyeon Yang a n d Gregory J. Salamo Depurtmerit qf'Physics, Universiv of'Arkunsas. Filyetteville, Arkansas 72701

(Receivcd 18 April 2002; accepted for publication 7 June 2002)

Self-assembled InP/InAs/InP quantum wires (QWRs) have been stacked for ten vertical periods a d characterized based on photolurninescence (PL) studies. Compared with single-period QWRs, behaviors in the PL spectra and some fundamental effects have been observed. Through the detailed analyses of the PL shapes. linewidths, and polarizations at different pump wavelengths, pump intensities, and sample temperatures, it is evidenced that the wire width and subband energy gradually decrease while the average wire thickness increases from the bottom period to the top one, period by period. Meanwhile, the average wire width gradually decreases. Following these results, growth conditions have been suggested, which can be essential to improving the optical quality of these self-assembled QWRs. Q 2002 American Institute of Physics. [DOI: 10.1063/1.1497993]

InAsIInP nanostructures have many potent~al applica- t ~ o n s In the near-infrared wavelength range, especially for optical communications. Consequently, InAsAnP quantum wells have been widely investigated in the past decade.' ' Recently, the InAs/InP quantum wells were converted to w~relike nanostructures or quantum dots at certain growth conditions due to phase transitions induced by a 3.2% lattice mismatch between InAs and 1 n ~ ( 0 0 1 ) . ~ - ' ~ In particular, six periods of 1 n A ~ / I n ~ , ~ ~ A l ~ , ~ ~ A s quantum wires (QWRs) were recently studied."hese nanostructures with the two- dimensional and threc-dimcnsional quantum confinement are expected to significantly improve the performances of opto- elcctro~iic devices." Furthermorc, a lot of fundamental effects on the quantum level can be explored on these s t r u c t u r ~ s . ' ~ To date, the severc size fluctuation in the self- assembled QWRs causes a broad emission linewidth that has masked the advantage of QWRs as emitters. Therefore, controlling the size fluctuation represents a critical issue for ex- ploring the finda~ncntal limit and practical applications of the self-assembled QWRs.

In this letter, the self-assembled InP/InAs/InP QWRs have been stacked for ten vertical periods. The photoluminescence (PL) spectra of these stacked QWRs were measured at different pump wavelengths, pump intensities, and sample temperatures. Following our experimental results and comparison with a single-period InP/InAs/InP QWRs. we have found s o n ~ e behaviors for the PL shapes, linewidths, and polarizations in our stacked QWRs. According to our analyses, we have determined the detailed structure for the stacked QWRs and provided some suggestions for narrowing the PL linewidth of the stacked self-assembled QWRs.

Our single-period and vertically stacked QW samples were grown on (001) InP substrates in a Riber 32 molecular- beam cpitaxy system. Thc depositcd thickncss of each InAs layer is 3.8 monolayers (MLs), grown at the temperature of 180 "C with a growth rate of 0.3 MLIs. The stackcd samplc consists of ten vertical periods of InAs/lnP superlattices with

"Electronic mall: [email protected]

a 30 nm thick InP cap layer. The thickness of each InP barrisr layer is 30 nni. The single-period sample has a 20 I?rn thick InP cap laycr that was grown without intcrruption i.11 thc top of the InAs layer. Our in sifu scanning-tunneling-111icroscope picture of the single-period structure shown in Ref. 10 re-':: veals that tlie InAs wirclikc nanostructurcs are forlncd 01: t.hc top of the InP layers after sample growth. T!;c average height, width, and length of the InAs QWRs are nieasurecl to be 23 A, 150 A, and 1 pm, respectively.

