1. Introduction

With a wide direct bandgap of 3.37 eV and a large exciton binding energy of 60 meV, ZnO has been widely investigated for a number of technologically advanced applications. Especially as a photodetector material, ZnO has many advantages over other semiconductors, which has been shown to be extremely resistant to high-energy photon irradiation, and therefore, it is appreciated more as photodetection materials for ultraviolet (UV) photons [1–4]. The photoconduction of various types of ZnO including single crystals [5], polycrystalline ZnO films [6], and ZnO nanostructures [7–11] has been investigated by many research groups and the oxygen chemisorption has been believed to play a central role in regulating the UV photosensitivity. With large surface-to-volume ratio and Debye length comparable to their small size, one-dimensional semiconductor nanostructures are considered as the most sensitive and fast responsive materials for sensors [12–17]. Furthermore, individual ZnO nanostructures have very high internal photoconductivity gain due to the surface-enhanced electron-hole separation efficiency [12]. Therefore, ZnO nanostructures become a promising material to make UV detectors. It is usually accepted that nominally undoped ZnO is an n-type semiconductor due to the presence of unintentionally introduced donor centers, like zinc interstitials or oxygen vacancies [18–25]. For nanostructured ZnO with the very large surface-to-volume ratio and typical n-type properties, dangling bonds due to a breaking of lattice periodicity on the surface can induce quantities of acceptor-type surface states, resulting into band bending upward and carrier-depletion layer in the vicinity of surfaces, which can be served as charge traps and electron barriers. Thus surface states will play a crucial role in the physical properties of nanostructures, and furthermore, there is no relationship with the work function of metal electrodes. In particular, Fermi level pinning at surface states would be expected. Therefore, there is a strong effect on the photoconductivity of ZnO, which makes the response and recovery time of photodetectors based on the ZnO NWs longer than 1s [26–29].

ZnO has relatively high sensitivity as UV photodetectors, while the bandgap of 3.37 eV is an important limitation for their applications in visible (VIS) and solar-blind UV (SBUV) range. For ZnO nanostructure photodetectors, although some efforts have been performed to obtain fast, sensitive, and enhanced UV photoresponse in ZnO thin films by microstructure modification, appropriate doping, or depositing a passivation layer on its surface [30–32], most of the early reports focus on the UV response and there have been few reports on their broader spectral response, especially at different bias voltages and illumination intensities. At present, few reports have been published on the photocurrent behavior of ZnO under below-gap VIS and above-gap SBUV light illumination [33–36], in particular, the effects of externally applied bias voltage and excitation light intensity.

Herein, we report on the possibility of creating highly sensitive NW switches by exploring the photoresponse properties of unintentionally doped individual ZnO NWs under UV light illumination, meanwhile expands its response range from UV to VIS and SBUV. The SBUV-VIS-NIR photodetectors with metal-semiconductor-metal (MSM) structure is on the basis of individual ZnO NWs. Under illumination at a low bias voltage, the device shows only a UV response related to ZnO band-edge absorption. At higher bias voltage, however, the band-edge response extends to SBUV region, and furthermore, it blueshifts with increasing bias voltage and illumination intensity. In addition, the detector also shows VIS and NIR response in the wavelength range of 500~900nm at higher bias voltage and illumination intnesity. By carrying out a series measurements on the responsivity of the ZnO NW photodetectors to the light irradiation with different wavelengths under different conditions, e.g., bias voltage, excitation intensity, and environment temperature, our experimental results cannot be explained by the photon-induced oxygen desorption from the surface of the ZnO NWs. Instead, we attribute the broad-spectral-response to the presence of surface states in n-type ZnO NWs, which can be modulated effectively by externally applying bias voltage and illumination intensity. This is of immense help in the realization of a practical ZnO-based SBUV-VIS-NIR photodetector.

2. Experimental

ZnO micro/nanowires used in our study were grown by a high-temperature thermal evaporation of pure ZnO powder (99.99%) at 1470 °C. Microstructure and composition characterization were carried out using a FEI Quanta 200F field emission gun scanning electron microscope (FESEM) and a JEOL 2010 high-resolution transmission electron microscope (HRTEM). The as-grown ZnO NWs are typically n-type and have diameters ranging from 100 to 700 nm.

