Electrical, photoelectrical and morphological properties of ZnO nanofiber networks grown on SiO2 and on Si nanowires

Electrical, photoelectrical and morphological properties of ZnO nanofiber networks grown on SiO2 and on Si nanowires

Nadia Celeste Vega^{I, II}; Monica TiradoIII, ^*; David Comedi^{I, II, III}; Andres Rodriguez^IV; Tomas Rodriguez^IV; Gareth M. Hughes^V; Chris R. M. Grovenor^V; Fernando Audebert^{II, V, VI}

^ILaboratorio de Física del Sólido, Departamento de Física, Facultad de Ciencias Exactas y Tecnología - FACET, Universidad Nacional de Tucumán - UNT, Av. Independencia, 1800, (4000) Tucumán, Argentina

^IIConsejo Nacional de Investigaciones Científicas y Técnicas - CONICET., Argentina

^IIILaboratorio de Nanomateriales y de Propiedades Dieléctricas, Departamento de Física, Facultad de Ciencias Exactas y Tecnología - FACET, Universidad Nacional de Tucumán - UNT, Av. Independencia, 1800, (4000) Tucumán, Argentina

^IVTecnología Electrónica, Escuela Técnica Superior de Ingenieros de Telecomunicación - ETSIT, Universidad Politécnica de Madrid - UPM, 28040 Madrid, España

^VDepartment of Materials, University of Oxford, Parks Road OXI 3PH, Oxford, UK

^VIGrupo de Materiales Avanzados, Facultad de Ingeniería, Universidad de Buenos Aires, Paseo Colón 850, Buenos Aires (1063), Argentina

ABSTRACT

ZnO nanofibre networks (NFNs) were grown by vapour transport method on Si-based substrates. One type of substrate was SiO₂ thermally grown on Si and another consisted of a Si wafer onto which Si nanowires (NWs) had been grown having Au nanoparticles catalysts. The ZnO-NFN morphology was observed by scanning electron microscopy on samples grown at 600 °C and 720 °C substrate temperature, while an focused ion beam was used to study the ZnO NFN/Si NWs/Si and ZnO NFN/SiO₂ interfaces. Photoluminescence, electrical conductance and photoconductance of ZnO-NFN was studied for the sample grown on SiO₂. The photoluminescence spectra show strong peaks due to exciton recombination and lattice defects. The ZnO-NFN presents quasi-persistent photoconductivity effects and ohmic I-V characteristics which become nonlinear and hysteretic as the applied voltage is increased. The electrical conductance as a function of temperature can be described by a modified three dimensional variable hopping model with nanometer-ranged typical hopping distances.

Keywords: nanostructures, semiconductors, ZnO, photoluminescence, photoconductivity

1. Introduction

ZnO nanofiber networks (NFNs) have been of interest in recent years due to their potential in future technological applications^1-7. ZnO is a n-type semiconductor with a direct wide band gap of 3.37 eV (UV), large exciton binding energy (60 meV), high superficial reactivity and high Curie temperature (with magnetic doping). Therefore, it is a predominant candidate for the development of photonic devices in the UV, like laser, LEDs and photodetectors. When combining the ZnO properties with those inherent of nano-structures, like mechanical flexibility, promising materials for the development of solar cells, sensors, optic memories and basic elements for the construction of flexible and transparent electronics are obtained^4,5,8. Last but not least, ZnO has high pizoelectricity, which makes it useful for the manufacturing of transductors and small electro-mechanical devices.

The use of low dimensional structures is a key technological factor in the creation of new functional and sensing devices which benefit from their large surface area to volume ratio^7-9. In addition, there has been interest in achieving Si NW/ZnO NW hybrid devices with enhanced functionality¹⁰ in electronic and photoelectronic properties. Carrier management in hetero-structures could provide ways of enlarging emission wavelength ranges and efficiencies of luminescent devices and also of mastering surface activities in sensors¹⁰.