The pump source used in our measurement is 2 I U I I ~ J ~ ~ -

Ti:Sapphire laser. Both the stacked and single-pe.riod samples wcrc niountcd insidc a cryostat in which the samb~c temperature can be set anywhere in the range of 4.3-30n K . Figure 1 shows the typical PL spectra of the two ~amples at 4.3 K for pump wavelength of 845 and 925 nm, ri,;pcctively., : The measured linewidths of the PL spectra are ahmt 74 and 54 meV for the stacked and single-period structures, respectively. Obviously, the additionally broadened PL lincwidtli for the stacked structure implies a lager wire size B ~ c t ~ ~ a t i o q ' due to vertical stacking. However, in our results, thz shapCs of the PL peaks for the stacked QWRs, which are ~.omplct~ly different from each other at the above ~ W O - P L I ~ ! ~ u:B~c'-

lengths. Indeed, for the stacked QWRs, one can kLe that'tlle short-wavelength (high-energy) component of th<: PL peak pumped by the 845 lun beam becomes a sho~~lJcr at !he pump wavelength of 925 nm. However, since he similar difference in the PL line shapes can not be observed i11,the single-period QWRs at those two pump \ :.:t*mc,ti~s ,. . !SC,C '

Fig. I) the band-filling effect should be ruled cut at :cell pump intensity.

Another behavior observed by us is the dccr.:ase of the PL linewidth for the stacked QWRs when the teniperaturc: is raised from 4.3 to 80 K, while the linewidth for rhc single. period QWRs increases as shown in Fig. 2. In ordtr to ;;I-

derstand this difference between the two samples. we have meawred a series of the PL spectra for the single-period and stackcd QWRs at diiTcrcnt tcmpcratures in thq rangc of 4.3-80 K, respectively, see Figs. 3(a) and 3(bi One can clearly see that the linewidth decrease is due 1.3 that the

0003-6951/2002/8l(6)/1107/31B19.00 1107 O 2002 Amerlcan lnst~tute of Physics Downloaded 27 Mar 2008 to 130.184.237.6. Redistribution subject to Alp license or copyright; see http://apl.alp.org/aplicok~yr~ght,j~p

APPLIED PHYSICS LETTERS VOLUME 80, NUMBER 25 24 I U N E 2002

Response to "Comment on 'Thermal annealing effect on the intersublevel . transitions in lnAs quantum dots'" [Appl. Phys. Lett. 80, 4867 (2002)l

I

Y. Berhane and M. 0. ~ a n a s r e h ~ ) Deparrmenr of'Eleclrical and Conlpurer Eugineering, University of'New ~Mexico. Alhuqtrerque. Neiew Mexico 871 Jl

H. Yang and G. J. Salamo Deparrmenr of Phvsics, Universi@ of Arkansus, Fuyetteville, Arkunstw. 72701

(Received 29 May 2001; acceptcd for publication 8 May 2002)

[DOI: lO.106311.1489488]

Poole and ~ e r s ' have commented regarding the wetting layer in our recent publication2 on "Thermal annealing effect on intersublevel transitions in InAs quantum dots." In Table I, in Figs. 1 and 2, and in the text of our letter2 we listed the InP buffer layer, of thickness 1 . I and 0.6 ,urn, as the "wetting layer." This assignment is incorrect. These layers are in fact the InP buffer layer and not the wetting layer. The wetting layer in our growths is composed of I ML of InAs, -,P,, where by growth x is intended (AslP exchange can change this) to be zero before thennal annealing but is likely not zero after thernlal annealing. The band assigned as "7" in Figs. 1 and 2 in our letter2 is due to the wetting layer. Sup- porting evidence for this assignment is based on the energy location (between 1.20 and 1.30 eV) of the band and the fact that the peak position is shifted upward in energy as the thermal annealing temperahtre is increased. This shift is likely due to the difhsion of P atoms from the buffer layer into the wetting layer further increasing the P mole fraction in the wetting layer. Additionally, intersublevel transitions terminology is used in this response and in our letter2 to indicate recombination from excited states in the quantum dots and not to transitions between electron or hole subbands.

