Chemical Physics Letters 462 (2008) 100–103

Contents lists available at ScienceDirect

Chemical Physics Letters

journal homepage: www.elsevier .com/locate /cplet t

Exciton states of quantum confined ZnO nanorods

Sun Young Kim a, Yun Seon Yeon a, Seung Min Park a,*, Jeong Hyun Kim b, Jae Kyu Song a,*

a Department of Chemistry, Kyunghee University, 1 Hoegi-dong, Dongdaemoon-gu, Seoul 130-701, Republic of Koreab School of Chemical and Biological Engineering, Seoul National University, Seoul 151-744, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history:Received 15 April 2008In final form 24 July 2008Available online 29 July 2008

0009-2614/$ - see front matter � 2008 Elsevier B.V. Adoi:10.1016/j.cplett.2008.07.079

* Corresponding authors. Fax: +82 2 966 3701 (J.K.E-mail addresses: smpark@khu.ac.kr (S.M. Park), ja

Photoluminescence (PL) of zinc oxide (ZnO) nanorods with an average thickness of 5 nm and a length of30 nm is blue-shifted compared to the bulk due to quantum confinement effects. The exciton statesremain relatively stable at a high carrier density due to a smaller exciton size and an enhanced excitonbinding energy in the quantum confined nanorods, whereas the electron-hole plasma states are formedin the bulk at the similar carrier density. A linear dependence of the PL intensity on the excitation inten-sity also corroborates the assumption that the stable exciton states are responsible for the undisturbedemission at a high carrier density.

� 2008 Elsevier B.V. All rights reserved.

Semiconductor nanocrystals such as quantum dots and quan-tum rods are of intense scientific and technological interest, be-cause electronic structures of nanocrystals can be tailored bychanging their sizes and shapes [1–3]. Zinc oxide (ZnO) has a wideband gap (3.37 eV) and a large exciton binding energy (60 meV) atroom temperature. The wide band gap is suitable for UV/blue opto-electronic applications, and lasing actions have been reported inZnO nanowires, bulk particles, and thin films [4–8]. The large exci-ton binding energy affords stable exciton states at room tempera-ture for better optical applications. However, a high carrierconcentration is usually required for a sufficient optical gain. Asthe exciton (carrier) density approaches ‘Mott density’, electron-hole plasma (EHP) states are formed [9–11]. Although EHP processfor emission is common for conventional laser diode operation, theoptical gain by the EHP recombinations is lower than that by theexciton-related processes, because binary recombination of freecarriers in EHP has a lower efficiency than exciton recombination[7]. To keep up the exciton states beyond the Mott density, a lowdimensionality is suggested to be an alternative approach, becausequantum confinement effects can alter the properties of excitonstates [12–16]. In addition, nanoparticles show enhancements ofthe exciton oscillator strength and the quantum efficiency, wherenanorods provide a better optical gain than quantum dots due toa larger absorption cross section and a lower nonradiative decay[17]. In this Letter, we present experimental evidence of the stableexciton states in ZnO nanorods beyond the Mott density of thebulk, i.e., the carrier density regime where EHP recombination isobserved in the bulk, which is suggested to stem from quantumconfinement effects in the nanorods.

The synthesis of ZnO nanorods in nonhydrolytic conditions wascarried out in a nitrogen atmosphere using standard Schlenk tech-

ll rights reserved.

Song).eksong@khu.ac.kr (J.K. Song).

niques, as described in other studies [18,19]. Briefly, the nanorodswere synthesized using a mixture of trioctylphosphine oxide(TOPO, 99%) and tetradecylphosphonic acid (TDPA, 98%) as stabiliz-ing ligands, which was added to a mixture composed of zincacetate (99.99%) and dioctyl ether (99%). After addition of 1,12-dodecanediol (>97%), the mixture was heated to 250 �C and main-tained at this temperature for 2 h. The prepared nanorods wereexamined without further treatments such as post-annealing.The synthesized nanorods were characterized by transmissionelectron microscopy (TEM, Technai 30). Relatively uniform-sizedZnO nanorods with an average size of 5 nm (thickness) � 30 nm(length) are found, as shown in Fig. 1a.