In this work, we present our study of ZnO NFN samples fabricated on either SiO₂ or Si wafers covered with Si NWs synthesized by low pressure chemical vapor deposition (LPCVD) technique. The morphology and properties of ZnO NFN are strongly dependent on the fabrication conditions, like the substrate temperature. In this work we show how the morphology of the network is modified by changing the substrate temperature from 720 °C to 600 °C, keeping all other growth conditions unmodified. Scanning Electron Microscopy (SEM) images show the morphology of the nanostructure obtained. The interfaces of ZnO NFN/Si NWs/Si and ZnO NFN/SiO₂ were studied with the use of a Dual Beam (FIB/SEM) Microscope.

Photoluminescence, electrical conductance and photoconductance in relation to temperature and to applied polarization voltage were also studied in a sample with ZnO NFN grown on SiO₂ substrate without Si NWs. The conductivity of the ZnO NFN can be represented by two parallel conductors, each of them dominated by Mott mechanism in three dimensions with different parameters.

2. Material and Methods

The substrates used for ZnO NFN growth were two types: (A) 0.5 µm thick thermal SiO₂ on Si (100) and (B) Si (100) covered with Si NWs that had been grown by LPCVD. In all cases, NF growth was promoted by Au nano-clusters deposited on the substrates¹¹. These metallic nanoislands act as catalysts for one-dimensional growth (nanowires, nanocolumns) for different systems^4,11-13.

Three ZnO NFN samples (S₀, S₁, S₂) were fabricated by the vapour-transport method. Their growth details are summarized in Table 1. The ZnO NFN sample (S₀) was grown on a SiO₂ substrate at 720 °C. Samples (S₁, S₂) were grown at 720 °C and 600 °C, respectively on Si (100) substrates that had Si NWs on it. The Si NWs had been synthesized by the LPCVD technique at 400 °C for 30 minutes using Si₂H₆ as precursor gas, with a flow ranging from 8 to 2 sccm. The H₂ flow was changed to achieve a total flow rate of 100 sccm and the total pressure was kept constant at 400 mTorr¹⁴.

The vapour-transport deposition of the ZnO NFN was performed in a pumped quartz tube furnace under Ar high purity (99.999%) flow. An alumina crucible containing ZnO and graphite powders (1:1 weight ratio) was placed inside the quartz tube at a position corresponding to the furnace center (1100 °C), while substrates were placed downstream the tube. The quartz tube, 1m long and 3.4 cm external diameter, was placed within the 0.6 m long tubular furnace. The furnace was controlled by a Honeywell DC1010 temperature controller, which allows setting different profiles of variation of temperature. The substrates were placed parallel to the tube axis at 15.7 cm and 17.9 cm distances from the furnace centre, which correspond to different temperatures: 720 °C and 600 °C respectively, due to the natural temperature gradient profile along the tube^13,15.

The quartz tube was purged with Ar until a base pressure of 250 mTorr was obtained (fixed by a small opening of the air inlet valve at maximum pumping speed) and a steady Ar flow was established by controlling the pumping speed through the vacuum valve, raising the pressure in the tube to about 1200 mTorr. The temperature in the centre of the furnace was set at 1100 °C at a heating rate of 25 °C/min. This temperature value was kept constant during 20 minutes and then, the furnace was naturally cooled to room temperature.

Morphology of all samples was characterized by means of SEM obtained in a JSM 6480 LV microscope and in a Zeiss NVision40 Dual Beam (FIB/SEM) microscope. The later technique was also used for studying the ZnO NFN/Si NW/Si and ZnO NFN/SiO₂ interfaces.

The electrical conductance was determined for sample S₀ by applying a DC bias between two Ag contacts formed on the network (separation: 1mm) and measuring the current with 100 µm diameter Cu wires. I-V curves were obtained with two cycles 0 V → 10 V → - 10 V → 0 V at room temperature. The sample was attached to a He closed circuit cryostat sample holder, using high vacuum grease. The pressure in the cryostat during the measurements was lower than 1mTorr, obtained with a diffusion pump. The sample resistance at room temperature was 80 KΩ, approximately. A voltage source was used to apply a bias on the sample, while the current was measured by recording with a nano-voltimeter the voltage drop across a test resistor connected in series with the sample.