In addition to calling attention to this error, Poole and ~ e r s ' also disputed our assignment of intersublevel transitions of the InAs quantum dots (QDs) to the well-defined peaks in Figs. 1 and 2 in the 0.80 to 1.0 eV energy range. Instead, Poole and Aers' state that the same peaks are instead due to families of islands varying in height by single monolayer steps with ground state energies dominated by dot height and not by lateral size, i.e., monolayer fluctuations of platelets. Poole and Aers base their conclusion on four "pieces of evidence." First, they stated that the relative intensity of the peaks remains approximately constant with varying excitation intensity. This description is incorrect. Nowhere in our letter do we make this claim. In fact, for our cxperiment the relative height of the PL peaks was observed to bc a function of excitation intensity. Second, they stated that the PL peak positions reported in our letter are highly reproducible from sample to sample, including san~ples grown Inore than one year apart. This description is incorrect. Nowhere in our letter do we make such a claim of


samples grown one year apart. Certainly, samples p~s-iwn tin- der the exact same conditions did yield reprobiucible PI. spectra, but as Figs. 1 and 2 in our letter show, the PL peaks are different both in separation and location if the InAs tieposition is varied from 3.1 104.0 ML. Third, they stalcil that the peak positions of our data fit a calculation of ~ I L N I I I ~ state energy versus InAs layer thickness. This descriptlun is i11col.- rect. An examination of our data shows in fact that the tle- viation of our data from the proposed theoretical cur?:? based on non no layer fluctuations of platelets is substanti,li. For ex; ample, the first point plotted by Poole and Aers which the; assign to a 1 ML platelet, deviates by almost 200 meV Itam thc theoretical prediction. In fact, they recognizcd this de\,ia- tion but suggest that deviations are expected f l ~ l - !ow ML platelcts. However, Poole and Aers chosu I( ; - ! I w r (]:?[a from Fig. 2 in their comment, had they chosen Fig ! ril.1h1 our letter: the deviation is substantial at high ML platelets (Fig. I of this reply). Fourth, they state that tli; PL positions are consistent from research group ti, resea1ci? group, with some increasing spread at low dot thickness. This description is incorrect. In fact. most of our tiara po~nts from Figs. 1 and 2 of our letter' differ significan[ly from the platelet data of others.'. For example, both Figs. I and ?-of .

our letter show a large number of low energy peaks that are -' separated by a constant -36 meV. These peal,s a1.c $,e- scribed in Fig. 3 of a previous letter2 as due to a touplingof the intersublevel transitions with the interfax phonon modes. In addition to these peaks separated by -36 nicV

r Curve by Poole et a1 .. 4.OML no annealing

A 3.1ML no annealing - .'

' -

800 - I I I I I I I

2 3 4 5 6 7 8 9 Monolayers

FIG. I. Solid line is a plot of the calculated ground state cncrgy of 111As layers reported by Poole and Aers (Ref. I ) ; triangles and sqr141.e-s are kom Figs. 1 and 2 of Ref. 2, respectively.

0003-695112002/80(25)14869121$19.00 4869 O 2002 American Institute of Phys~cs Downloaded 27 Mar 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcop!,right.)ap

Investigation of ultra-low-load nanoindentation for the patterning of nanostructures

Curtis ~ a ~ l o ; ~ , Robin Princea, Ajay ~ a l s h e * * ~ , Laura ~ i e s t e<**~ , Gregory Salamoc, Seong Oh ChoC 'Department of Mechanical Engineering, University of Arkansas, Fayetteville, AR 72701 (USA)

b High Temperature Materials Laboratory (HTML), Oak Ridge National Laboratory, Oak Ridge, Tennessee 3783 1 (USA)

"Department of Physics, University of Arkansas, Fayetteville, AR 72701 (USA)

