title artificial control of zno nanodots by ion-beam ... · pdf filemasaya ueda, mitsuru...

Title Artificial control of ZnO nanodots by ion-beam nanopatterning

Author(s) Kim, Sang-Woo; Ueda, Masaya; Funato, Mitsuru; Fujita,Shigeo; Fujita, Shizuo

Citation JOURNAL OF APPLIED PHYSICS (2005), 97(10)

Issue Date 2005-05-15

URL http://hdl.handle.net/2433/39713

Right

Copyright 2005 American Institute of Physics. This article maybe downloaded for personal use only. Any other use requiresprior permission of the author and the American Institute ofPhysics.

Type Journal Article

Textversion publisher

Kyoto University

Artificial control of ZnO nanodots by ion-beam nanopatterningSang-Woo KimDepartment of Metallurgical Engineering, Kumoh National Institute of Technology, 1 Yangho-dong, Gumi,Gyeongbuk 730-701, Republic of Korea

Masaya Ueda, Mitsuru Funato, and Shigeo FujitaDepartment of Electronic Science and Engineering, Kyoto University, Katsura, Nishikyo-ku, Kyoto 615-8510, Japan

Shizuo Fujitaa!

International Innovation Center and Advanced Research Institute for Nanoscale Science and Technology,Kyoto University, Katsura, Nishikyo-ku, Kyoto 615-8510, Japan

sReceived 13 September 2004; accepted 7 March 2005; published online 5 May 2005d

The use of focused ion-beamsFIBd nanopatterning for manipulating self-assembled ZnO nanodotsis described. Highly aligned ZnO-nanodot arrays with various periodicitiesse.g., 750, 190, and 100nmd on FIB-nanopatterned SiO2/Si substrates were prepared by metal-organic chemical-vapordepositionsMOCVDd. The artificially assembled ZnO nanodots had an amorphous structure. Gaatoms incorporated into the surface areas of FIB-patterned nanoholes during FIB engraving werefound to play an important role in the artificial control of ZnO, resulting in the production of ZnOnanodots on the FIB-nanopatterned areas. The nanodots evolved into single-crystalline dot clustersand rods with increasing MOCVD-growth time. In addition, microphotoluminescencemeasurements showed that the ZnO-nanodot arrays have low-dimensional quantumcharacteristics. ©2005 American Institute of Physics. fDOI: 10.1063/1.1898446g

I. INTRODUCTION

Self-assembled ZnO nanostructures such as nanodots,1

nanowires,2 nanorods,3 nanobelts,4 and nanotubes5 have at-tracted a great deal of attention for use in various applica-tions including nanoscale optical devices, field emitters,nanoscale circuits, and nanosensors. In addition, a specialinterest exists in the optoelectronic research community, asthe result of the promising optical properties of ZnOse.g., alarge exciton binding energy of 60 meV and biexciton bind-ing energy of 15 meVd.6 However, in spite of a number ofreports on the fabrication of self-assembled ZnO nanostruc-tures, one of the major obstacles to achieving ZnO-basednanostructure-building blocks is the ability to control ZnOnanostructures with a desired size and forming position inthe synthesis process. In this regard, some groups have at-tempted to manipulate the position and size of ZnOnanostructures.7,8 However, only a few systematic reportshave appeared with detailed information relative to control-ling the dimension and chirality of nanostructures by synthe-sis methods. Recently, our group reported on the control ofthe position and size of ZnO nanodots using metal-organicchemical-vapor depositionsMOCVDd in conjunction withnanopatterning by focused ion beamsFIBd. However, themechanism of formation of artificially controlled ZnO nan-odots is currently unclear. Information on the mechanism forthe selective formation of ZnO nanodots on FIB-engravednanohole arrays would attract a great deal of attention inother interdisciplinary fields of nanotechnologies as well asthe ZnO-related research community. We report herein on thepreparation of position- and size-manipulated ZnO nanodots,

in which a systematic study of the formation mechanism andthe characterization of ZnO nanodots was carried out.

II. EXPERIMENT

In order to fabricate ZnO nanodots with a controlledmorphology and dimension, we introduced a combination ofa “bottom-up”–“top-down” approach based on MOCVD andFIB sSEIKO SMI-2050d-functionalized substrates.7,9 Thenanopatterned substrates for the growth of manipulated ZnOnanodots were prepared by the FIB nanoengraving ofSiO2/Si substrates. Typical flow rates of source materials forthe MOCVD growth of ZnO, diethylzincsDEZnd and N2O,were 1 and 7000mmol/min, respectively, where the pressurein the MOCVD reactor was maintained at 200 Torr. Thesubstrate temperatures were in the range of 500–700 °C so asto study the effect of temperature on the selective growth ofZnO nanodots in nanohole arrays on the substrates preparedby FIB.