Photoconductance was also evaluated for sample S₀ by illuminating the area between contacts with ultraviolet radiation (3.1 eV) obtained from a pair of commercial light emitting diodes and observing the changes in the electrical current with respect to the conductance without illumination.

The network conductance was also measured as a function of the sample temperature by applying on the sample a constant voltage in the ohmic (linear) range. It was found that, by applying identical contacts on a SiO₂ substrate with no deposit on it, the substrate conductance is below the detection limit of the measurement equipment used (~10 ^{- 9}Ω ^{- 1}) and therefore negligible.

Photoluminescence (PL), also in sample S₀, was measured using the 325 nm wave length of a HeCd laser as excitation source, and a CCD (charge-coupled device) spectrometer from Ocean Optics. The measurements of the PL were performed in the 4 K to 300 K temperature range.

3. Results and Discussion

3.1. Morphology

SEM images corresponding to sample S₀ are shown in Figure 1. The ZnO nanofibers are curled with ramifications and they conform a sponge-like three-dimensional random network of randomly oriented nanofibers (Figure 1a, b). Figure 1c shows a detail of the ZnO NFN/SiO₂ interface.

Figure 2a clearly shows that in sample S₁ ZnO NFN growth occurs in the regions among Si NWs and not from the Si NWs themselves. The cross section view (Figure 2b) shows Si NWs (of ~140 nm length) embedded in the ZnO network. In Figure 2c Au islands and a thin Si film are observed. Au islands are localized at the Si film/Si substrate interface. The Si layer is 100 nm thick and corresponds to Si deposition during the LPCVD process.

Figure 3 shows images of sample S₂, which correspond to the lower ZnO growth temperature (600 °C). The ZnO NFN exhibits a higher density than the network obtained at 720 °C (sample S₁, Figure 2). Hence, for 600 °C, the ZnO NFN obtained is a very dense layer of several microns thick, which covers the whole substrate surface. The cross section view (Figure 3c) shows the Au islands and the Si layer as was observed for the sample S₁.

In the three samples studied, the ZnO NFs do not grow in an ordered way, presumably due to a large O vacancy density, as expected in the present oxygen starving deposition conditions. The lack of O₂ could inhibit coherent one-dimensional crystal growth. Studies of ZnO NW growth on SiO₂ with increasing O₂ reveal the development of a 2D ZnO layer (not observed here) followed by an ordered growth of high crystallinity NW vertically aligned¹⁶.

The morphology of the ZnO structure does not seem to strongly depend on the substrate type or the the presence of Au catalyst on the growth surface. This can be seen by comparing Figures 1 and 2, which essentially show very similar ZnO morphologies for samples grown on SiO₂ with Au catalysts as for samples with ZnO grown on Si NW. Note than in the latter case the Au catalysts on the Si (100) were covered by a thin Si film, while some remained at the tip of Si-NWs, as expected from the vapour-liquid-solid mechanism. Nevertheless, the ZnO NFN grew from the regions where Au nanoclusters had been covered and not from the Si NW tips. Furthermore, the presence of Si NWs (and a Si thin film on the areas between the wires as well) did not significantly affect the growth mode. This can be understood by considering that ZnO tends to grow in a one-dimensional anisotropic way without a metal catalyst due to its polar hexagonal crystalline structure, which favours the ZnO growth along the c axis. In addition, self catalytic growth has also been reported, where Zn clusters act as the catalysts. Indeed, it is very probable that the O₂ starving conditions prevailing in the present experiments favoured the formation of non-oxidized Zn clusters on the SiO₂ and Si substrates.

Further research is clearly needed to understand the growth processes. Future work could concentrate, for instance, on studying the effects of adding an O₂ flux on the resulting ZnO NFN/Si NW morphology. The effect of adding O₂ on the ZnO NWN/SiO₂ has been reported by some of us^15,16.

3.2. Electrical, photoelectrical and optical characterization

I-V curves, obtained on sample S₀ at room temperature, change with the cycling of voltage sweeps, producing some hysteresis effects (Figure 4a). Such behavior is typical of oxide semiconductor materials, and has been studied for the development of electronic memories¹⁷.