ABSTRACT

In this study, nanoindents are characterized for the patterning of nanostructures. Nanoindentation is performed on Si- doped (n-type) Vertical Gradient Freeze (VGF) GaAs (100) and epitaxial GaAs (100) using a diamond Cube Comer indenter. Nanoindentation of GaAs has been studied in the past, but not at extremely low loads. Previous research has been done on high load (50-200 mN) and low load (0.2-8 mN) nanoindentation. The size of nanostructures range from 10-100 nm, requiring a nanoindentation width in the same size range. The loads required to produce such small indentations are in a lower regime of <0.2 mN, which to the authors' knowledge is an area that has not been previously studied using a Cube Comer indenter. Ultra-low-load (<0.2 mN) nanoindentations are characterized in order to study the selective growth of nanostructures on the indentation sites. Indentations of less than 200 nrn in width are produced, and mechanical properties of the two materials including hardness and elastic modulus are calculated and compared. The geometry of the indents is characterized using atomic force microscopy (AFM).

Keywords: ultra-low load nanoindentation, GaAs, nanostructures, patterning, AFM

1. INTRODUCTION

1.1 Motivation The development of techniques for the selective growth of nanostructures is essential for the realization of future nano- mechanical, electronic, and optical devices. Many techniques have been explored to achieve control over nanostructure growth. However, few techniques exist for highly selective patteming of nanostructures. In particular, controlled growth or patteming of quantum dots is made possible by introducing areas on epitaxial films where the surface energy is low or where there is a substantial density of seed material surrounding a given site. Quantum dots also grow on strain-relaxed surfaces, where the lattice misfit is decreased.

Recently, progress has been made in the formation of quantum dot array patterns by the creation of mesa, trench, and hole features by etching or mechanical modification of the growth surface. Mesa and trench features fabricated in GaAs by wet'" and electron beam lithography4 are shown to provide controlled growth of bunched linear positioned quantum dot structures, while hole features are shown to provide highly selective sites for positioning fewer dots and in some cases the growth of a single quantum dot at each hole site. It is speculated that one of the primary reasons for dot nucleation at the hole sites is due to the presence of multi-atomic steps inside of the holes. These steps act as barriers to material diffusion and allow accumulation of seed material at the sites, which lead to the formation of dots.

Another technique used to produce trench and hole features in GaAs is the use of scanning probe tips. Highly selective growth of quantum dots has been demonstrated on nitrogen-passivated GaAs (001) surfaces patterned using a scanning tunneling microscope (sTM)'. The tip of the STM is used to depassivate specific areas of the GaAs surface by

[email protected]; phone 479-575-5583; fax 479-575-6982; "[email protected]; phone 479-575-6561 ; fax 479-575-6982; ... [email protected]; phone 865-574-2588; fax 865-574-6098

Nano- and Microtechnology: Materials, Processes, Packaging, and Systems, Dinesh K. Sood, Ajay P. Malshe, Ryutaro Maeda. Editors, Proceedings of SPlE Vol. 4936 (2002)

Q 2002 SPlE .0277-786)(1021$15.00


Origin of step formation on the GaAs(311) surface

Z. M. Wang, V. R. Yazdanpanah, C. L. Workman, W. Q. Ma, J. L. Shultz, and G. J. Salaino Ph~2.sic.s lleparlrnerz/, lJnic.eeri)?i oJilrkan.ra,v, Faj~etfeville, .41.kans~~s 72701

(Received 27 August 2002; published 27 November 2002)

GaAs(3 1 I ) surfaccs grown by molecular bca~n cpitaxy arc invcstigatcd by i l l si/rc ultrahigh-vacuum scannlng tunnelling microscopy. The observation of an atomically flat Ga(2 X 1 )-reconstructed GaAs(311) surface and its transformation to a 8 X 1 -rcconstructcd GaAs(3 11) surface lcads to an improvcd undcrstanding of thc processes involved in the step fornlation. The high density of steps observed on the 8X 1-reconstructed - GaAs(3 11) surrace along the [233] direction originates from the change of surface atomic density required 10

accommodate tl~c surfacc transition from thc Ga(2 X 1 ) surfacc to the 8 x 1 surfacc. This undcrstanding IS

further supported by the observation of independent step formation.

