Chemical Physics Letters 409 (2005) 208–211

www.elsevier.com/locate/cplett

Band gap variation of size-controlled ZnO quantum dots

synthesized by sol–gel method
Kuo-Feng Lin a, Hsin-Ming Cheng a,b, Hsu-Cheng Hsu a, Li-Jiaun Lin b,
Wen-Feng Hsieh a,*
a
Department of Photonics and Institute of Electro-Optical Engineering, National Chiao Tung University, 1001 Tahsueh Road,
Hsinchu 30050, Taiwan, Republic of China
b
Material Research Laboratories, Industrial Technology Research Institute, Hsinchu 310, Taiwan, Republic of China

Received 25 April 2005; in final form 7 May 2005

Available online 3 June 2005

Abstract

ZnO quantum dots were synthesized successfully via a simple sol–gel method. The average size of quantum dots can be tailored
using well-controlled concentration of zinc precursor. Size-dependent blue shifts of photoluminescence and absorption spectra
revealed the quantum confinement effect. The band gap enlargement is in agreement with the theoretical calculation based on
the effective mass model. Furthermore, as the particle size decreases, we observed an increase in the size-dependent Stokes shift
of the photoluminescence peak relative to the absorption onset.
Ó 2005 Elsevier B.V. All rights reserved.

1. Introduction have been employed to produce ZnO quantum QDs [14–

21]. For instance, Guo et al. [15] experimentally estab-
Semiconductor nanoparticles have recently attracted lished that the third-order nonlinear susceptibility of
significant attention for their role in fundamental studies ZnO nanoparticles is almost 500 times larger than that
and technical applications [1,2], mainly due to their unu- of bulk ZnO. Vanmaekelbergh and co-workers [22] dis-
sual photonic characteristics. Zinc oxide (ZnO) is a covered that the optical transitions in artificial atoms
versatile material that has achievable applications in consist of one to ten electrons occupying the conduction
photo-catalysts, varistors, sensors, piezoelectric trans- levels in ZnO nanocrystals. Fonoberov et al. [23] theo-
ducers, solar cells, transparent electrodes, electrolumi- retically investigated that, depending on the fabrication
nescent devices and ultraviolet laser diodes. As a technique and ZnO QD surface quality, the origin of
result, it has stimulated extensive research [3–13]. Com- UV photoluminescence (PL) in ZnO QDs is either
pared to other wide band gap materials, ZnO has a large recombination of confined excitons or surface-bound
exciton binding energy of 60 meV, which results in effi- ionized acceptor–exciton complexes. More and more
cient excitonic emission at room temperature. ZnO unique behaviors are continuously being explored.
nanocrystals or quantum dots (QDs) have superior opti- Understanding the electronic and optical character-
cal properties of the bulk crystals owing to quantum izations in ZnO QDs and nanoparticles is important
confinement effects. In the past decade, various methods from both a fundamental science and a proposed pho-
tonic application point of view. Accordingly, absorption
*
Corresponding author. Fax: +886 3 5716631. spectra were widely used to investigate the band
E-mail address: [email protected] (W.-F. Hsieh). edge emission from ZnO QDs [14–20]. However, direct

0009-2614/$ - see front matter Ó 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2005.05.027
 K.-F. Lin et al. / Chemical Physics Letters 409 (2005) 208–211 209

observation of the band gap variation upon particle size

from PL is relatively rare [21]. In this Letter, we show
growth of high-quality ZnO QDs via a simple sol–gel
method. The average size of nanoparticles can be tai-
lored for use of the appropriate concentration of zinc
precursor. Furthermore, size-dependent PL and absorp-
tion spectra are carefully discussed and compared with
the theoretical calculation from the effective mass
model.