In order to obtain photoluminescence (PL) spectrum (Fig. 1b),nanorods drop-coated on glass substrates were excited with aHe-Cd laser (325 nm). For comparison, PL spectrum of ZnO bulkpowders (Aldrich) drop-coated on glass substrates was also ob-tained, where the exciton emissions are found at the peak energyof 3.25 eV with a FWHM of 0.13 eV. The peak energy and theFWHM are not identical to the known values of ZnO bulk(3.28 eV and 0.10 eV, respectively [20]) presumably due to impuri-ties in ZnO in addition to the homogeneous broadening [15,20–22].On the other hand, the emission peak of the nanorods in UV regionis blue-shifted by 0.05 eV compared to that of the bulk, which is as-cribed to the quantum confinement effects. When the dimension ofa ZnO nanoparticle is comparable to a Bohr radius of the exciton(1.4 nm 6 aex 6 2.3 nm [14,23]), the band gap increases. Accordingto an effective mass approximation, the band gap depends on thesize of spherical nanoparticles [24]:

DEgðRÞ ¼p2h2

2R2

1m�eþ 1

m�h

� �� 1:8e2

eRþ EsolðRÞ ð1Þ

where DEg(R) is the shift of the lowest excited state such as theexciton state in the spherical nanoparticle with a radius of R, m�e

20 nm

2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6Energy (eV)

600 550 500 450 400 350 Wavelength (nm)b

a

bulk

nanorod

Fig. 1. (a) The transmission electron microscopy (TEM) image shows relativelyuniform-sized ZnO nanorods with an average size of 5 nm (thickness) � 30 nm(length). (b) Photoluminescence (PL) spectra of ZnO nanorods and bulk powdersdrop-coated on glass substrates. The emission of the nanorods in the UV region isblue-shifted compared to that of bulk. The dotted line indicates the peak position ofthe bulk powders.

2.9 3.0 3.1 3.2 3.3 3.4Energy (eV)

2.9 3.0 3.1 3.2 3.3 3.4Energy (eV)

a

b

2.8 mJ/cm2

1.0 mJ/cm2

5.6 mJ/cm2

11.2 mJ/cm2

20.0 mJ/cm2

2.8 mJ/cm2

1.0 mJ/cm2

5.6 mJ/cm2

11.2 mJ/cm2

20.0 mJ/cm2

Fig. 2. (a) Excitation intensity dependence of photoluminescence (PL) in ZnOnanorods. The PL is not shifted with an increase in the excitation intensity. Thedotted line indicates the peak position at the excitation intensity of 1 mJ/cm2.Overall shape of PL is virtually unchanged with an increase in the excitationintensity. PL spectra are offset for clarity. (b) Excitation intensity dependence of PLin ZnO bulk. The PL is red-shifted with an increase in the excitation intensity. Thedotted line indicates the peak position at the excitation intensity of 1 mJ/cm2. PLspectra are offset for clarity.

S.Y. Kim et al. / Chemical Physics Letters 462 (2008) 100–103 101

is the effective mass of the electron, m�h is the effective mass of thehole, e is the dielectric coefficient, and Esol (R) is the solvation en-ergy. The first term is the band gap change due to the quantum con-finement, while the second term is the Coulomb attraction relatedto the exciton binding energy. The shift of the lowest excited stateof a spherical nanoparticle with the diameter of 5.0 nm is calculatedto be 0.17 eV from Eq. (1). However, the ZnO nanorods investigatedin this study are not in spherical shapes, which needs complex the-oretical methods to be tested to figure out the shift of the lowest ex-cited state. In fact, the band gap is also dependent on the length ofthe nanorods in the given diameter, although the band gap variationwith respect to the length is not as sensitive as to the diameter[25,26]. In addition, the length dependence of the band gap is notsignificant beyond the aspect ratio of 2 [26]. For the rod shape, a re-cent study reports that the ZnO nanorod of a diameter of 2.2 nm (aneffective diameter of 2.6 nm) and a length of 43 nm shows the shiftof 0.23 eV [15], suggesting that the shift in the diameter of 5.0 nm isin the range of 0.04–0.06 eV, if the shift is only sensitive to thediameter. This range coincides with the observed blue-shift of theemission (0.05 eV), which supports that the UV emission resultsfrom the exciton recombination in the quantum confined ZnO nano-rods. The broad green emissions around 2.5 eV do not appear clearlyin the nanorods. The mechanism of the green emission has beenextensively investigated, and the oxygen vacancy is generally ac-cepted as the origin of the green emission [7,18,19]. We note that

the nearly absent visible emission in the PL spectrum strongly sug-gests the good crystal quality of the nanorods, i.e., a low defect den-sity such as the oxygen vacancy in the nanorods.