DOI: 10.1 103/PhysRevB.66.193313 PACS number(s): 81.15.Hi, 61.14.Hg. 81.07.-b. 68.37.Ef

Self-assembled nanostructures, made by depositing a few monolaycrs of onc sclniconductor matcrial onto anothcr with a significant lattice mismatch, have been the focus of much rcscarch. The potential of these nanos t r~c tures '~~ is particularly noted by recent achievements with quantum dot lasers and detectors, although their pcrformancc havc bccn rathcr limited by their lack of uniformity of both size and shape. In particular, the current lack of uniformity typically associated with self-assembled growth has resulted in a rather large broadening in the density of states, presenting a fornlable obstacle for the application of nanostructures. The difficulty In obtaining uniformity has, in turn. sparked recent interest in the growth on high index surfaces, with an interest in patterning growth. In addition, not only are investigations of such surfaces meaningful for the application of nanostructures, they simultaneously provide a significant opportunity to gain a greater understanding of the GaAs ~ u r f a c e ~ - ~ in order to uncover its unique electrical and optical properties for advanccd optoclcctronic d c v i c c ~ . ~ ~ ' ~

An example of the importance of surfaces with high Millcr indices, such as the (3 11) GaAs surface, to the formation of nanostructures is their significant role in determining thcir shape1'-" and size u r ~ i f o n n i t ~ . l ~ . ' ~ Along these lines, the (3 11) surface (reportedly the first stable high index GaAs sur~ace'-~) was recently used as a template to foml highly unifornl quantum dot These exciting observations led to the possibility of using othcr high index surfaces, and consequently to the Inore fundamental question of when and why high index surfaces are stable. In order to understand the origin of high index surfaces, it is necessary to study them directly. Such an understanding may led to the ability to control and select the inorphology of surfaces.

In this investigation a surface phase, with atomically flat Ga(2 X 1 ) reconstructed wide tclyaccs, is obscrvcd under Ga- rich conditions, and its transformation into many narrow 8 X 1 -reconstructed terraces separated by randomly nucleated

steps along the [ z 3 ] direction is studied. These steps, interesting in their own right, can have a strong impact on the physical properties of GaAs(3 11 )-based structures, such as the optical polarization and electric transport anisotropy. For example, by assembling these steps, a one-dimensional quantum structure was observed and fi~rther combined with pat-

tern growth to provide uniform quantum Our investigation denlonstratcs that thcsc steps are folnlcd to accomnlodate the change of the surface atom density required by the Ga(2 X 1 ) to 8 X 1 phase transition.

Experiments were carried out in a combined Riber molecular beam epitaxy1Omicron scanning tunneling rnkcro- scope (STM) ultrahigh-vacuum system. For the experilllent described here, epiready n-type GaAs substrates with both sides polished were used for comparative studies of the (3 1 l)A and (3 11)B ~urfaces.~"n this paper, however, we focus on the (3 1 l)B surface, which is located about halfway in between the As-te~nlinatcd (1 1 l)B and Ga-tcrminatcd (100) surface orientations? although a similar step behavior was observed on the (31 l)A surface. A I -,urn-thick GaAs buffer layer is grown by molecular beam epltaxy at 580 O C with a growth ratc of 1.0 MLloo/S ( 1 ML,o ,= 2.83 A) and an As4/Ga beam equivalent ratio of 20. Reflection high- energy electron diffraction (RHEED) was used in . r i m to monitor the growth process. A (1 X 1 ) -like RHBBD pattern was obscrvcd during growth, which was denoted as Ga( 1 X 1) before.' This pattern was kept stable by closing the Ga and As shutters at the same time, and quickly removing the sample to the transfer module and cooling it down there out of the residual As, background influence. The sample was then transferred to thc STM chamber without breaking ultra- highvacuum conditions. The base pressure of the transfer module and the STM chamber are around 2 X 10- 'O~orr. Filled-state STM images were collected at sample voltages of 2-3 V and tunnelling currents of about 0.1 nA.