For the device fabrication, the ZnO wires were dispersed on an Al₂O₃ substrate and then each end of the wire was fixed to the substrate using silver paste; metal wires were also bonded to a ZnO wire for electrical measurement. For photodetection experiments, the monochromatic and continuous wavelength light source was provided by a fluorescence spectrophotometer (Hitachi F4600) with a 150 W Xe lamp. The illumination intensity was controlled by the slot width of incident light. For micro-optical detection experiments, the excitation light source was provided by a microscope objective. Current-voltage (I-V) characteristics of the devices were studied using a Keithley 2400 sourcemeter. All of the measurements were carried out in ambient condition.

3. Results and discussion

The XRD pattern in Fig. 1(a) indicates that the as-synthesized ZnO NWs have a wurtzite structure (JCPDS:36-1451). Figure 1(b) shows a field emission scanning electron microscopy (FESEM) image of the ZnO NWs, showing diameters of around 0.5 μm and lengths of several hundreds of micrometers. Compositional analysis of the nanostructures probed by energy-dispersive X-ray spectroscopy (EDS) shows that they are only composed of Zn and O elements (the below left inset in Fig. 1(c)). The high resolution electron microscopy (HRTEM) image and corresponding fast Fourier transform (FFT) analysis are given in Fig. 1(c), indicating that the ZnO NWs are grown as single crystal with a growth direction along [001], namely c-axis.

 

Fig. 1 Structural characterization of as-synthesized ZnO. (a) XRD pattern. (b) FESEM image. (c) HRTEM image of single NW, taken from [100] zone axis, indicating that ZnO NW is single crystal and grows along [001] direction (c-axis), and the below left and below right insets in (c) correspond to EDS spectrum and FFT pattern, respectively.

Download Full Size | PDF

Figure 2 shows the spectral responsivity characteristics of as-fabricated ZnO-based NW detectors under the illumination with wavelength ranging from 200 to 900 nm at the dc bias voltage of 1, 10, and 20 V. The illumination intensity is controlled by the slot width of incident light. The schematic diagram of the “metal-semiconductor-metal” (M-S-M) device structure is inset in Fig. 2(a). It can be clearly seen that the spectra strongly depend on the illumination intensity and bias voltage. It only shows a strong response peaks centered at about 370 nm under a low illumination intensity, which corresponds to the near-band-edge (NBE) absorption of ZnO. With increase in excitation intensity, the NBE absorption peak widens, and therefore, the detectors show an obvious response to SBUV light. In addition, the detectors also show relatively strong responsivity in VIS and NIR range under higher illumination intensity. In particular, at higher bias voltage, the responsivity to SBUV, VIS and NIR is more obvious. In order to further verify the broader spectral response characteristics, the light with different wavelengths was illuminated in turn on the detector at the dc bias voltage of 1 and 10 V, as shown in Fig. 3. It is more obvious that the detector can show a relatively strong responsivity to the light with wavelength range from 200 to 900 nm.

 

Fig. 2 At the dc bias voltage of (a) 1 V, (b) 10 V and (c) 20 V, the spectral photoresponse of a detector vs wavelengths at different illumination intensities controlled by the slot width of incident light. The inset corresponds to a schematic diagram of a photoconductive apparatus for measuring the photoresponse.

Download Full Size | PDF

 

Fig. 3 At the dc bias voltage of (a) 1 V and (b) 10 V, the photoresponsivity of the detector to the light with different wavelengths, which was illuminated in turn, indicating that all the wavelength light can show a responsivity in range of 200~900 nm.