The crystal structure of the artificially controlled ZnOnanodots on SiO2/Si substrates was analyzed by transmis-sion electron microscopysTEM, HITACHI H-9000NAR,300-kV accelerating voltaged. The formation of not only nan-odots but also a combination of clusters and rods were con-firmed by atomic force microscopysAFM, SEIKO SPA-400dand field-emission scanning electron microscopysFE-SEM,JEOL JSM-7400d. The composition of the selectively grownZnO nanodots in nanopatterned arrays was determined bymicroenergy-dispersive x-ray spectroscopysm-EDXd linkedwith a TEM instrumentsHITACHI HF-2000, 200-kV accel-erating voltaged, in addition with SEM-EDX sattached toJEOL JSM-6500Fd measurements. The optical properties ofthe ZnO nanodots were characterized by cathodolumines-adElectronic mail: [email protected]

JOURNAL OF APPLIED PHYSICS97, 104316s2005d

0021-8979/2005/97~10!/104316/9/$22.50 © 2005 American Institute of Physics97, 104316-1

Downloaded 31 May 2007 to 130.54.110.22. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

http://dx.doi.org/10.1063/1.1898446

http://dx.doi.org/10.1063/1.1898446

cencesCLd measurements at room temperaturesRTd with asystem equipped with a FE-SEM instrumentsJEOL JSM-6500Fd and microphotoluminescencesm-PLd measurementsperformed by fluorescent microscopysNIKON Y-FL d usingthe 325-nm line of a He–Cd lasers20 mWd as the excitationsource.

III. RESULTS AND DISCUSSION

A. FIB Nanopatterning

In order to fabricate spatially regular arrays of nano-structures, appropriately shaped patterns were introducedinto the substrate followed by MOCVD growth of the ZnOinto the patterns. Recently, nontraditional nanofabricationtechnologies such as FIB lithography,10 near-field opticallithography,11 and contact printing techniques12 have beenactively developed. Of these, FIB, utilizing a direct sputter-ing process, has many unique advantages over other similartypes of techniques for nanofabrication:s1d maskless rapidprototyping,s2d its sub- 10-nm spot size enables the pattern-ing of diverse nanostructures at very high resolution, ands3dallowing the production of high-resolution patterns withoutrefocusing or changing the sample height.

In this work, SiO2s50 nmd /Si substrates were patternedin a FIB chamber to create two-dimensionals2Dd arrays ofnanoholes. A Ga+ liquid-metal ion source, in which the beamenergy was fixed at 30 keV, with an ion dose of 131016 cm−2 was used. The ion-beam size is a function of thebeam current as programmed FIB properties. The beam cur-rents were varied from 1 pA to 1.3 nA in order to control thedimensions of the nanoholes engraved by FIB. Details con-cerning the patterning modes of the FIB system for high-speed nanoengraving over a wide area have been describedelsewhere.13

Numerous sputtering conditions were examined to real-ize the desired nanohole profiles. Figure 1 shows AFM im-ages of 2D nanohole arrays patterned under a programmingcommand of fabricating a 6-nm constant depth with varyingthe beam sizes of 13sad, 23 sbd, and 100 nmscd. The beam

diameter directly corresponded to that of the nanohole at thetop surface. On the other hand, the sputtering time requiredfor achieving the same sputtering depth among the three pat-terns was different. A large beam size leads to a short pro-cessing time at the expense of increase in the individualholes. Using optimized FIB-sputtering conditions we suc-ceeded in forming 2D nanohole arrays with various period-icities se.g., 750, 190, and 100 nmd over a wide area.

B. Selective formation of ZnO nanodots

Figure 2 shows AFM images of highly aligned 2D ZnO-nanodot arrays formed in FIB-engraved nanoholesspattern-ing depth: 6 nmd with different periodicities at a MOCVD-growth temperature of 700 °C for 40 s. From the AFM studyshown in Fig. 2 the average heightsf9 sad, 5 sbd, and 4 nmscdg and average widthsf130 sad, 40 sbd, and 25 nmscdg ofindividual nanodots selectively grown on each nanohole ar-ray are closely related to the dimensions of the nanoholes,indicating that a nanohole with a larger dimension accommo-dates a larger ZnO nanodot. The perfect one-by-one accom-modation of a ZnO nanodot in a nanohole was realized in thesamples of 2D nanohole arrays with 750- and 190-nm peri-odicities, while the packing percentage of ZnO nanodots inthe nanoholes with a 100-nm periodicity was about 64% inaddition to a large distribution in their widthsfFig. 2sddg. Thelow packing percentage is due to the relatively high growthtemperature of 700 °C, resulting in an increase in the atomicdesorption of reacting species from the surface of the pattern.Very recently, the selective formation of ZnO nanodots witha packing percentage higher than 75% into the nanoholesshown in Fig. 2scd was realized, using more optimizedgrowth conditions, which will be reported elsewhere.

In order to study the characteristics of the selectivegrowth and evolution of ZnO nanodots as a function ofMOCVD-growth time, we varied the growth time from 40 sto 600 and 900 s. Figure 3 shows that, when the MOCVD-growth time is increased, the nanodots are converted to clus-tersf600 s, Fig. 3sbdg and rodsf900 s, Figs. 3scd and 3sddg onFIB-nanopatterned holes. An increase in the average dot

FIG. 1. Schematic image of the FIB sputtering of a SiO2/Si substrate with30-keV Ga+ ions. AFM imagessb, c, and dd of 2D nanopattern arrays en-graved by FIB are shown here. Nanopatterns were produced under a pro-gramming command of fabricating a constant depth of 6 nm, by varying thebeam diametersbeam currentd of 13 sb, 1 pAd, 23 sc, 48 pAd, and 100 nmsd,1.3 nAd.