Photoluminescence spectra from the ZnO NFN typically show a narrow peak in the UV and a broad peak centered in the green wavelength range (see Figure 4b). The UV peak is centered at 367 nm, which corresponds to photon energy of 3.38 eV. This energy corresponds to the ZnO bandgap energy, indicating that this photoluminescence peak results from the recombination of an electron in the conduction band and a hole in the valence band, or from exciton emission. The broad peak is related to recombination of electrons trapped in deep states within the bandgap, due to defects, such as oxygen vacancies or Zn intersticials. The great intensity of this peak indicates that the nanofibers have a large number of defects, which could be related to their large surface area¹⁸.

Photoconductivity effects are observed in the ZnO NFN when the sample is illuminated with UV, as it can be seen in Figure 4c. The energy of the incidental photons (3.1 eV) is a little lower than the ZnO bandgap energy at 300 K (3.3 eV), thus the photons are weakly absorbed in every nanofiber layer, hence penetrating the whole ZnO structure. When interrupting the illumination, photoconductivity has two decay regimes. The first is very fast (characteristic time much lower than a second), and the second one is very slow and non-exponential, which tends to a conductance value a 30% higher than the conductance in initial darkness. This behavior is usually observed in ZnO and other semiconductor oxides films/sheets^18-21, and is explained by the band-band recombination (which occurs in very short times, in the order of 10 ^{- 9} seconds) limited by the extremely slow release of charges from a deep state in the bandgap, due to defects (see Figure 4a).

To measure the electrical resistance as a function of temperature, the applied voltage is kept constant at a value within the ohmic regime (0.5 V) and the temperature is varied in a range 10-300 K. The conductance (G) versus the temperature (T) (Figure 4d) shows a semiconductor behaviour in the sense that G increases with T, however, the dependence does not follow an Arrhenius expression as would be expected for the thermal activation of carriers. In contrast, they can be described by a exp( - bT ^{- 1/4}), as seen in Figure 4e. Such behaviour is expected in accordance with a conduction mechanism corresponding to variable range hopping (VRH) between localized states in three dimensions²². For higher temperatures, T > 70 K, a new mechanism is established, which is characterized by a faster growth of G with T. As it can be seen in Figure 4e, when we tried to represent the conductance behaviour in the whole range of temperatures as the sum of a contribution of a Mott type (which dominates at low T) and another of an Arrhenius type (which dominates at high T), a good fit is not obtained (Model 1). On the contrary, a satisfactory fit is only obtained when the sum of two Mott mechanisms (A and B) is assumed, one at high T and another at low T (Model 2):

From Mott's VRH expressions¹⁸ we estimated the density of localized states at the Fermi energy (g_F) and the mean hopping distance (R) from the fit values of T_0A and T_0B (See Table 2). The g_F value obtained for mechanism B is reasonable, but for mechanism A, it is too large. However, it must be remembered that this kind of analysis is known to overestimate g_F in many cases²³. The R values obtained are significantly smaller than the typical NW diameters (50 nm). This is consistent with the 3D character of the transport revealed by the T ^{- 1/4} dependence.

The conductance of the dense ZnO nanofiber network is determined by the conductivity of each nanofiber, the connectivity between them and their density. The conductivity of each of them, at the same time, is probably fixed by the density of O vacancies, as this defect behaves as an n dopant in ZnO²¹.

4. Conclusion

Under the present fabrication conditions, the transport in Ar of Zn vapours leads to the formation ZnO NFNs on both, SiO₂/Si and Si NWs/Si substrates whose morphology can be described by three-dimensional networks of interconnected, randomly oriented, curled nanofibers. For high substrate temperature (720 °C) the final product is a sponge-like ZnO structure, while for 600 °C a thick and relatively compact network is obtained. The ZnO NFN/Si NWs/Si samples have the Au islands below a Si thin film, whereas no ZnO thin film could be observed at the ZnO NFN/SiO₂ interface.