Download Full Size | PDF

The response behavior of the device was further characterized by measuring the current at fixed bias voltage of 1 and 10 V as a function of time when the device was periodically exposed to the light with wavelength of 330, 370, 650, and 745 nm at different power densities, which are shown in Figs. 4 and 5, respectively. As seen from Fig. 4, under the UV, VIS and NIR illumination the detector all shows a relatively high responsivity at the bias voltage of 1 V, especially compared with the initial dark current. Furthermore, the responsivity increases with increasing excitation intensity. Additionally, when the light is turned on, the photoresponse is monotonously increasing without showing any sign of saturation, especially illuminating of sub-gap lights. The decay of the photoresponse on the cessation of radiations is also very slow, indicating a presence of persistent photoconductivity (PPC), and moreover, it shows a much slower decay with increasing illumination intensity. At the bias voltage of 10V, however, the photocurrent shows a relatively rapid rise firstly, and subsequently increases slowly. When the 370, 650 and 760 nm illumination light is turned off, the photocurrent also shows a slowly decay decrease, and similarly, the decay is much slower with increasing illumination intensity. Under the 330 nm UV illumination, whose photon energy is higher than the energy gap, the photocurrent shows a relatively quick response and decay at 10 V bias voltage, especially under low illumination intensity.

 

Fig. 4 At the dc bias voltage of 1 V, the excitation intensity dependence of photoresponse of the ZnO-based NW detector under the periodic illumination with the wavelength of (a) 330 nm, (b) 370 nm, (c) 650 nm and (d) 745 nm.

Download Full Size | PDF

 

Fig. 5 At the dc bias voltage of 10 V, the excitation intensity dependence of photoresponse of the ZnO-based NW detector under the periodic illumination with the wavelength of (a) 330 nm, (b) 370 nm, (c) 650 nm and (d) 745 nm.

Download Full Size | PDF

Generally, the photocurrent on-off ratio as $\Delta I/I_{dark}$ is critical parameter to determine the sensitivity for an optoelectronic device, where $\Delta I/I_{light-}-I_{dark-}$, and I_light and I_dark are the current under the illumination and in the dark, respectively [37–39]. As seen from Fig. 4 and 5, the dark current is larger at higher bias voltage. At the low bias voltage of 1V, therefore, the photocurrent on-off ratio is relatively large as compared to the high bias voltage of 10V. Another key parameter to evaluate the sensitivity of the UV detector is the photoconductive gain (G), which is defined as the number of electrons collected by electrodes due to the excitation by one incident photon. One can express G as $G=(hv/e)(\Delta I/P)$, where P is the effective light power irradiated on the NW, h is Planck’s constant, ν is the frequency of the incident photon, and e is the electronic charge. Considering the length between Ag electrodes (~um) and the diameter of the ZnO NW (~nm), the photoconductive gain of the UV detector was calculated to be ~1000 irradiated by 330 nm light at the bias voltage of 1 V. Such a relatively high value of $\Delta I/I_{dark-}$ and G indicates high UV and NIR light sensitivity of the present device. From Fig. 5, however, it can be seen that the G shows a decrease with increasing illumination intensity at the high bias voltage of 10 V.

Figure 6 further shows the I-V characteristics of the single ZnO NW-based photodetector measured under different illumination intensities, different excitation wavelengths, and different sweep voltages. It is clear that the ZnO NWs are highly sensitive to the UV, VIS and NIR light irradiation. Compared with the high bias voltage, moreover, the photocurrent on-off ratio is large at the low bias voltage. At higher illumination intensity and bias voltage, the I-V curves show high linear I-V characteristics, namely good Ohmic contact, especially under the UV illumination, implying the disappearance of surface states at the electrode/ZnO NW interface. Moreover, the $\Delta I/I_{dark-}$and G both decrease when I-V curves show Ohmic characteristics, indicating that the surface barrier plays a dominant role in photoresponse. At a high bias voltage, additionally, the photocurrent shows firstly an increase and subsequently changes linearly, indicating that there exists an obvious hysteresis phenomenon in the current response to light.

