Journal of Materials Science: Materials in Electronics

, Volume 17, Issue 9, pp 651–655

Fluorescence and infrared spectroscopy of electrochemically self assembled ZnO nanowires: evidence of the quantum confined Stark effect

Authors

  • S. Ramanathan
    • Department of Electrical and Computer EngineeringVirginia Commonwealth University
  • S. Patibandla
    • Department of Electrical and Computer EngineeringVirginia Commonwealth University
    • Department of Electrical and Computer EngineeringVirginia Commonwealth University
  • J. D. Edwards
    • US Army Engineer Research and Development Center
  • J. Anderson
    • US Army Engineer Research and Development Center
Article

DOI: 10.1007/s10854-006-0021-4

Cite this article as:
Ramanathan, S., Patibandla, S., Bandyopadhyay, S. et al. J Mater Sci: Mater Electron (2006) 17: 651. doi:10.1007/s10854-006-0021-4

Abstract

We report room temperature fluorescence (FL) and infrared absorption (IR) spectra of spatially ordered two-dimensional arrays of vertically standing ZnO nanowires. The wires are produced by selective electrodeposition of Zn in 10-, 25- and 50-nm pores of a porous anodic alumina film, followed by chemical oxidation. Wires of different diameters show distinctly different FL emission characteristics associated with either deep level traps, or exciton recombination. The intensity of the peak caused by exciton recombination is larger than that caused by deep level traps, which is unusual in nanostructures, and attests to the high structural purity. We also see an anomalous red-shift in the FL emission spectrum which appears to be evidence of quantum confined Stark shift caused by built-in electric fields in the alumina template. The IR absorption spectra are mostly featureless and show no significant peaks indicating the absence of shallow level traps.

Introduction

ZnO is a wide band gap (3.2 eV) semiconductor with many possible applications in short wavelength (blue or ultra-violet) electro-optical devices such as light emitting diodes and diode lasers, as well as piezoelectric devices, sensors, field emitters and solar cells [15]. ZnO nanowires and nanorods have been fabricated by a variety of techniques [612]. In this paper, we report fluorescence (FL) and infrared absorption (IR) spectra of electrochemically self-assembled ordered arrays of ZnO nanowires. Three sets of samples were fabricated, with diameters of 10, 25 and 50 nm. In Sect. 2, we describe the synthesis procedure and in Sects. 3 and 4, we present the FL and IR spectra. Finally, in Sect. 5, we present the conclusions.

Electrochemical self-assembly of ZnO nanowires

In order to produce ordered arrays of ZnO nanowires, we first prepare a porous anodic alumina film by anodization of aluminum. A 99.997% pure Al foil is degreased and electropolished in a solution of perchloric acid, butyl cellosolve, ethanol and distilled water to produce a mirror like surface [13, 14]. This film is then washed in distilled water, air dried and anodized in a suitable acid at room temperature using a constant anodizing voltage. The anodization is carried out for several hours to form a thick porous alumina film on the surface. This film is stripped-off in hot chromic/phosphoric acid and the anodization repeated again for 5 min to form another porous alumina film of thickness less than 1 µm. The two-step anodization process [15] yields a film with a well ordered array of pores as shown in the scanning electron micrograph of Fig. 1. The cross-section micrograph in Fig. 2 shows that the pores are indeed straight and do not meander even when the aspect ratio (length-to-diameter ratio) is large.
https://static-content.springer.com/image/art%3A10.1007%2Fs10854-006-0021-4/MediaObjects/10854_2006_0021_f1.jpg
Fig. 1

Scanning electron micrograph of a porous anodic alumina film showing a hexagonal close packed arrangement of pores. The pore diameter is 50 nm. This membrane was produced by anodization in oxalic acid with a dc voltage of 40 V

https://static-content.springer.com/image/art%3A10.1007%2Fs10854-006-0021-4/MediaObjects/10854_2006_0021_f2.jpg
Fig. 2

Cross section scanning electron micrograph of pores showing that the pores are straight and do not meander. Thus, nanowires produced by electrodeposition of materials within the pores remain straight even if their aspect ratios (length-to-diameter ratios) are large

Table 1 lists the anodizing acid and the anodizing voltage used to produce pores of a given diameter.
Table 1

Parameters for pore formation

Pore diameter (nm)