The resulting surfacc topography is shown in Fig. l(a). More than 100-nm-wide atomic-flat terraces are separated by 1 -ML3, ,-height (1 ML3, ,= 1.7 A) steps. This surface morphology is conlparable in quality with the GaAs(100) surface, and consistent with the observed excellent RHEED oscillations on GaAs(311)B. The high-resolution STM nnage of Fig. l(b) reveals a reconstruction of the GaAsl31 l)B surface, 2 X 1 . The basic unit cell as outlined in Fig. I (b) is 8 X 13.3 A'. The period along [011] is 8 A, double the face- centered ideal substrate unit, although RHEED shows a dif- fusive 1 X pattern. 2 X 1 is the smallest unit cell for GaAs(3 11) located between (100) and (1 11). concerning the atomic dimerization of GaAs(100) surfaces and the energy

01 63-1 82912002/66(19)11933 13(4)/$20.00 66 193313-1 02002 The American Physical Soc~ety

CThE2

OSA Trends in Optics and Photonics Series, v 88, p 1422-1424,2003

Yongan Tang, Baolai Liang, kqiang Guo, and Gregory J. Salamo Department of Physics, University of Arkansas, Fayeticville, Arkansas 72701

Phone (501) 575-5931; F.4X: (501) 575-4580 ; Email: salan~o~.uarlicdi~

Mordechai Segev, , Physics Department, Tecbnion - Israel Institute of technology. Haifa 32000,

Israel, and Electrical Engineering departmen& Princeton University, Princeton, NJ 08544

Abstract: We demonstrate experimentally that information can be transferred from one spatial soliton to other solitons by multicollisions, The results of these experiments can be used for vector algebra. 02002 Optical Society of America OCLS codes: (190.5330 ) Photorefractive nonlinear optics; (230.7380) Spatial soliton

l, a2B(x.z) 2 i- + --- + ( l ~ ( ~ , ~ ) 1 + ~ p ~ x , z ~ ~ 2 ~ ~ ( x , z l = 0 az ax2

background beam is perpendicular to the c-axis of the crystal. The intensity ratio of the soliton b beam is about 2.5. The image of the input and output face of the crystal is captured by a CCD camera.

--a - -

APPLIED PHYSICS LEITERS VOLUME 81, NUMBER 16 14 OCTOBER ZOO2 '

GaAs(311) templates for niolecular beam epitaxy growth: surface morphologies and reconstruction

Z. M. ~ a n g , ~ ) V. R. Yazdanpanah, J. L. Shultz, and G. J. Salamo Physics Department, UniversiQ of Arkansas, Fayetteville, Arizona 72701

(Received 8 July 2002; accepted 22 August 2002)

Morphologies of GaAs(3 11) surfaces grown by molecular beam epitaxy were investigated by in situ reflection high-energy electron diffraction and scanning tunnelling microscope. In addition to the (8 X I ) reconstruction, two surface phases, GaAs(3 1 1)A-(4 X I ) and GaAs(3 1 l)B-(2 X 1 ) were observed. Both of these surfaces are characterized by wider, atomically smooth terraces with much lower structural anisotropy, when compared to the (8 X 1 ) reconstructed GaAs(3 11) surfaces. The observed surfaces have potential as templates for the growth of organized quantum dots, wires, and wells. 0 2002 American Institute of Physics. [DOI: 10.106311.15 148221