2. Experimental

The ZnO colloidal solutions were produced from zinc

acetate dihydrate (99.5% Zn(OAc)2, Riedel-deHaen) in
diethylene glycol (99.5% DEG, EDTA), similar to what
we presented exhaustively before [24]. The slight differ-
ence is that we placed the final product in a centrifuge
operating 3000 rpm for 30 min. After this procedure,
the solution was separated into two gradations. The
white bottom layer included the secondary ZnO clusters
Fig. 1. TEM image of the ZnO QDs fabricated using 0.06 M
[24] and the upper suspension was more transparent and
Zn(OAc)2.
included the dispersive single crystalline ZnO QDs. Un-
like the secondary clusters, the single crystalline ZnO
QDs were almost monodispersive because of the stable Fig. 2 demonstrates the XRD profiles of the ZnO
surface during the chemical reaction. The average size QDs prepared with various concentrations of Zn(OAc)2.
of ZnO QDs can be tailored under well-controlled con- The diffraction peaks and their relative intensities of
centration of precursor, Zn(OAc)2, including 0.04, 0.06,
0.08, 0.1, 0.16 and 0.32 M. The specimens were prepared 6000
by placing a drop of transparent suspension on glass
substrates and carbon-coated copper grids were then
allowed to air dry to remove excess solvent.
The shapes and sizes of ZnO QDs were analyzed 5000
using JEOL JEM-2100F field emission transmission 0.32 M
electron microscope (FETEM) operated at 200 keV.
The average crystallite sizes were also characterized 4000
XRD Intensity (a. u.)

using Bede D1 thin film X-ray diffractometer (TFXRD) 0.16 M

with grazing incident mode. Optical absorption spectra
were recorded on a Hitachi U4001 spectrophotometer
with xenon lamp. The PL measurement was made using 3000
0.1 M
a 20 mW He–Cd laser at wavelength of 325 nm and the
emission light was dispersed by a TRIAX-320 spectrom-
eter and detected by a UV-sensitive photomultiplier tube. 2000 0.08 M

3. Results and discussion 0.06 M

1000

Fig. 1 shows a typical TEM micrograph of the ZnO

QDs formed using 0.06 M Zn(OAc)2. The nanoparticles 0.04 M
are essentially little aggregated but still appear to be 0
sphere and ellipsoid in shape individually. The mean-par- 30 40 50 60 70
ticle size is estimated to be
4.0 ± 0.3 nm. Presumably 2θ (degree)
due to the viscosity of DEG, the solvent may have mod-
Fig. 2. XRD profiles of the ZnO QDs prepared with various
ified the Ostwald ripening kinetics such that the growth concentration of Zn(OAc)2. The crystalline size can be approximately
rate decreases with the size of the ZnO QDs. This would estimated to be 12, 7.4, 6.5, 5.3, 4.2 and 3.5 nm, respectively (top to
narrow the size distribution of ZnO QDs effectively. bottom), for concentration 0.32, 0.16, 0.1, 0.08, 0.06 and 0.04 M.
210 K.-F. Lin et al. / Chemical Physics Letters 409 (2005) 208–211

spectra all coincide with JCPDS card no. 36-1451, so 3.5 nm. Since the exciton Bohr radius, aB, of bulk
that the observed patterns can be unambiguously attrib- ZnO is 2.34 nm [25], the Coulomb interaction should
uted to the presence of hexagonal wurzite crystallites be relatively small as compared to the kinetic energy
with cell constants of a = 3.251 Å and c = 5.208 Å. No resulting from confinement in our samples and the
excess peaks detected, which indicates that all the pre- ZnO QDs are in the moderate to strong confinement re-
cursors have been completely decomposed and no other gime. In general, quantum confinement shifts the energy
complex products were formed. It should be noted here levels of the conduction and valence bands apart, giving
that the full width at half maximum (FWHM) of the rise to a blue shift in the transition energy as the particle
diffraction peaks increase while decreasing the concen- size decreases. Such phenomenon is also revealed in the
tration of zinc precursor duo to the size effect. Further- absorption spectra, although the faint excitonic absorp-
more, the crystalline size can be estimated to be 12, 7.4, tion peaks due to the moderate size distribution of ZnO
6.5, 5.3, 4.2 and 3.5 nm, for concentration 0.32, 0.16, QDs. However, from this figure, it can clearly be seen
0.1, 0.08, 0.06 and 0.04 M, respectively, by using the De- that the absorption onset exhibits a progressive blue
bye–Scherer formula. The statistical result is consistent shift from 3.43 to 3.65 eV as the size of ZnO QD de-
with the observation from TEM. creases. Similar observations of such size dependence
Fig. 3 shows typical PL and absorption spectra of upon optical properties have been made previously for
samples with different average QD sizes. The UV emis- other semiconductor quantum dots [26–28].
sion represents a relaxed state of exciton near the band The relationship between band gap and size of QD
edge in ZnO QDs. The nature of the UV-PL from can be obtained using a number of models [29–32].
ZnO QDs itself is still a matter of controversy. Some Using the effective mass model for spherical particles
authors attributed the UV-PL to the recombination of with a Coulomb interaction term [29], the band gap
confined excitons [14], while others argued that the emis- Eg ½eV can be approximately written as
sion comes from surface impurities or defects [15]. In  
 bulk h2 p2 1 1 1.8e2
our case, high efficient UV emission near band edge is Eg ffi Eg þ þ  ; ð1Þ
attributed to free exciton emission with high electronic 2er2 me mh 4pee0 r
density of states, which shift to higher energies from where Ebulk
g is the bulk energy gap, r is the particle ra-
3.30 to 3.43 eV as the QD size decreases from 12 to dius, me is the effective mass of the electrons, mh is the
effective mass of the holes, e is the relative permittivity,
e0 is the permittivity of free space, h is PlanckÕs constant
divided by 2p , and e is the charge of the electron. The
polarization term included in this model is usually
12 nm negligible.
Fig. 4 shows both the UV emission peaks, absorption
onsets and the dependence of the band gap enlargement
7.4 nm