In order to study the exciton states of ZnO nanorods at high car-rier densities, nanorods were excited with the third harmonic(355 nm) of a Nd:YAG laser (Surelite I, Continuum, 20 Hz, 6 ns)using a UV microscope objective. The emission was collected bythe same objective, spectrally resolved by a monochromator, anddetected by a photomultiplier tube. Fig. 2a shows PL spectra ofthe nanorods as a function of the excitation intensity, where theemission peak of the nanorods is hardly shifted upon the increasein the excitation intensity. For comparison, PL spectra of the bulkobtained in the same excitation intensity range are presented inFig. 2b. The emission peak of the bulk becomes red-shifted withan increase in the excitation intensity, which is plotted in Fig. 3a.The red-shift observed in the bulk is closely related to the EHPstates, where the exciton–exciton scattering emission is not ob-served due to the reduced crystallinity of the bulk powder [9,11].With the increase in the carrier density, the band gap decreasesdue to exchange and correlation effects. In addition, the Coulombinteractions between an electron and hole become screened. Whenan average exciton–exciton distance is comparable to the excitondiameter at a high exciton density (above the Mott density), theexciton states become destabilized to form EHP states, while theband gap decreases [9–11]. The amount of the red-shift in the bulk

0 5 10 15 20 25

0

100

200

300

400

Inte

nsity

(a.u

.)

Excitation intensity (mJ/cm2)

0 5 10 15 20 25

3.20

3.22

3.24

3.26

3.28

3.30

3.32

Peak

cen

ter (

eV)

Excitation intensity (mJ/cm2)

b

abulknanorod

bulknanorod

Fig. 3. (a) Excitation intensity dependence of peak positions of photoluminescence(PL). The emission peak in the bulk is red-shifted with an increase in the excitationintensity, whereas emission peak in the nanorods is not shifted as much.(b) Excitation intensity dependence of the total PL intensity. The emissionintensities of the bulk are not linear over the excitation intensity, whereas theemission intensities of the nanorods are nearly linear.

102 S.Y. Kim et al. / Chemical Physics Letters 462 (2008) 100–103

is observed to be about 0.04 eV at the excitation intensity of 20 mJ/cm2�pulse. The carrier density is estimated according to the modelthat one absorbed photon creates one carrier [11]:

np ¼Iexcs

hmexcLð2Þ

where Iexc is the excitation power, hmexc is the photon energy, s isthe characteristic time, and L is the effective length. Because thelifetime of the excited carrier is reported to be 200 ps in the bulk[5], which is smaller than the pulse width of the excitation laser(6 ns), s is set to be 200 ps [11]. The effective length of 500 nm(the average diameter of the bulk powder) is used for the bulk, be-cause the diameter of the bulk powder is between the penetrationdepth of the excitation laser (50 nm) and the diffusion length(1 lm) [11]. Then, the carrier density is estimated to be about2 � 1019 cm�3 at the excitation intensity of 20 mJ/cm2�pulse. There-fore, the observed shift (0.04 eV) in the bulk seems to agree wellwith other studies [9–11], because the band gap renormalizationis estimated to be 0.10 eV (the exciton binding energy in additionto the observed shift).

In the confinement regime, however, Coulomb interactions andquantum confinement effects show distinctive dependences on thesize of nanoparticles. In other words, while the exciton size isdetermined by the interactions between the electron and the holein the bulk limit, the exciton size in nanoparticles is mainly depen-dent on the boundaries of the confinement potential [12–14], lead-

ing to a smaller exciton radius with a decrease in the nanoparticlesize [12]:

aex ¼1R

� ��1

ð3Þ

where aex is the exciton radius and R is the nanoparticle radius. Arecent study calculates that the exciton radius is 0.98 nm for theZnO quantum dot of a diameter of 5.3 nm [16], which is smallerthan the Bohr radius of the exciton in the bulk. Thus, the spatialoverlap of excitons in nanoparticles occurs at a higher exciton den-sity, and the critical exciton (carrier) density such as the Mott den-sity in the nanorods is expected to be higher than that in the bulk.For example, the Mott density of the nanorods can be higher (up to10 times) than that of the bulk, when the exciton radius of thenanorods is similar to that of the quantum dot. Accordingly, theexciton states of nanorods are stable even at a high carrier density,where the EHP states are readily formed in the bulk. However, a di-rect comparison may not be possible without more information onthe carrier density calculated with the absorption cross sections andlifetimes of the nanorods. In our preliminary study, the absorptionefficiency of the nanorods is estimated to be similar to that of thebulk. Although the lifetime measurements remain for further stud-ies, the lifetimes of the nanorods with a diameter of 29 nm are sim-ilar to those of the bulk [5,6,27], suggesting that the lifetimes of thenanorods with a diameter of 5 nm might not be much different fromthe bulk. On the other hand, because the diameter of the nanorods(5 nm) is smaller than the penetration depth (50 nm), the effectivelength (L) in Eq. (2) is 50 nm in the nanorods, while L is 500 nm inthe bulk, resulting in a higher carrier density in the nanorods at thesame excitation intensity. Therefore, the carrier density of the nano-rods is supposed not to be smaller than that of the bulk, unless thelifetime or the absorption efficiency of the nanorods is more than 10times smaller. In PL spectra of the nanorods obtained in the sameexcitation intensity range, however, the emission peak of the nano-rods is not shifted as much as that in the bulk (Fig. 3a), presumablybecause the EHP states are not formed in the nanorods even abovethe Mott density of the bulk due to a smaller exciton size.

The stable exciton states at a high carrier density are also ex-plained by the enhanced exciton binding energy. With a decreasein the nanoparticle size (R), the second term of Eq. (1) related tothe exciton binding energy increases with R�1, which results inthe exciton binding energy of 0.29 eV in the spherical nanoparticlewith the diameter of 5.0 nm [24]. Using the temperature depen-dence of the shift in PL spectra, the exciton binding energy is esti-mated to be 0.13 eV in a quantum dot (diameter of 5.3 nm) [16].Because the exciton binding energy is also dependent on the lengthof the nanorods [13], it is not easy to estimate the exact excitonbinding energy in the nanorods. However, the recent study findsthat the exciton binding energy in the ZnO nanorod of a diameterof 2.2 nm (an effective diameter of 2.6 nm) is 6–10 times largerthan that in the bulk [15], which suggests that the exciton bindingenergy in ZnO nanorod with the diameter of 5.0 nm is larger than0.10 eV. Thus, the large exciton binding energy in the nanorods canstabilize the exciton state even at a high carrier density. In otherwords, the exciton states in the nanorods at a high carrier densityare affected by the confinement energies, the screened Coulombinteractions, and the band gap renormalization effects. With theincrease in the exciton density, the Coulomb interactions arescreened and the exciton binding energy decreases. However, theexciton state in the nanorods is not influenced as much due toits large binding energy that originates from quantum confinementeffects, resulting in the stable exciton states at a high carrierdensity.

Another evidence for the stable exciton states beyond the Mottdensity of the bulk is found in a nearly linear increase in the PLintensity as a function of the excitation intensity, as presented in

S.Y. Kim et al. / Chemical Physics Letters 462 (2008) 100–103 103

Fig. 3b. While the total PL intensity of the bulk levels off at the highcarrier density regime due to a less efficient emission of the EHPstates [7], the PL intensity of the nanorods shows a nearly lineardependence on the excitation intensity, supporting that the exci-ton states are mainly responsible for the emission of the nanorodsabove the Mott density of the bulk. Therefore, the emission effi-ciency of the nanorods can be larger than that of the bulk at a highcarrier density, because the exciton recombination has a betterefficiency than the binary recombination of free carriers in EHPstates.