Anodizing acid

Anodizing voltage

50

3% oxalic

40 V dc

25

3% oxalic

25 V dc

10

15% sulfuric

10 V dc

The pores are then selectively filled up with Zn using ac electrodeposition in a non-aqueous solution. The solution consists of ZnClO4 (10.5 gm), LiClO4 (2.5 gm) and dimethyl sulfoxide (250 ml). The porous film is immersed in the solution and an ac potential of 25 V rms and frequency 250 Hz is imposed between the aluminum substrate and a graphite counter electrode. The temperature is maintained at 75°C. The Zn atoms are selectively electrodeposited within the pores which offer the least impedance path for the ac current (displacement current) to flow. This method is slightly different from the technique employed in [16]. Metallic Zn within the pores is then oxidized to ZnO by immersion in H2O2 at room temperature for 5 h. This process results in ZnO nanowires (of varying heights) housed within the pores which have diameters of 10, 25 and 50 nm.

In Fig. 3, we show a transmission electron micrograph of a single ZnO nanowire captured on a TEM grid. For imaging purposes, the wires were released from the alumina host by dissolving the alumina in hot chromic/phosphoric acid. The released wires were then captured on a TEM grid and imaged. The image shows a ZnO nanowire of about 50 nm diameter, with a length of about 0.6 µm.
https://static-content.springer.com/image/art%3A10.1007%2Fs10854-006-0021-4/MediaObjects/10854_2006_0021_f3.jpg
Fig. 3

Transmission electron micrograph of a single ZnO nanowire. The wires were released from their alumina host by dissolution in hot chromic/phosphoric acid. The released wires were captured on a TEM grid and imaged

Fluorescence spectra

Photoluminescence spectra of ZnO nanostructures have been reported by a large number of groups [6, 1619]. The spectra typically consist of two bands: a low frequency (green) band associated with emissions from deep levels induced by structural defects, and a high frequency (ultra-violet) band associated with bound exciton recombination. Vanheusden et al. have argued that the low frequency peak arises from optical transitions in singly ionized oxygen vacancy in ZnO [20].

In Fig. 4, we show the room temperature FL emission spectra from 10-, 25- and 50-nm diameter nanowires. The spectra are collected using a JY Horiba Fluoromax-3 and Fluorolog Tau-3 spectrometer (xenon arc lamp).
https://static-content.springer.com/image/art%3A10.1007%2Fs10854-006-0021-4/MediaObjects/10854_2006_0021_f4.gif
Fig. 4

Fluoresence spectra of ZnO nanowires of 10, 25 and 50 nm diameter taken at room temperature. (a) and (b) show the emission characteristics of two sets of samples. In (b), the emission peaks of the 50-nm samples are slightly more red-shifted, presumably because this sample has a larger built-in electric field due to space charges trapped in the alumina. Consequently, it displays a more pronounced Stark shift

The 10-nm diameter wires do not show the high frequency peak, but show the low frequency peak centered at 490 nm wavelength. Ref. [6] also failed to observe the high frequency peak in 20-nm diameter wires produced in alumina templates, and instead observed only the low frequency peak.

The 25-nm diameter wires show the high frequency peak centered at a wavelength of 390 nm but not the low frequency peak. The full-width-at-half-maximum (FWHM) linewidth is about 430 meV which is much larger than the thermal energy kT. Therefore, this linewidth is clearly the result of inhomogeneous broadening accruing from variations in the length of the nanowires.

The 50-nm diameter wires show both the high and the low frequency peaks, in addition to a peak at 430 nm wavelength originating from the anodic alumina film. The 430 nm peak is due to optical transitions in singly ionized oxygen vacancies (F+ centers) in the alumina templates [21, 22] and its origin has been confirmed by electron paramagnetic resonance measurements [22].

In the 50-nm samples, the ratio R of the high frequency peak intensity to the low frequency peak intensity is 1.5, whereas in most other work involving ZnO nanostructures, it has been less than unity. For example, ref. [19], which reports photoluminescence of 400–500 nm nanorods, found R = 0.25. Since the low frequency peak is associated with structural defects (vacancies), the quantity R is a measure of the structural quality. Only in heteroepitaxial thin films (not nanostructures), has R been found to be large [17]. Thus, an R-value larger than unity in our nanostructures attests to the high degree of structural integrity.