4.0
0.24
PL Intensity (arb. units)

Absorption (arb. units)

Stokes Shift (meV)

3.9 0.22
6.5 nm
0.20
3.8
0.18

3.7 0.16
Eg* (eV)

5.3 nm 0.14
3.6 4 6 8 10 12
Diameter (nm)

3.5
4.2 nm
3.4

3.3

3.5 nm 2 4 6 8 10 12 14
Diameter (nm)

2.8 3.2 3.6 4.0 Fig. 4. The dependence of the band gap enlargement versus the ZnO
Photon Energy (eV) QDs diameter as calculated from the effective mass model and the
corresponding experimental data of PL peak maximum (h) and the
Fig. 3. PL (solid line) and absorption (dashed line) spectra near the absorption onset (n). The inset shows the size-dependent Stokes shift
band edge of various ZnO QD size. of various QD diameter.
 K.-F. Lin et al. / Chemical Physics Letters 409 (2005) 208–211 211

on the ZnO QD diameter as calculated from the effec- References

tive mass model (Eq. (1)) with Ebulk g ¼ 3.35 eV, me =
0.24m0, mh = 0.45m0, and e = 3.7, where m0 is the free [1] J.R. Heath, Ed. Acc. Chem. Res. 32 (1999) 389 (Special issue for
electron mass [33]. Enlargement effects are expected to Nanoscale Materials).
[2] A.P. Alivisatos, Science 271 (1996) 933.
be predominant when the QD size is less than 6 nm, [3] F.C. Lin, Y. Takao, Y. Shimizu, M. Egashira, Sensors Actuators
meanwhile PL data and absorption data of the same ten- B 24–25 (1995) 843.
dency indicates effective mass theories accurately, which [4] K.S. Weissenrieder, J. Muller, Thin Solid Films 30 (1997) 300.
predict the size-dependent energy gap. We note here that [5] J. Muller, S. Weissenrieder, J. Anal. Chem. 349 (1994) 380.
the absorption data closely coincides with the curve cal- [6] S.C. Minne, S.R. Manalis, C.F. Quate, Appl. Phys. Lett. 67
(1995) 3918.
culated from the effective mass model. However, the PL [7] Karin. Keis, Nanostructured ZnO Electrodes for Solar Cell
data is not fixed since it represents the emission from a Applications, Acta Universitatis Upsaliensis, Uppsala, 2001.
relaxed state and the exciton binding energy should be [8] J.B. Baxter, E.S. Aydil, Appl. Phys. Lett. 86 (2005) 053114.
considered. We also observed a size-dependent Stokes [9] R.L. Hoffman, B.J. Norris, J.F. Wager, Appl. Phys. Lett. 82
(2003) 733.
shift of the PL maximum relative to the absorption on-
[10] W.I. Park, G.C. Yi, Adv. Mater. 16 (2003) 1907.
set as shown in the inset of Fig. 4. This Stokes shift in- [11] R. Könenkamp, R.C. Word, C. Schlegel, Appl. Phys. Lett. 85
creases as the particle size decreases and has been (2004) 6004.
observed in other II–VI and III–V QDs (e.g., InP, CdSe) [12] Z.K. Tang, G.K.L. Wong, P. Yu, M. Kawasaki, A. Ohtomo, H.
and extrapolated exponentially or hyperbolically [34]. Koinuma, Y. Segawa, Appl. Phys. Lett. 72 (1998) 3270.
Such a presentation was previously reported as either [13] T. Makino, C. Chia, Y. Segawa, M. Kawasaki, A. Ohtomo, K.
Tamura, Y. Matsumoto, H. Koinuma, Appl. Surf. Sci. 189 (2002)
size-dependent electron–phonon processes or size- 277.