FIG. 2. Plan-view AFM images of 2D ZnO-nanodot arrays with differentperiodicities.sad 750, sbd 190, andscd 100 nm. Distribution of the nanodotsshown insad and scd in their widths is presented atsdd.

104316-2 Kim et al. J. Appl. Phys. 97, 104316 ~2005!


width was linear with respect to growth time, while the av-erage dot height was less sensitive to growth time. Whensufficient growth timesover 600 sd was used, the averagevolume sFig. 4d of the grown ZnO nanostructures increasedsuperlinearly with the increment of growth time, resulting inthe formation of rod structures with random growth direc-tions. The superlinear increase in volume can be attributed toa dramatic increase in the height of the ZnO nanostructures,leading to the dots being converted to rods with high aspectratios. This indicates that FIB-patterned nanoholes act asseeds for the growth of ZnO nanodots, clusters, and rods as afunction of growth time. It should be noted, however, that,contrary to the one-by-one accommodation of nanodots innanoholes, 3–6 clustersslarge-sized dotsd and rods are ac-commodated in a nanohole, as shown in Figs. 3sbd–3sdd. Themechanism for the selective formation of nanodots, clusters,and rods on nanopatterned holes will be described in detail inSec. III C.

Figure 5 shows a plan-view SEM image of ZnO clustersselectively grown in nanoholes for 600 s and a RT-CL imagetaken at a wavelength of 374 nm in addition to the measuredCL spectrum. The strong emission band at 374 nm in thescanned spectrum taken with the CL system indicates that theorigin of the emission can be attributed to the near band-edgeemission of ZnO considering its band gap of 3.37 eV. Theresulting monochromatic CL image taken using the 374-nm

luminescence peak represents an excellent map of the lumi-nescence from the area designated by a box in the SEMimage, thus permitting the formation of stoichiometric ZnOclusters to be visualized.

C. Formation mechanism of ZnO nanodots on FIB-patterned nanoholes

The bottom and sidewall of the sputtered nanohole arealso SiO2 with amorphous structural properties because thethickness of the thermally grown SiO2 layer on Si substratesis 50 nm and the depth sputtered by FIB is 6 nm in allsamples used in this work. Thus, it can be concluded that achange in the surface structure, such as the generation ofsurface atomic steps and kinks by energetic Ga+ ions, is notrelated to the selective formation of ZnO nanodots on theFIB-sputtered SiO2 areas.

Our previous study suggested the possibility that Ga at-oms are incorporated into patterns during the FIBengraving.9 In order to investigate the effect of Ga atomsincorporated into patterns from the ion beam for nucleationof ZnO, we carried out the growth of ZnO using a substratessubstrate 1d which is different from the previous onessub-strate 2d shown in Fig. 1. A schematic image for comparisonbetween the two types of substrates is shown in Fig. 6 andsurface AFM images for substrate 1 observed at each step areshown in Fig. 7. For substrate 1, after preparing FIB-engraved 2D nanohole arrays on a Si substratefFig. 7sadg, wecarried out thermal oxidation of it in a furnace at 1000 °C for10 min in an oxygen atmosphere to allow the incorporatedGa atoms in the patterned holes to form GaxO1−x. In order toexclude the effect of Ga on ZnO nucleation, the thermallyformed GaxO1−x was lifted off, and the thermally grown SiO2

layer eliminated using a diluted HF solutionfFig. 7sbdg.SEM-EDX measurements confirmed that Ga atoms that wereincorporated during the FIB nanopatterning were completelyeliminated by the lift-off process. The reoxidation of the HF-treated substrate with an oxide thickness of 50 nms1000 °Cfor 40 mind was then carried out to achieve a SiO2/Si sub-strate with 2D nanohole arrays, with a 750-nm periodicity

FIG. 3. FE-SEM images of selectively grown ZnO nanodotssad, clusterssbd, and rodsscd. The image shown insdd is a magnified image ofscd. Thegrowth time for the formation of nanodotssad, clusterssbd, and rodsscd is40, 600, and 900 s, respectively. Although some rods fall down on thesubstrate surface, all are selectively well grown on the 2D nanohole arrays.

FIG. 4. Volume evolution of grown ZnOsfrom nanodots to clusters androdsd as a function of MOCVD-growth time.

FIG. 5. sad FE-SEM image of well-aligned ZnO clusters grown for 600 s.sbd CL spectrum of the FE-SEM imagesad measured at RT. The inset showsRT-CL image taken at the wavelength of 374 nm from the area designated asa dotted square in the FE-SEM imagesad.



fFig. 7scd, the same as the substrate shown in Fig. 2sadg.After the reoxidized substrate was transferred to theMOCVD reactor, Zn and O species were supplied for ZnOgrowth. The feature of ZnO grown under the same growthconditions, as were used for the ZnO nanodots on the reoxi-dized substrate presented in Fig. 2sad, is shown in Fig. 7sdd.A number of smaller nanodots compared with those in Fig.2sad falso shown at Fig. 7sfdg are formed into nanoholes aswell as on the planar SiO2 surface between the nanoholesfFig. 7sedg, indicating that the nanoholes did not act as arti-ficial traps for migrating Zn and O adatoms and manipulatednucleation seeds for ZnO any longer. The random formationof these nanodots in their positions irrespective of the exis-

tence of nanoholes is similar to that of self-assembled ZnOnanodots on planar SiO2/Si substrates, as reported in ourprevious studies.1,14