ZnO NFN has high photoluminescence and considerable conductance and photoconductance. The conductivity of the ZnO NFN can be represented by two parallel conductors, each of them dominated by Mott mechanism in three dimensions with different parameters. It could be possible that the three-dimensional characteristic of the nanofibers conductivity prevails because their diameter is significantly larger than the average hopping lengths. Also, we speculate that a large vacancy density may be responsible for the disordered character of the NFN, their high PL in the visible, and the hopping conduction. These hypotheses should be verified in future research, together with other possible conductivity models in nanofiber networks.

The network shows quasi-persistent photoconductivity and their I-V characteristics exhibits hysteresis at high V. These properties could be useful for memory applications¹⁷.

The high superficial area in the NFN is also remarkable for the development of ultrasensitive sensors, where the variations of photoconductivity, photoluminescence or conductance could be measured when selected molecules are absorbed at nanofiber walls.

Acknowledgements

This work was partially funded by the Argentinean agencies CONICET, CIUNT 26/E419, FONCyT (PICT 2010 No. 400 and PICT-RAICES No. 2351), MINCyT (Ministerio de Ciencia, Tecnología e Innovación Productiva de la Argentina) and the Spanish Agency AECID.

Received: November 12, 2012

Revised: December 5, 2012

• 1. Datta N, Ramgir N, Kaur M, Kailasa Ganapathi S, Debnath AK, Aswal DK et al. Selective H₂S sensing characteristics of hydrothermally grown ZnO-nanowires network tailored by ultrathin CuO layers. Sensors and Actuators B: Chemical. 2012; 166-167:394-401. http://dx.doi.org/10.1016/j.snb.2012.02.079
• 2. Ko SH, Lee D, Kang HW, Nam KH, Yeo JY, Hong SJ et al. Nanoforest of Hydrothermally Grown Hierarchical ZnO Nanowires for a High Efficiency Dye-Sensitized Solar Cell. Nano Letters. 2011; 11(2):666-671. PMid:21207931. http://dx.doi.org/10.1021/nl1037962
• 3. Na CW, Woo HS, Kim ID and Lee JH. Selective detection of NO₂ and C₂H₅OH using a Co₃O₄-decorated ZnO nanowire network sensor. Chemical Communications. 2011; 47(18):5148-5150. PMid:21416083. http://dx.doi.org/10.1039/c0cc05256f
• 4. Wang N, Cai Y and Zhang RQ. Growth of nanowires. Materials Science and Engineering 2008; R60:1-51.
• 5. Unalan HE, Yan Z, Hiralal P, Dalal S, Kuo D, Milne WI et al. Zinc Oxide Nanowire Networks for Macroelectronic Devices. In: Proceedings of the 8th IEEE Conference on Nanotechnology; 2008; Arlington, Texas USA; 2008. p. 561-564. http://dx.doi.org/10.1109/NANO.2008.166
• 6. Al-Heniti S, Badran RI, Umar A, Al-Ghamdi A, Kim SH, Al-Marzouki F et al. Temperature Dependant Structural and Electrical Properties of ZnO Nanowire Networks. Journal of Nanoscience and Nanotechnology 2011; 11:1-7.
• 7. Yang Q, Guo X, Wang W, Zhang Y, Xu S, Lien DH et al. Enhancing Sensitivity of a Single ZnO Micro-/Nanowire Photodetector by Piezo-phototronic Effect. ACS Nano 2010; 4:6285-6291. PMid:20919691. http://dx.doi.org/10.1021/nn1022878
• 8. Wang ZL. Self-Powered Nanosensors and Nanosystems. Advanced Materials 2012; 24:280-285. PMid:22329002. http://dx.doi.org/10.1002/adma.201102958
• 9. Lupan O, Emelchenko GA, Ursaki VV, Chai G, Redkin AN, Gruzintsev AN et al. Synthesis and characterization of ZnO nanowires for nanosensor applications. Materials Research Bulletin 2010; 45:1026-1032. http://dx.doi.org/10.1016/j.materresbull.2010.03.027