In summary, the ZnO nanorods synthesized in nonhydrolyticconditions exhibit strong UV emission. The absence of visible emis-sion indicates that the nanorods have a low level of defects such asoxygen vacancy. Due to the quantum confinement effects, theemission band is blue-shifted in the nanorods. With an increasein the excitation intensity, the PL spectra of the nanorods remainnearly unchanged, which gives a sharp contrast to the red-shift ob-served in the bulk ZnO. Because of a smaller exciton size and an en-hanced exciton binding energy, the exciton states of the nanorodsare stable at a high excitation intensity that the EHP states areformed in the bulk. A linear dependence of the PL intensity onthe excitation intensity also suggests that the exciton states arestable in the nanorods beyond the Mott density of the bulk, whichpromises a more efficient exciton-related emission in the ZnOnanorods for applications of ZnO nanorod-based laser with a lowerlasing threshold and an improved optical gain.

Acknowledgement

This work was supported by the Korea Research FoundationGrant funded by the Korean Government (MOEHRD, Basic ResearchPromotion Fund, KRF-2006-331-C00145).

References

[1] V.I. Klimov et al., Science 290 (2000) 314.[2] J.T. Hu, L.S. Li, W.D. Yang, L. Manna, L.W. Wang, A.P. Alivisatos, Science 292

(2001) 2060.[3] J. Wang, M.S. Gudiksen, X. Duan, Y. Cui, C.M. Lieber, Science 293 (2001)

1455.[4] H. Cao, Y.G. Zhao, H.C. Ong, S.T. Ho, J.Y. Dai, J.Y. Wu, R.P.H. Chang, Appl. Phys.

Lett. 73 (1998) 3656.[5] H. Cao et al., Phys. Rev. Lett. 84 (2000) 5584.[6] M.H. Huang et al., Science 292 (2001) 1897.[7] J.C. Johnson, H. Yan, P. Yang, R.J. Saykally, J. Phys. Chem. B 107 (2003) 8816.[8] J.K. Song, J.M. Szarko, S.R. Leone, S. Li, Y. Zhao, J. Phys. Chem. B 109 (2005)

15749.[9] D.M. Bagnall, Y.F. Chen, Z. Zhu, T. Yao, M.Y. Shen, T. Goto, Appl. Phys. Lett. 73

(1998) 1038.[10] A. Yamamoto, T. Kido, T. Goto, Y. Chen, T. Yao, A. Kasuya, Appl. Phys. Lett. 75

(1999) 469.[11] C. Klingshirn, R. Hauschild, J. Fallert, H. Kalt, Phys. Rev. B 75 (2007) 115203.[12] F. Rossi, G. Goldoni, E. Molinari, Phys. Rev. Lett. 78 (1997) 3527.[13] Y. Zhang, A. Mascarenhas, Phys. Rev. B 59 (1999) 2040.[14] R.T. Senger, K.K. Bajaj, Phys. Rev. B 68 (2003) 45313.[15] Y. Gu, I.L. Kuskovsky, M. Yin, S. O’Brien, G.F. Neumark, Appl. Phys. Lett. 85

(2004) 3833.[16] W.-T. Hsu, K.-F. Lin, W.-F. Hsieh, Appl. Phys. Lett. 91 (2007) 181913.[17] H. Htoon, J.A. Hollingworth, A.V. Malko, R. Dickerson, V.I. Klimov, Appl. Phys.

Lett. 82 (2003) 4776.[18] J. Joo, S.G. Kwon, J.H. Yu, T. Hyeon, Adv. Mater. 17 (2005) 1873.[19] P.D. Cozzoli, M.L. Curri, A. Agostiano, G. Leo, M. Lomascolo, J. Phys. Chem. B 107

(2003) 4756.[20] C. Klingshirn, Phys. Status Solidi. B 244 (2007) 3027.[21] B.K. Meyer et al., Phys. Status Solidi. B 241 (2004) 231.[22] Z.D. Fu et al., Appl. Phys. Lett. 90 (2007) 263113.[23] B. Gil, A.V. Kavokin, Appl. Phys. Lett. 81 (2002) 748.[24] L.E. Brus, J. Chem. Phys. 80 (1984) 4403.[25] J. Hu, L.-W. Wang, L.-S. Li, W. Yang, A.P. Alivisatos, J. Phys. Chem. B 106 (2002)

2447.[26] D. Katz, T. Wizansky, O. Millo, E. Rothenberg, T. Mokari, U. Banin, Phys. Rev.

Lett. 89 (2002) 086801.[27] S.S. Hong, T. Joo, W.I. Park, Y.H. Jun, G.C. Yi, Appl. Phys. Lett. 83 (2003) 4157.