The high index GaAs(311) surface has been intensively studied over the last twenty years. Depending on the surface termination, the GaAs(311) surface is categorized into GaAs(3 1 l )A and GaAs(3 1 l)B types. The type A surface contains twofold coordinated (100)-like As atoms and threefold coordinated (111)A-like Ga atoms while the type B surface contains twofold coordinated (1 00)-like Ga atoms and threefold coordinated (1 11)B-like As atoms. The persistent interest in GaAs(311) originates for the following reasons. First, it has been reported that the conductivity of Si-doped GaAs(3 1 l)A can be controlled by changing growth temperatures and Ga/As flux ratios while maintaining high crystal q ~ a l i t y . ' . ~ This characteristic promises physical phenomena and exciting optoelectronic device applications. Second, the observed surface corrugation on GaAs(31 l)A offers considerable potential for the fabrication of quantum wires.3 Third, both the GaAs(3 1 l)A and GaAs(3 1 l)B surfaces have been used as templates for growth of uniform quantum dot arrays4.' and for engineering specific quantum dot shapes6 Fourth, self-assembled three-dimensional island formation is strain-driven and is currently the leading method for semiconductor nanostructure fabrication. The (31 1) surface was the first observed as a side wall of the self-assembled i~lands. ' .~

Despite being the subject of considerable study, the exact experimental determination of the surface morphology of GaAs(3 1 l )A or GaAs(3 1 l)B is still controversial. For GaAs(311)A, the first surface corrugation model was proposed with a lateral periodicity of 3.2 nm and a corrugation depth of 1.02 nm.339 The lateral periodicity, later confirmed by scanning tunnelling microscope (STM) results, is called ( 8 X 1 ) reconstruction. The depth of modulation, however, has been the subject of controversy and is argued to be 0.34 nm.lo There is even conflicting evidence that reports the GaAs(311)A surface to be as smooth as, if not smoother than, the GaAs(100) surface.11212 Meanwhile the GaAs(311)B was also reported to be unstable and faceted with low index surfaces.13 In addition, recently an (8 x 1) reconstruction, similar to that of GaAs(3 11)A, has also been

"Elcctronic mail: [email protected]

observed.14915 In general, the observed spectrum of' behaviors of the GaAs(311) surfaces may even suggest the possibility of other reconstructions of GaAs(31 I), which together with known surfaces, would play an important role in determining surface behavior. In fact, in this letter, we report the observation of surface morphologies of GaAs(31 I ) . We have observed a (4 X 1) reconstruction of the GaAs(3 I l )A surface and a (2 X 1 ) reconstruction of the GaAs(3 1 l )B surface. Both surfaces are characterized by wide atomically flat terraces. Based on the characteristics of these sul.t:Jces, they may act as good templates for the growth of quantum dots, wires, and wells.

While taken together, these previous reports 011 the mor- '

phology of GaAs(311) may at first appear inconsistent, we believe that the work reported here clarifies these observations.

The GaAs(311) experiments reported here were performed in a combined molecular beam epitaxy (MBE)-STM UHV system equipped with an optical system th;r ~r~nr:!ors the substrate band edge to give accurate growth temperatures. Epitaxial ready n-type GaAs(3 11) wafers, p~lished on both sides, were used for comparable studies on both the A and B surfaces. MBE growth under various growth conditions was monitored in situ by reflection high-energy electron diffraction (RHEED). The (8X 1) RHEED pattern is observed on both the GaAs(31 l)A and GaAs(3 1 1 ) ~ surfaces by introducing growth interruption at various substrate temperatures and As pressures. The temperature and As pressure window needed to stabilize the GaAs(31 l)A surface reconstruction was larger than that for GaAs(311)13.~'' The ( 8 X 1) reconstructed surface was quenched by simultaneously reducing the substrate temperature and As pressure. The . samples were subsequently transferred through a 0-V transfer chamber into the STM chamber. Constant curcent STM images were obtained using a tunnelling current of around 0.1 nA and a sample bias of - 3.0 V. Typical STM images of the (8 X 1 ) reconstructed GaAs(3 1 l)B surface are 'shown in .. Fig. 1. The two most important characteristlcs 01 :;iiksurface .

are the high density of straight steps running alorlg [233], as shown in Fig. l(a), and the uniform height corrugation re- s i t ing from the two monolayer zigzag As d im~r s along. [233] with a lateral periodicity of 3.2 nm, as sho\vn in Fig.