There is also evidence of quantum confined Stark effect in the emission spectra. It is known that there are strong built-in electric fields in the alumina templates due to trapped charges [23, 24]. These fields induce the quantum confined Stark effect [25, 26] which tends to quench the high frequency peak associated with exciton recombination and also induces a red-shift in the peak position. The band gap of ZnO is 3.37 eV and the exciton binding energy is 60 meV [27, 28]. Therefore, we expect the high frequency peak to be centered at an energy of 3.37 − 0.06 = 3.31 eV, or a wavelength of 375 nm, if we neglect any blue shift caused by quantum confinement. In the 50-nm sample, we find the peak to be around this value, but in the 25-nm sample, the peak is centered at around 390 nm, meaning there is at least a 15 nm red-shift. This is a surprising result and counter-intuitive. We actually expect the emission peak in the 25-nm sample to be blue-shifted compared to the 50-nm sample, because of stronger quantum confinement. Yet we consistently observe the opposite trend. This can be explained by either strain effects or the quantum confined Stark effect. Strain effects are negligible since the nanowires are not single crystals (they are polycrystalline). Therefore, the likely cause is the quantum confined Stark effect which renormalizes the subband energies in a nanowire and induces a red-shift in the emission peak. Note that the actual red-shift in the 25-nm sample is larger than 15 nm, since we did not account for any blue-shifts caused by quantum confinement.

By invoking the quantum confined Stark effect, we can explain all of the observed trends. The built-in electric fields are likely to be progressively larger in anodic alumina films with smaller pores because the “porosity” (defined as the ratio of pore volume to the volume of alumina) decreases with decreasing pore diameter. Thus, samples with smaller pores have more alumina available in a given volume to host more fixed charges, and therefore a larger built-in electric field. Accordingly, the 10-nm samples have the largest built-in electric fields which ionize the excitons and completely quench the high frequency peak associated with exciton recombination. That is why we do not observe the high frequency peak in these samples. The 25-nm samples have intermediate values of built-in fields which are not strong enough to quench the high frequency peak completely, but induce a red-shift of at least 15 nm. Finally, the 50-nm samples have the weakest built-in fields. Therefore, the high frequency peak is least affected in these samples, both in terms of intensity and frequency shift.

Infrared absorption spectra

Figure 5 shows the IR absorption spectra of 10-, 25- and 50-nm diameter wires. These spectra are obtained after subtracting the spectra of the bare anodic alumina membrane from the total absorption spectra. We see no peaks (the ripples are due to interference effects caused by the finite thickness of the membranes) in the near infrared region, indicating that there are no shallow level traps. There is a weak peak in the long wavelength region (centered at a wavelength of 13.3 µm) that appears in the 25-nm samples. The transition energy associated with this peak is 92 meV. Since the effective mass of electrons in the conduction band is 0.27 times the free electron mass [29], the energy separation between subbands in the conduction band is ∼ ∼6.6 meV for quantum dots with dimensions of 25-nm on the edge (assuming hard-wall boundary conditions and a square well potential). Even if we account for side depletion due to interface states, which is known to reduce the effective carrier confinement volume in these nanostructures [13, 30], we cannot explain the large transition energy of 92 meV by ascribing it to inter-subband excitation. Therefore, this peak is most likely caused by carrier transfer from unintentional dopants producing impurity levels near the conduction band edge.
https://static-content.springer.com/image/art%3A10.1007%2Fs10854-006-0021-4/MediaObjects/10854_2006_0021_f5.jpg
Fig. 5

Infrared spectra of 10-, 25- and 50-nm diameter ZnO nanowires taken at room temperature. There are no absorption peaks in the near infrared region, indicating the absence of shallow level traps. There is a weak peak in the long wavelength region in the 25-nm samples, which could be caused by unintentional dopants

Conclusion

In conclusion, we have self-assembled 10-, 25- and 50-nm diameter ZnO nanowires, which show fluorescence characteristics that attest to the high degree of structural integrity. They also show the quantum confined Stark effect, which can be used to modulate the emission frequency with an externally applied electrostatic potential [25]. This can lead to wavelength-tunable ultra-violet lasers and electro-optic devices.

Acknowledgements

The authors are indebted to Dr. Feng Yun for help with SEM imaging.

Copyright information

© Springer Science+Business Media, LLC 2006