Figure 8 shows cross-sectional bright-field TEM imagesof selectively grown ZnO nanodotsfthe sample shown in theFig. 3sadg on the nanoholes. No significant crystalline latticeswere observed in the ZnO dot area, indicating that the struc-ture of the ZnO grown on the nanopattern is amorphous. Theamorphous structure of the ZnO nanodot was also confirmedfrom the selective area transmission electron-diffractionsTEDd pattern shown in the inset of Fig. 8sbd. In order toclarify the mechanism for the selective formation of ZnOnanodots on nanoholes, we carried outm-EDX measure-ments at points corresponding to the cross sections of thesample. Them-EDX spectra were recorded at different posi-tions of the nanodot and the substrate area for compositionanalyses. The spectrum shown in Fig. 9sad was obtained

FIG. 6. Schematic image for comparison between two kinds of substratesssubstrates 1 and 2d patterned by FIB for ZnO growth to clarify mechanismon the selective formation of ZnO nanodots on 2D nanoholes.

FIG. 7. Plan-view AFM imagesfthe image scale ofsad–sdd and sfd is 333 mm2 and that ofsed is 6003600 nm2g of sad FIB-engraved 2D nanoholearrays on a Si substrate,sbd chemicallysa dilute HF solutiond treated sub-strate after a thermal oxidation at 1000 °C for 10 min,scd reoxidation of theHF-treated substrate at 1000 °C for 40 min,sdd randomly grown ZnO nan-odots onscd, sed a nearby area of a single nanohole in the samplesdd, andsfdwell-aligned ZnO nanodots selectively grown on the substratesad, which isalso shown in Fig. 2sad.

FIG. 8. Side-view bright-field TEM images of selectively grown ZnO nan-odots on nanoholes shown in Fig. 3sad. sad TEM image of nanodots grownon three arrayed nanoholes.sbd TEM image of a ZnO nanodot on a nano-hole. The inset ofsbd is selective area TED pattern of the ZnO nanodot withan amorphous-crystalline structure grown on the nanohole. The scale bar inthe imagesbd indicates 50 nm.

FIG. 9. sad andsbd showm-EDX spectra recorded at the points marked as*1and*2 in Fig. 8sbd, respectively.



from the planar unpatterned SiO2 surface between nanoholesfmarked as “*1” in Fig. 8sbdg, which clearly shows the char-acteristic peaks corresponding to Zn and Ga as well as Si andO. The intensity of the Zn- and Ga-related peaks is compa-rable, but is quite weak. On the other hand, the intensity ofthe Ga peak is much stronger than that of the Zn peak in thespectrum shown in Fig. 9sbd, which was measured at theinterface areafmarked as “*2” in Fig. 8sbdg between the dotand the surface of the nanohole. This result is reasonable,considering the Gaussian profile of the ion beam.15 The in-corporation of Ga atoms in area ‘*1’ might be due to ion-beam broadening caused by the surface charging of the insu-lating SiO2 surface. The dark areafdenoted as “*3” in Fig.8sbdg on the Si surface below the thermally grown SiO2 layeris a damage-induced area, produced by energetic Ga+ ionsduring the FIB nanoengraving. Ga and Cu atoms as well asmajor Si and O atoms were detected from the clusters in theSiO2. The formation of clustersf“ *4” area in Fig. 8sbdg com-posed of Ga and Cu atoms might be due to the incorporationof Ga atoms during FIB patterning and Cu diffusion from theSi to the SiO2 layer during the MOCVD-growth temperatureat 700 °C. Our simulation showed that Ga+ ions that are ac-celerated by 30 keV can penetrate to depths of more than 60nm and that the maximum Ga concentration is at around 30nm from the SiO2 surface. From these findings and the SiO2

thickness of 50 nm, it can be concluded that the damagedareas*3d is induced by Ga+ ions and that the clusters consist-ing of Ga and Cus*4d are formed at the depth where the Gaconcentration is at a maximum.

From the above discussion, in conjunction with Figs.6–9, we propose the mechanism for the evolution of ZnOnanodots to clusters and rods on FIB-patterned nanoholes asa function of MOCVD-growth time. The schematic imagesare presented in Fig. 10. A number of Ga atoms are incorpo-rated into the SiO2 layer during FIB nanoengraving, eventhough Si and O atoms are simultaneously etched out byenergetic Ga+ ions. The Ga atoms that are incorporated in thepatterned nanohole were molten before reaching the ZnO-growth temperature of 700 °C in this study because the melt-ing point of Ga is about 30 °C. At the early stage of MOCVDgrowth, Zn atoms that decompose from the DEZn are ab-sorbed into the molten Ga atoms, which are mainly anchoredon the nanoholes, forming a solid solution of Zn–Ga. Oncethe Zn–Ga solid solution is supersaturated, the continuoussupply of Zn and O species results in the formation of amor-phous ZnO nanodots on the supersaturated surface16 of theGa-incorporated area into the nanohole. Therefore, we con-clude that the amorphous ZnO nanodot with a Zn-abundantphase, as confirmed by them-EDX measurements, directlygrew out from the supersaturated surface. As mentionedabove, Zn atoms were also detected on the planar unpat-terned SiO2 surface between the nanoholes. However, noZnO-related emission by selective excitation of the surfacearea, denoted as ‘*1’, was observed in the RT-CL measure-ments, while a ZnO band-edge emission was clearly ob-served from a single dot on a nanohole.13 This indicates thatonly a small number of Zn and O adatoms were incorporatedinto area ‘*1’. However, spontaneous dot nucleation followedby growth and coarsening only occurs in area ‘*2’ of the

nanohole. From them-EDX spectra in Fig. 9, the quantity ofincorporated Ga atoms in the ‘*1’ area is clearly muchsmaller than that in area ‘*2’ in the nanohole, indicating thatsufficient numbers of Ga atoms to act as a catalyst are re-quired for effective ZnO nucleation.