dependent spin–orbit exchange interaction, which may [14] D.W. Bahnemann, C. Kormann, M.R. Hoffmann, J. Phys. Chem.
contribute to the Stokes shift. However, the origin of 91 (1987) 3789.
the size-dependent Stokes shift in semiconductor QD [15] L. Guo, S. Yang, C. Yang, P. Yu, J. Wang, W. Ge, G.K.L.
systems remains unclear at present [34,35]. Wong, Appl. Phys. Lett. 76 (2000) 2901.
[16] E.A. Meulenkamp, J. Phys. Chem. B 102 (1998) 5566.
[17] E.M. Wong, J.E. Bonevich, P.C. Searson, J. Phys. Chem. B 102
(1998) 7770.
4. Conclusion
[18] S. Sakohara, M. Ishida, M.A. Anderson, J. Phys. Chem. B 102
(1998) 10169.
In summary, we have demonstrated successfully the [19] S. Mahamuni, K. Borgohain, B.S. Bendre, V.J. Leppert, S.H.
ZnO QDs synthesized by a simple sol–gel method and Risbud, J. Appl. Phys. 85 (1999) 2861.
the average size of ZnO QDs can be tailored under [20] H. Zhou, H. Alves, D.M. Hofmann, W. Kriegseis, B.K. Meyer,
G. Kaczmarczyk, A. Hoffmann, Appl. Phys. Lett. 80 (2002) 210.
well-controlled concentration of zinc precursor. Size-
[21] E.M. Wong, P.C. Searson, Appl. Phys. Lett. 74 (1999) 2939.
dependence of efficient UV photoluminescence and [22] A. Germeau, A.L. Roest, D. Vanmaekelbergh, G. Allan, C.
absorption spectra of various QD sizes give evidence Delerue, E.A. Meulenkamp, Phys. Rev. Lett. 90 (2003) 097401.
for the quantum confinement effect. Furthermore, band [23] V.A. Fonoberov, A.A. Balandin, Appl. Phys. Lett. 85 (2004)
gap enlargement is also in agreement with the theoretical 5971.
[24] H.M. Cheng, H.C. Hsu, S.L. Chen, W.T. Wu, C.C. Kao, L.J.
calculation based on the effective mass model. We also
Lin, W.F. Hsieh, J. Cryst. Growth 277 (2005) 192.
observed an increase in size-dependent Stokes shift of [25] R.T. Senger, K.K. Bajaj, Phys. Rev. B 68 (2003) 045313.
the PL maximum relative to the absorption onset as [26] D.J. Norris, Al.L. Efros, M. Rosen, M.G. Bawendi, Phys. Rev. B
the particle size decreases. 53 (1996) 16338.
[27] C.A. Smith, H.W.H. Lee, V.J. Leppert, S.H. Risbud, Appl. Phys.
Lett. 75 (1999) 1688.
[28] H. Zhang, L.P. Wang, H.M. Xiong, L.H. Hu, B. Yang, W. Li,
Acknowledgements Adv. Mater. 15 (2003) 1712.
[29] L.E. Brus, J. Chem. Phys. 80 (1984) 4403.
[30] M.S. Hyberstsen, Phys. Rev. Lett. 72 (1994) 1514.
Authors gratefully acknowledge financial support [31] A.L. Efros, M. Rosen, Annu. Rev. Mater. Res. 30 (2000) 475.
from the National Science Council (NSC) in Taiwan un- [32] K.E. Andersen, C.Y. Fong, W.E. Pickett, J. Non Cryst. Solids
der Contract No. NSC-93-2112-M-009-035. We also 299 (2002) 1105.
thank the Nano Technology Research Center and [33] S.A. Studenikin, N. Golego, M. Cocivera, J. Appl. Phys. 84
Energy & Resources Laboratories of ITRI for facilitates (1998) 2287.
[34] H. Fu, A. Zunger, Phys. Rev. B 56 (1997) 1496.
support and S.Y. Lai (TEM group of MRL/ITRI) for [35] M. Nirmal, D.J. Norris, M. Kuno, M.G. Bawendi, Al.L. Efros,
great help on electron microscopy measurements. M. Rosen, Phys. Rev. Lett. 75 (1995) 3728.