 

Fig. 6 Excitation wavelength, bias voltage, and illumination intensity dependence of I-V characteristics of individual ZnO-based NWs, and the bias voltage sweeping from negative to positive. (a), (b) and (c) illuminated under 745 nm NIR light at the bias voltage of 1, 10, and 20 V, respectively. (d), (e) and (f) illuminated under 650 nm red light at the bias voltage of 1, 10, and 20 V, respectively. (g), (h) and (i) illuminated under 370 nm UV light at the bias voltage of 1, 10, and 20 V, respectively. The insets in (a) and (b) correspond to the dark current at the bias voltage of 1 and 10V, respectively.

Download Full Size | PDF

In order to switch the NWs to the PPC state, the device was firstly put for 24 h in dark, and then irradiated by the 370 nm UV and 650 nm red light for 600 s, respectively, as shown in Fig. 7. From Fig. 7(a), it can be seen that the photocurrent can almost reach 80% of saturation value within 10s under the UV illumination. After the UV light is turned off, the photocurrent shows two fast decay processes. A quite slow decay tail is observed subsequently for a very long duration due to the PPC effect. Under the red light illumination (Fig. 7(b)), however, the current of the device rises from 0.15 to 1.3 uA within 5 s on illumination. The rapid photocurrent raise is followed by a slower component, in which the photocurrent increases by a further factor of 2 in about 600 s, and furthermore, monotonously increasing without showing any sign of saturation. The decay of the photoresponse on the cessation of red radiations is also very slow. The current with magnitude approximately ten times of its initial value was maintained even after 1 h of decay in the dark condition and the decreasing rate is very low. As seen from the insets in Fig. 7, the detector show a very high initial resistance in the dark, and however, the current can show an slow increase upon applying a bias voltage even without light illumination, indicating that the externally applied bias voltage can affect the charge transport of ZnO-based detector.

 

Fig. 7 Under the illumination with (a) 370 nm UV light and (b) 650 nm red light, the typical photoresponse and decay characteristics of the ZnO-based NW detector at a dc bias voltage of 1V. The insets in (a) and (b) correspond to the current characteristics in logarithmic scale. Besides the persistent photoconductivity, upon applying a bias voltage the current also shows a clear increase in the dark.

Download Full Size | PDF

In general, it is accepted that oxygen molecules are adsorbed onto ZnO surfaces by capturing free electrons from the n-type ZnO [$O_2(g)+e^{-}\to O_2^{-}(ad)$], which creates a low-conductivity depletion layer near the surface [1,12,26,30,40–43]. However, the electronic injection into/from interface states by an externally applied bias voltage and the surface photovoltaic effect seem to be the dominant mechanism in our photodetectors based on individual undoped ZnO NWs. Figure 8 shows the schematic diagrams of the photoresponse mechanism of the photodetector in the dark without an external bias voltage and under illuminating one end subjected to a negative bias voltage, respectively.

 

Fig. 8 Schematic diagrams of the photoresponse mechanism of the photodetector based on individual ZnO NWs. (a) Before contact, showing the work function of Ag and the electron affinity of ZnO. (b) In dark at zero bias voltage, the presence of surface states leads to the formation of back-to-back Schottky-like diode, preventing charges from passing through the Ag electrode/Zn NW interface, and correspondingly, showing a relatively low dark current. (c) Upon illuminating the end subjected to at a negative bias voltage, electrons can be injected into surface states from the electrode, resulting into a decrease and elimination of surface barrier, and moreover, photogenerated electron-hole pairs can be separated efficiently by surface built-in electric field, resulting the free transport of charges at the electrode/ZnO interface. However, electrons are injected into the electrode from interface states upon applying a positive bias voltage, resulting into an increase of surface barrier.