The formation of amorphous ZnO nanodots may cause acatalyst-poisoning effect,17 inactivating the interface and cut-ting off the Zn and O supply to the supersaturated surface.Some large dots, as shown in Fig. 3sad, are also formed ir-regularly on the substrate in addition to uniform small nan-odots. From a careful observation of the SEM image in Fig.3sad, the large dots are formed not on the central area of thenanohole surface but in the vicinity of the nanoholes. TheTEM observation of a large dotfFig. 11sbdg also clearlyshows that a large dot with a single-crystalline structure isformed in proximity to an amorphous ZnO dot grown on thecentral area of a nanohole surface. The catalyst-poisoningeffect causes the Zn and O supply beneath the amorphousZnO to be cut off. Thus, the source supply is concentratedaround the amorphous ZnO nanodot, and new ZnO precipi-tates with a single-crystalline structure are formed in closeproximity to the amorphous ZnO nanodotfFig. 11sadg. Theabsorption and diffusion of Zn and O can be enhanced at thesurfaces of the ZnO precipitates during growth.18 Furthersuccessive supply of Zn and O species results in the coars-ening of the ZnO precipitates rather than amorphous ZnOnanodots.

FIG. 10. Schematic images showing the evolution of ZnO nanodots intoclusters and rods in nanoholes, as a function of MOCVD-growth time.sadGa atoms are incorporated with a Gaussian profile into the patterned SiO2

layer. The incorporation of Ga atoms is propagated up to the damaged areassdark areasd on the Si surface below the thermally grown SiO2 layer. Simu-lation results showed that the maximum propagation depth of Ga atomsthrough SiO2 is about 60 nm in the patterning condition of this study.sbdThe incorporated Ga atoms melt before the main MOCVD-growth tempera-ture at 700 °C. Clusters into the SiO2 layer are composed of Ga and Cuatoms.scd Formation of an amorphous ZnO nanodot on the supersaturatedsurface of the Ga-incorporated area into a nanohole.sdd ZnO precipitateswith a single-crystalline structure are formed on the close vicinity of theamorphous ZnO nanodot due to catalyst-poisoning effects.sed The ZnOprecipitates rather than the amorphous ZnO nanodots are enlarged by suc-cessive supply of Zn and O species, resulting in large ZnO dots.sfd Thelarge ZnO dots lead to the formation of ZnO rod structures through suffi-cient growth.



Although the number of the ZnO precipitatesfFig. 11sadgand large dotsfFig. 11sbdg is small, these structures coexistwith uniform amorphous ZnO nanodotsfFig. 8sbdg on thesame sample. The cause of this is unclear at present, butmight be due to different critical evolution times that dependon the degree of the catalyst-poisoning effect. Ga atoms in-corporated by FIB result in the selective growth of ZnO. Atthe same time, however, Ga atoms prevent ZnO nuclei fromcrystallizing, which results in the formation of amorphousZnO nanodots that serve as the origin of the catalyst-poisoning effect. We conclude that the number of Ga atomsvaries from position to position due to local ion-beam fluc-tuations during the FIB patterning, causing changes in thedegree of the catalyst-poisoning effect. Figure 12 clearlyshows that self-assembled ZnO nanodots on planar unpat-terned SiO2/Si substrates have a single-crystalline structurewith hexagonal atomic arrays without any types of defects inspite of the amorphous nature of SiO2.

1,19 From the experi-

ment results, it could be confirmed that the formation ofartificially controlled ZnO nanodots with an amorphousstructure is due to the Ga atoms incorporated during the FIBengraving process.

As a consequence, Zn and O species supplied duringMOCVD growth for ZnO rods are concentrated beneath theprecipitates. The FE-SEM image of Fig. 3scd shows that onlya small number of ZnO rods appeared to grow perpendicularto the surface of the nanohole. Moreover, the lengths of therods are much shorter than those of the rods inclined fromthe surface normal, which serves as additional evidence for acatalyst-poisoning effect.

D. Low-dimensional quantum characteristics ofselective grown ZnO nanodots

We previously reported on quantum confinement effectsfrom self-assembled ZnO nanodots by MOCVD on SiO2/Sisubstrates.1 However, since the spatial distribution of thenanodots fabricated by self-assembly is still random, it isdifficult to achieve desirable characteristics for ZnOnanostructure-based devices. In this regard, the realization oflow-dimensional ZnO nanodots well manipulated in size andpositions is strongly desirable. Them-PL measurements wereperformed to confirm low-dimensional quantum characteris-tics of the selectively grown ZnO nanodots. Thetemperature-dependent PL measurements were carried outfrom 4.2 to 300 K in a cryostat. Approximately 100 nanodotswith a uniform sizefshown in Fig. 2sadg arranged in thediameters7.5 mmd of the focused laser spot were excited.