• 10. Rami Reddy Devarapalli RR, Deodatta R. Shinde DR, Barka-Bouaifel F et al. Vertical arrays of SiNWs - ZnO nanostructures as high performance electron field emitters. Journal of Materials Chemistry. 2012; 22(43):22922-22928. http://dx.doi.org/10.1039/c2jm34224c
• 11. Plante MC, Garrett J, Ghosh SC, Kruse P, Schriemer H, Hall T et al. The formation of supported monodisperse Au nanoparticles by UV/ozone oxidation process. Applied Surface Science 2006; 253(4):2348-2354. http://dx.doi.org/10.1016/j.apsusc.2006.05.113
• 12. Wagner RS and Ellis WC. Vapor-Liquid-Solid Mechanism of Single Crystal Growth. Applied Physics Letters. 1964; 4:89-90. http://dx.doi.org/10.1063/1.1753975
• 13. Comedi D, Tirado M, Zapata C, Heluani SP, Villafuerte M, Mohseni P et al. Randomly Oriented ZnO Nanowires Grown on Amorphous SiO₂ by Metal-Catalyzed Thermal Evaporation. Journal of Alloys and Compounds 2010; 495(2):439-442. http://dx.doi.org/10.1016/j.jallcom.2009.10.070
• 14. Rodríguez A, Sangrador J, Rodríguez T, Ballesteros C, Prieto AC and Jiménez J. SiGe Nanowires Grown by LPCVD: Morphological and Structural Analysis. Materials Research Society Symposium Proceedings 2010; 1258:P05-05.
• 15. Vega NC, Wallar R, Caram J, Grinblat G, Tirado M, LaPierre RR et al. ZnO nanowire co-growth on SiO₂ and C by carbothermal reduction and vapour advection. Nanotechnology. 2012; 23(27):275602. PMid:22706726. http://dx.doi.org/10.1088/0957-4484/23/27/275602
• 16. Grinblat G, Capeluto MG, Tirado M, Bragas AV and Comedi D. Hierarchical ZnO nanostructures: Growth mechanisms and surface correlated photoluminescence. Applied Physics Letters 2012; 100:233116. http://dx.doi.org/10.1063/1.4724195
• 17. Villafuerte M, Juárez G, Heluani SP and Comedi D. Hysteretic Current-Voltage Characteristics in RF-Sputtered nc-TiO₂ Thin Films. Physica B 2007; 398(2):321-324. http://dx.doi.org/10.1016/j.physb.2007.04.035
• 18. Comedi D, Heluani SP, Villafuerte M, Arce RD and Koropecki RR. Power-law photoconductivity time decay in nanocrystalline TiO₂ thin films. Journal of Physics: Condensed Matter. 2007; 19:486205. http://dx.doi.org/10.1088/0953-8984/19/48/486205
• 19. Moazzami K, Murphy TE, Phillips JD, Cheung MC-K and Cartwright AN. Sub-bandgap photoconductivity in ZnO epilayers and extraction of trap density spectra. Semiconductor Science and Technology 2006; 21:717-723. http://dx.doi.org/10.1088/0268-1242/21/6/001
• 20. Liao ZM, Lu Y, Zhang JM and Yu DP. Temperature dependence of photoconductivity and persistent photoconductivity of single ZnO nanowires. Applied Physics A 2009; 95:363-366. http://dx.doi.org/10.1007/s00339-008-5058-1
• 21. Tiwari A, Jin C, Narayan J and Park M. Electrical transport in ZnO_1-δ films: Transition from band-gap insulator to Anderson localized insulator. Journal of Applied Physics. 2004; 96:3827-3830. http://dx.doi.org/10.1063/1.1783591
• 22. Mott NF. Localized states in a pseudogap and near extremities of conduction and valence bands. Philosophical Magazine 1969; 19(160):835:852.
• 23. Marshall JM and Main C. A new procedure for calculating the density and energy distribution of localized hopping sites in disordered semiconductors, using low-temperature electrical conductivity data. Journal of Physics: Condensed Matter. 2008; 20:285210. http://dx.doi.org/10.1088/0953-8984/20/28/285210

Publication Dates

• Publication in this collection
  05 Mar 2013
• Date of issue
  June 2013

History

• Received
  12 Nov 2012
• Reviewed
  05 Dec 2012