Download Full Size | PDF

The work function of Ag (Φ_Ag = 4.26 eV) is lower than the electron affinity of ZnO (χ_ZnO = 4.35 eV), and therefore, it would be expected to form Ohmic contact when Ag is in contact with n-type ZnO NWs. In the dark, however, the I-V curves show asymmetrical characteristics rather than typical Ohmic contact, especially at relatively low bias voltage and illumination intensity. It is usually accepted that nominally undoped ZnO is an n-type semiconductor due to the presence of unintentionally introduced donor centers, like zinc interstitials or oxygen vacancies. For the nanostructured ZnO with very large surface-to-volume ratio and typical n-type properties, dangling bonds can induce quantities of acceptor-type surface states to exist on the NW surfaces due to the breaking of lattice periodicity on the surface, and thus electrons will diffuse from NW interior toward its surface, resulting into band bending upward and the existence of carrier-depletion layer in the vicinity of surfaces, where majority carriers (electrons) are almost depleted completely. For the surface potential barrier height, moreover, there is no relationship between the work function of metal electrodes and the electron affinity of ZnO. In additional, quantities of singly charged $V_O^{+}$ and doubly charged $V_O^{++}$ exist in depletion region, which can trap electron and then change into neutral V_O [$V_O^{+}+e\to V_O$and $V_O^{++}+2e\to V_O$,]. For wurtzite structure ZnO, positive and negative charge centers are not overlapped in lattice, and furthermore, in this study, ZnO NWs grow along c-axis. Therefore, it is difficult for the surface barriers at the two ends of NWs to equalize completely. As a consequence, an asymmetrical back-to-back Schottky-like diode is formed, and the detector shows a relatively low dark current. The corresponding schematic illustration is shown in Fig. 8(b).

Upon applying a relatively low negative bias voltage at one end of the detector, electrons cannot cross the electrode/ZnO NW interface since the direction of externally applied electric field direction is the same as that of surface built-in electric field, namely reverse bias. Upon applying a relatively high negative bias voltage, however, electrons can be injected into surface states of ZnO NWs from the Ag electrode, and moreover, excess electrons in surface states can drift toward the NW interior. Thus electrons can also be filled into depletion region, resulting into a reaction of $V_O^{+}$ and $V_O^{++}$ to V_O, namely trap filling. Correspondingly, the potential barrier height and depletion region width decrease. Therefore, the current can show an increase upon applying a relatively high negative bias voltage both in dark and under light illumination, which is consistent with the experiment results shown in Fig. 6 (in dark) and 7 (under illumination). When the depletion region width is lower than the mean free path of electrons, it will be very easy for electrons to cross the contact interface by tunneling mechanism. According to oxygen chemisorption mechanism, however, more electrons on the ZnO surface, injected from the electrode applied a negative bias voltage, can lead to the chemisorption of more oxygen molecules onto the surface, and correspondingly, the surface barrier should increase. It is a contradiction with our experiment results, indicating that surface states mainly originate from dangling bonds while oxygen chemisorption can only modulate it.

Upon UV illumination in the vicinity of electrode subjected to a negative bias voltage, quantities of electron-hole pairs are generated since photon energy is close to ZnO bandgap. Subsequently, they can be separated efficiently by surface built-in electric field, and electrons migrate to the NW interior while holes migrate to the NW surface along the potential gradient produced by band-bending, generating surface photovoltaic effect. Thus the trap centers $V_O^{+}$ and $V_O^{++}$ in depletion region can capture electrons and then change into VO, further resulting into a huge decrease in potential barrier height and depletion region width. In additional, the increased electron concentration lifts the Fermi level of ZnO. When it is higher than the conduction band (CB) level, the degenerate semiconductor is formed. As a consequence, charges can cross the electrode/ZnO NW contact interface freely and a typical Ohmic contact is formed, and correspondingly, showing a strong photoresponsivity. Additionally, the rise of Fermi level leads to a blueshift of the bandedge response, expanding to SBUV range. On cessation of UV illumination, firstly, the photogenerated electrons from the CB combine rapidly with free holes, resulting into a rapid decay. Thereafter, owing to a spatial separation of photogenerated charges, the combination rate decreases, accompanying a slow decay. With further combining, electron concentration decreases, and correspondingly, Fermi level reduces as well. When it is close to the CB level, the current will show a rapid decay again and ZnO will return to nondegenerate semiconductor, and thus the surface states will be re-established. After that, VO will release gradually electrons by thermal excitation and then change back into $V_O^{+}$ and $V_O^{++}$, namely detrapping process, simultaneously accompanying a PPC state. Thus the depletion region width will increase gradually and the device will return to its initial state eventually, allowing a smaller conduction channel for electrons to move from the source to the drain in the ZnO NW. In additional, under the VIS and NIR illumination at a low bias voltage, as shown in Fig. 6, I-V curves show non-Ohmic contact properties, indicating that ZnO is still nondegenerate semiconductor, and therefore, the detector only shows one fast decay process compared with UV illumination .