The m-PL spectra are presented in Fig. 13. At 4.2 K, themain peak position of the spectrum with a broad full width athalf maximumsFWHMd of about 80 meV is around 3.380eV. Usually, a free-exciton emissionsEXd from ZnO thinfilms on SiO2/Si, Si, and Al2O3 is located at about 3.377 eVat low temperatures10 Kd,1,20,21 and the energy position ofthe near-band-edgesNBEd emission is barely related to thecrystal structure of ZnO. It has been reported that the PLspectrum of amorphous ZnO films shows a strong ultravioletemissionsNBE emissiond.22 Therefore, it is reasonable toconclude that the emission from 100 nanodots in this study is

FIG. 11. Cross-sectional bright-field TEM images as supporting evidencefor the process shown in Figs. 10sdd and 10sed. sad Formation of a ZnOprecipitate on the close vicinity of the amorphous ZnO nanodot.sbd A largeZnO dot is formed on the close vicinity of the amorphous ZnO nanodotgrown on the center of a nanohole surface because the absorption and dif-fusion of Zn and O are enhanced at the surfaces of the ZnO precipitateduring growth. The inset is selective area TED-pattern image of the largeZnO dot, which clearly shows that the large dot has a single-crystallinestructure.

FIG. 12. sad Plan-view TEM image of self-assembled ZnO nanodots on aSiO2/Si substratesdot density: 1.131011 cm−2d by MOCVD. The widths ofthe individual nanodots range from 5 to 15 nmsan average height of thenanodots is about 7 nmd. sbd High-resolution TEM image of a self-assembled ZnO nanodot with a single-crystalline structure showing hexago-nal atomic arrays.

FIG. 13. Temperature-dependentm-PL spectra of 100 nanodots in the ZnO-nanodot sample presented at Fig. 2sad.



shifted to the higher-energy side. Moreover, the emission en-ergy was independent of temperature below 80 K. Sharppeaks that were not shifted with temperature originated fromthe emission of the He–Cd laser, which is frequently ob-served when the intensity of the PL emission from a sampleis weak and which is a good reference for calibrating PLenergy.

First, the cause of the observed blueshift is discussed.There are two possible explanations: one is the quantum sizeeffects from the nanodots,1,23 and the other is band-gap wid-ening by the Burstein–MosssBMd effect24 which could occurdue to the unintentional doping effect of Ga atoms into thenanodots during MOCVD growth. Calculations were per-formed concerning the quantum effect. The infinite potentialbarrier was introduced to clarify the band-gap enhancementby the quantum size effect in the ZnO nanodots. The averageheight and average width of nanodots are 9 and 130 nm,respectively, as mentioned in Sec. III B, indicating that thedimension of the dots is much larger than the excitonic ZnOBohr radius of 1.8 nm. In this case, the center-of-mass mo-tion of excitons is quantized by the confinement potentialand the relative carrier motion is dominated by the Coulombinteractions. This is the so-called weak confinement.25 As-suming the Coulomb potential to be a small perturbation, theconfinement energies were calculated by a combination ofthe Kronig–Penny model adopting “a particle in a box” andthe Kayanuma model,26 including the Coulomb potential.The band-gap enhancement was estimated from the shift inthe EX band. The energyE of the EX band from the ZnOnanodots can be given by

E = EEX.ZnO+p2h2

2S 1

me* +

1

mh* DFSnx

LxD2

+ Sny

LyD2

+ Snz

LzD2G −

1.789e2

«ÎSnx

LxD2

+ Sny

LyD2

+ Snz

LzD2

− 0.248ERyd, s1d

where the exciton binding energy of 60 meV in bulk ZnOsthe Rydberg value,ERydd was employed and introduced intothe Kayanuma model. The effective masses of electrons andholes areme

* =0.24m0 and mh* =0.45m0, respectively.27 The

average dot dimension derived from AFM measurement,Lx

=Ly=130 nm andLz=9 nm, is substituted into the Eq.s1d. Inaddition, the third term can be neglected due to the fact thatits calculated value is very small. Since the dot structures inthis study are not capped, half of the dot surfaces are understress-free conditions. From this fact, the strain-inducedband-gap shift was not taken into account. Thus, the Eq.s1dcan be simply expressed as

E = EEX,ZnO+ CSnx2 + ny

2

1302 +nz

2

92D − 0.014 88 eV. s2d

Here,EEX,ZnO senergy position of the EX band in ZnO thinfilmsd and C sthe calculated confinement parameterd are3.377 and 2.40 eV, respectively. The calculated EX energy atthe lowest excited levelsnx=ny=nz=1d is 3.392 eV, whichagrees well with the experimental PL results shown in Fig.