0003-6951 12002181 ( I 6)12965131$19.00 2965 63 2002 American Institute of Physics Downloaded 03 Nov 2008 to 130.184.237.6. Redistribution subject to AIP license or copyright; see http:llapl.aip.orglapllcopyright.jsp

Fixed three-dimensional holographic images

Clint Wood, Gregory J. Salamo, John Goff, Gary L. Wood, Richard J. Anderson, and David J. McGee

Three-dimensional holograms were recorded in a cerium-doped, strontium barium niobate (SBN:75) photorefractive crystal. These holograms are shown to not degrade after more than one week of continuous readout and to reconstruct reproductions of the original object with an observable field of view of approximately 35". 0 2002 Optical Society of America

OCZS codes: 090.0090, 100.0100, 190.0190.

1. Introduction

In the current era of high-speed computers, the need for an ultrahigh-speed, large-capacity data storage system becomes greater than ever. Holographic storage in general and photorefiactive crystals in particular offer the potential to meet this need. Pho- torefractive crystals have the unique property of al- lowing optical interference patterns to redistribute charge among traps, which then generate local space- charge fields that replicate the original optical interference patterns.1-4 The resulting space-charge field induces an index of refraction grating that can be used to reconstruct a holographic image of the original object. Theoretical predictions indicate that a crystal's storage capacity is of the order of -v/k3 bits, where V is the volume of the crystal and A is the wavelength of light.5 For a typical crystal volume of 1 cm3 and green light, this prediction is of the order of eight trillion bits or approximately 4 orders-of-magnitude times larger than the storage capacity of a typical compact disk.2 Not only does this technology promise large storage capacity, recent results demonstrate that a randomly chosen data element could be accessed in approximately 100 ps, a figure expected to decrease to around 10 ps in the

C. Wood and G. J . Salamo ([email protected]) are with the Department of Physics, University of Arkansas, Fayetteville, Ar- kansas 72701. J . Goff and G. L. Wood are with the U.S. Army Research Laboratory, 2800 Powder Mill Road, Adelphi, Maryland 20783-1197. R. J . Anderson is with the National Science Foun- dation, 4201 Wilson Boulevard, Arlington, Virginia 22230. D. J . McGee is with the Department of Physics, Drew University, Mad- ison, New Jerwy 07940.

Received 17 October 2001; revised mailuscript received 22 July 2002.

0003-6935/02/326796-06$15.00/0 0 2002 Optical Society of America

foreseeable future.6 That figure is also several orders-of-magnitude faster than conventional magnetic disk drives. These two attributes of photorefractive crystals offer substantial advantages over conventional devices and make photorefractive holographic storage an attractive alternative for next- generation computing.

In addition to computing applications, photorefractive crystals also offer the possibility of storing a library of three-dimensional (3-D) holographic images in a relatively small volume.7-14 Of the many 3-D image display techniques, holography provides one of the most aesthetically pleasing (possibly because of the different perspectives observed as the viewer moves from side to side) 3-D images for the human eye-brain system. As a result, applications of holography to cinematography, artificial intelli- gence, and security have long been the pursuit of scientists and engineers. Recently, Ketchel et al.15 reported storing 3-D holograms in cerium-doped, strontium barium niobate (SBN:60). The experiment demonstrated that high-resolution 3-D objects could be stored and retrieved in real time, meaning that no processing is required to view the image. The stored 3-D image was characterized by a large depth of field, high resolution, and a wide field of view of -35". To take advantage of the large storage capacity of these crystals, Ketchel e t al.15 used angle multiplexing of images to store several 3-D color holograms in one crystal. In this technique one image is stored, and the crystal is rotated by an angle of 0.082" to allow another distinct image to be written. By use of angle, wavelength, and other multiplexing techniques, many 3-D images can be written into a single crystal. This ability to store large numbers of high-quality 3-D images allows one to imagine a wide area of applications for photorefiactive holography

6796 APPLIED OPTICS / Vol. 41, NO. 32 / 10 November 2002

intervalley up-transfer for electrons in type-11 gaas ... · index term-charge carrier processes,...

Documents