Under the illumination with sub-bandgap VIS and NIR light, although the excited electrons cannot jump to the CB, they can combine with $V_O^{+}$ and $V_O^{++}$ and then make them changed into $V_O^{}$, resulting into a decrease in depletion width. Therefore, the devices can also show a strong VIS and NIR responsivity at high bias voltage and illumination intensity. The recombination rate of electrons with $V_O^{+}$ and $V_O^{++}$ is relatively slow, and therefore, the detector shows a relatively slow photoresponse to VIS and NIR. Without external bias voltage, in additional, the depletion region may extend throughout the entire NWs, and thus majority carriers (electrons) are almost depleted completely inside NWs. Therefore, the concentration of $V_O^{+}$ and $V_O^{++}$ is quite high, resulting a pinning of Fermi level. It is insufficient for barrier decrease to induce Fermi level to unpin at a low bias voltage and illumination intensity, and thus the device hardly responses to the VIS and NIR light.

To further confirm the proposed hypothesis, one end of the photodetctor based on an individual NW was applied a positive or negative bias voltage, respectively, and then illuminated using a micro-beam VIS light from an optical microscopy objective, as shown in Fig. 9. It can be clearly seen that the $\Delta I/I_{dark}$ and G is more obvious when the illuminated electrode is subjected to a negative bias voltage as compared to a positive bias voltage with the same magnitude. Although the surface barrier increases upon applying a positive bias voltage to the electrode, charges can cross freely at the electrode/ZnO NW interface since the direction of external electric field is just opposite to that of the surface built-in electric field, namely forward bias. For light illumination, therefore, it only shows very week responsivity. Furthermore, the photoconductive gain G also decreases with increase in bias voltage and excitation light density, as shown in Figs. 4 and 5. These results verify solidly that the photoresponse characteristics of the undoped ZnO NWs are strongly dependent on surface state-related barrier height and width.

 

Fig. 9 At the dc bias voltage of a) −1 V, b) 1 V, c) −10 V and d) 10 V, the photoresponse of the ZnO-based NW photodetector which was irradiated periodically the same end by a micro-beam VIS light, showing a larger photocurrent on-off ratio and photoconductive gain upon illuminating the end subjected a negative bias voltage compared with a positive bias voltage. The Figures above a) and b) correspond to a schematic diagram of a photoconductive apparatus for measuring the photoresponse where the illuminated end is applied negative and positive bias voltage, respectively.

Download Full Size | PDF

4. Conclusion

In summary, the large surface-to-volume ratio of ZnO NWs results into a presence of surface states due to a breaking of lattice periodicity on the surface, which strongly influences their photoresponse characteristics. Externally applied bias voltage and illumination intensity can modulate the surface states. Upon applying a negative bias voltage at one end of the detector, electrons can be injected into surface states from the electrode, and moreover, photogenerated electron-hole pairs can be separated and transferred efficiently by surface built-in electric field, resulting into a decrease in depletion region width and potential barrier height. Therefore, a strong photoresponsivity is shown since electrons can cross freely the electrode/ZnO NW interface upon applying a negative bias voltage and light illumination. The increased electron concentration in ZnO lifts its Fermi level, resulting in an extension of photoresponse to SBUV range when its Fermi level is higher than its CB level. The $V_O^{+}$ and $V_O^{++}$, existed in depletion region, result into a response to sub-bandgap VIS and NIR light. The gradual release of electrons from VO by thermal excitation, namely detrapping process, leads to a PPC effect, and consequently, the surface band bending is increased and the surface states is re-established. These results demonstrate that nominally undoped ZnO NWs can indeed serve as high-performance photodetectors in SBUV and VIS light range besides UV light.