13. Another possible explanation for the blueshift, that is, theBM shift is discussed later.

Concerning broad linewidths80 meV at 4.2 Kd, size dis-tribution within the laser spots7.5 mmd does not seem re-sponsible because only 100 nanodots were excited. Since theZnO nanodots in this work have nonburied structures and theemissions from them are only observed below 175 K, wespeculate that the broad emission is related to the generationof nonradiative recombination centers at the surface of theZnO nanodots. In addition, the asymmetry of the spectralshape with a long tail toward the lower-energy side of thespectrum might be interpreted as being due to the generationof surface defect levels.28

The anomalous PL temperature dependence mentionedin Fig. 13 can be mainly attributed to the thermal excitationof carriers among discrete energy levels.29 The width of thenanodots in this study was as large as 130 nm, resulting inquantum-well-like 2D quantum confinement. In this situa-tion, the energy gap between each level is very small, result-ing in the easy excitation of a carrier from lower-energy tohigher-energy levels by thermal energy. From the calculationof energy levels in the conduction band near the band gap ofthe ZnO nanodots using Eq.s2d, the findings show that thereare five discrete energy levels within 2 meV from the lowestexcitation-energy level, showing that the band-gap shrinkageby thermal energysestimated by Varshni equation30 to be 6.7meV up to 80 K, which is the temperature upper limit forshowing no peak shiftd can be compensated by the thermalexcitation of carriers in the conduction band. Regretfully, wewere not able to obtain more information from other sampleswith smaller dot dimensionsfshown in Figs. 2sbd and 2scdgdue to the very weak emission at this stage.

The higher-energy shift of the PL spectra in Fig. 13 canbe explained by the BM shift as another possibility, consid-ering the possible unintentional Ga doping into the nanodots.The Fermi level in the conduction band of the degeneratesemiconductor leads to a widening of the energy band, whichis related to carrier concentration,n, by

DE =h2

8m* S3n

pD2/3

. s3d

The electron effective mass is expressed asm* . Generally,however, the experimental results for the band-gap shift ver-sus the carrier concentration do not exactly follow Eq.s3d.Experimental results reported from some ZnO researchgroups have shown various exponent valuess2/3, 1/3, and3/5d31–33 for Eq. s3d. In heavily doped polar ZnO, the band-gap shift cannot be simply explained by only a BM shiftbecause the widening of the band gap caused by the BM shiftcompetes with the narrowing of the band gap by Coulombinteractions among the charged carriers and their scatteringagainst ionized impurities.

Assuming that the BM shift is also observed at low tem-peratures4.2 Kd, we estimated the carrier concentration inthe nanodots using the energy shift of the main peak ob-served in them-PL measurement at 4.2 KsFig. 13d comparedwith the main peak position of a reference sample. The ref-erence sample was a ZnO thin film on a SiO2/Si substrate,used in our previous report.1 The main peak position of the



D0X emission from the reference sample is 3.366 eV, whilethe main peak position from the 100 ZnO nanodots in them-PL measurement at 4 K is 3.380 eV. In order to achievethis energy blueshift of 14 meVs=3.380−3.366 eVd only bya BM shift, the carrier concentration estimated from the Eq.s3d should be 6.2331018 cm−3, which looks a possible dop-ing concentration. However, we did not observe a blueshiftin the m-PL measurements of selectively grown amorphous100 nanodots with a larger dimensionfaverage width andheight sAFM resultsd is 170 and 20 nm, respectivelyg com-pared with the nanodots used in Fig. 13, as shown in Fig. 14.Moreover, the neutral donor-bound excitonsD0Xd emissionlocated at 3.363 eV is sharp at 4.2 K and the peak shifttoward a lower-energy side with temperature increase isclearly observed, which is very similar to the optical behav-ior of ZnO thin films. The fact that a blueshift was not ob-served inm-PL measurements of nanodots with larger di-mensions, in spite of the same FIB-patterned substratesnamely, the same doing concentration of Ga atomsd, indi-cates that the blueshift is less related to the BM shift. Inaddition, regarding the temperature-dependent PL resultshown in Fig. 13, the peak position of the main peak ishardly shifted up to the measurement temperature of 80 K.This result clearly shows that the contribution of the BMshift to the band-gap widening is insignificant because theband gap of a sample with a high carrier concentration gen-erally shrinks with increasing temperature more drasticallythan that with a low carrier concentration.34 Thus, we canconclude that the blueshift of the PL spectra from the nan-odots is mainly due to low-dimensional quantum confine-ment in the ZnO nanodots.

IV. SUMMARY

The preparation of highly ordered ZnO-nanodot arrayswith the periodicities of 750, 190, and 100 nm on FIB-manipulated SiO2/Si substrates is described. A systematicstudy was performed, using TEM measurements to clarifythe structural properties of the selectively grown ZnO nan-odots. The findings indicate that the artificially assembled

ZnO nanodots have amorphous structural properties and thatGa atoms, incorporated during the FIB engraving, play animportant role in the artificial control of ZnO nanodots. Thesupply of Zn and O species results in the formation of amor-phous ZnO nanodots on a surface that is supersaturated withrespect to a Zn–Ga solid solution in the Ga-incorporated areainto the FIB-patterned nanohole. An increase in MOCVD-growth time permitted nanodots to be changed into clustersand rods. The lengths of the rods perpendicular to the nano-hole surface are much shorter than those of the rods inclinedfrom the surface normal, which clearly demonstrates thecatalyst-poisoning effect. Finally, fromm-PL measurements,we confirmed that the blueshift in the PL spectrum measuredat the low temperature of 4.2 K from ZnO-nanodot arrayscan be mainly attributed to the low-dimensional quantumcharacteristics of the ZnO nanodots.

ACKNOWLEDGMENTS

This work was partly supported by a grant-in-aid forscientific research and also by grants for regional science andtechnology promotion from the Ministry of Education, Cul-ture, Sports, Science, and Technology of Japan.

1S.-W. Kim, Sz. Fujita, and Sg. Fujita, Appl. Phys. Lett.81, 5036s2002d.2M. H. Huanget al., Science292, 1897s2001d.3W. I. Park, D. H. Kim, S.-W. Jung, and G.-C. Yi, Appl. Phys. Lett.80,4232 s2002d.

4Z. W. Pan, Z. R. Dai, and Z. L. Wang, Science291, 1947s2001d.5J. J. Wu, S. C. Liu, C. T. Wu, K. H. Chen, and L. C. Chen, Appl. Phys.Lett. 81, 1312s2002d.

6K. Hummer, Phys. Status Solidi B56, 249 s1973d.7S.-W. Kim, T. Kotani, M. Ueda, Sz. Fujita, and Sg. Fujita, Appl. Phys.Lett. 83, 3593s2003d.

8X. D. Wang, C. J. Summers, and Z. L. Wang, Nano Lett.4, 423 s2004d.9S.-W. Kim, M. Ueda, T. Kotani, Sz. Fujita, and Sg. Fujita, Jpn. J. Appl.Phys., Part 242, L568 s2003d.

10T. Aign et al., Phys. Rev. Lett.81, 5656s1998d.11M. M. Alkaisi, R. J. Blaikie, S. J. McNab, R. Cheung, and D. R. S.

Cumming, Appl. Phys. Lett.75, 3560s1999d.12A. Kumar, H. A. Biebuyck, and G. M. Whitesides, Langmuir10, 1498

s1994d.13M. Ueda, S.-W. Kim, Sz. Fujita, and Sg. Fujita, Jpn. J. Appl. Phys., Part 2

43, L652 s2004d.14S.-W. Kim, Sz. Fujita, and Sg. Fujita, Jpn. J. Appl. Phys., Part 241, L543

s2002d.15G. Benassayag, C. Vieu, J. Gierak, P. Sudraud, and A. Corbin, J. Vac. Sci.

Technol. B 11, 2420s1993d.16X. Wang, Y. Ding, C. J. Summers, and Z. L. Wang, J. Phys. Chem. B108,

8773 s2004d.17Y. Li, W. Kim, Y. Zhang, M. Rolandi, D. Wang, and H. Dai, J. Phys.

Chem. B 105, 11424s2001d.18B. D. Yao, Y. F. Chan, and N. Wang, Appl. Phys. Lett.81, 757 s2002d.19J. Zhong, S. Muthukumar, Y. Chen, Y. Lu, H. M. Ng, W. Jiang, and E. L.

Garfunkel, Appl. Phys. Lett.83, 3401s2003d.20K. Ogata, S.-W. Kim, Sz. Fujita, and Sg. Fujita, J. Cryst. Growth240, 112

s2002d.21H.-J. Ko, Y. F. Chen, Z. Zhu, T. Yao, I. Kobayashi, and H. Uchiki, Appl.

Phys. Lett. 76, 1905s2000d.22Z. J. Wang, H. M. Zhang, Z. J. Wang, L. G. Zhang, and J. S. Yuan, J.

Mater. Res.18, 151 s2003d.23K.-K. Kim, N. Koguchi, Y.-W. Ok, T.-Y. Seong, and S.-J. Park, Appl.

Phys. Lett. 84, 3810s2004d.24E. Burstein, Phys. Rev.93, 632 s1954d.25U. Woggon,Optical Properties of Semiconductor Quantum Dots, Springer

Tracks in Modern Physics, Vol. 136sSpringer, Berlin 1996d.26Y. Kayanuma, Phys. Rev. B38, 9797s1988d.27E. M. Wong and P. C. Searson, Appl. Phys. Lett.74, 2939s1999d.28E. Hanamura, J. Phys. Soc. Jpn.28, 120 s1970d.

FIG. 14. Luminescence spectra obtained by temperature-dependentm-PLmeasurements, using the same PL configuration introduced in Fig. 13, ofselectively grown 100 nanodots with the dimensionsaverage width andheight is 170 and 20 nm, respectivelyd.



29C. Y. Liu, S. Yuan, J. R. Dong, S. J. Chua, M. C. Y. Chan, and S. Z. Wang,J. Appl. Phys.94, 2962s2003d.

30E. Y. P. Varshni, PhysicasAmsterdamd 34, 149 s1967d.31N. Malkomes, M. Vergohl, and B. Szyszka, J. Vac. Sci. Technol. A19,

414 s2001d.

32A. P. Roth, J. B. Webb, and D. F. Williams, Phys. Rev. B25, 7836s1982d.33B. H. Choi and H. B. Im, Thin Solid Films193/194, 712 s1990d.34H. Teisseyre, P. Perlin, T. Suski, I. Grzegory, S. Porowski, J. Jun, A.

Pietraszko, and T. D. Moustakas, J. Appl. Phys.76, 2429s1994d.



title artificial control of zno nanodots by ion-beam ... · pdf filemasaya ueda, mitsuru...

Documents