Introduction

Titanium dioxide (TiO2) is a popular photocatalytic material [1–3] which lacks visible light absorption due to its high optical band gap energy (3.2 eV) [4]. The presence of plasmonic nanoparticles (NPs) such as silver (Ag) can enhance the absorption of TiO2 in the visible region and therefore, its photocatalytic activity [5]. The presence of metal NPs (Ag NPs) produced localized surface plasmon resonance (LSPRs), which coupled with the medium like TiO2 to enlarge the optical cross-sections of the LSPRs and increase the optical path length of the incident photons and thus the absorption also [6]. Therefore, the overall backscattering and dissipation of light absorption over short wavelength were suppressed, resulting in an increase in the photocurrent conversion efficiency. Again, the Ag NPs on the TiO2 thin film (TF) attract the oxygen molecules from the TiO2 and introduce defect states [7]. The absorption of photon can be enhanced due to the sub band gap transitions of the said material via the defects. The introduced trap states can enhance the photosensitivity of the fabricated device by hole trapping process as discussed elsewhere [8]. Presently, the superior performances of multilayered TiO2 TF-based memory devices have been investigated by introducing the metal NPs like platinum (Pt) in between two successive layers [9]. The enhanced charge-trapping ability of the devices was related to the interface defects [10–12], which appeared due to the metal NPs into the system. Again, the ability and advantages of glancing angle deposition (GLAD) technique for synthesis of metal NPs like silver (Ag) and indium (In) were discussed by the authors [13, 14]. In this paper, we have synthesized the Ag NPs on the TiO2 TF by GLAD. The production of enhanced current under illumination condition was examined and discussed elaborately in view of hole trapping process under reverse bias condition. The maximum and enlarged UV light detection by the Ag NPs-patterned TiO2 TF-based devices were experimented as compared to bare TiO2 TF. The enhanced charge-trapping capability of Ag NPs/TiO2 TF/n-Si device was investigated by comparing with that of the TiO2 TF/n-Si. The low impedance and high dark conductivity of the bare TiO2 TF was observed as compared to Ag NPs/TiO2 TF device.

Experiments

The GLAD technique has been carried out to synthesize the pure Ag (MTI, USA) NPs on n-type Si <100> substrate coated with 150 nm TiO2 TF. The substrates were rotated azimuthally with a constant speed of 460 rpm at an orientation of 85 ° with respect to the perpendicular line between the metal source and the planar substrate holder. The depositions were carried out at a base pressure of ∼2 × 10−5 mbar inside the e-beam evaporator chamber. A deposition rate of 1.2 Å/s was kept constant. A 50 nm TiO2 cover layer was then deposited on Ag NPs-patterned TiO2 TF. Gold (Au) and Ag contacts were made separately as top electrodes (area of ∼1.77 × 10−6 m2) of the fabricated plasmonic devices on the cover layer (50 nm TiO2 TF).

The current (I)–voltage (V) characteristics of the samples were investigated by using a Keithley 2400 source measure unit through Au electrode. The wavelength-dependent photo responsivity of the detector was measured using 300 W xenon arc lamp (650–0047) through a monochromator (Sciencetech Inc., Canada) in open beam configuration. The capacitance of the Ag NPs/TiO2 TF-based plasmonic and TiO2 TF devices were measured by Agilent (E4980A) LCR meter using Ag top contact.

Results and Discussion

Device Conduction

Figure 1a shows the schematic of the bare TiO2 TF and Ag NPs-patterned TiO2 TF (plasmonic)-based devices. The top view SEM image of the Ag NPs-patterned TiO2 TF on n-Si substrate was reported by the authors [14]. Figure 1b shows the I–V characteristics for the TiO2 TF device and the Ag NPs/TiO2 TF-based device using Au Schottky contact. The devices were tested under dark and illumination condition.

Fig. 1
figure 1

a The schematic of Ag NPs-patterned TiO2 TF (plasmonic) and the bare TiO2 TF-based devices. b I–V characteristics of TiO2 TF device and Ag NPs/TiO2 TF plasmonic device through Au Schottky contact. c Light/dark current of the plasmonic and thin film devices through Au contact

Full size image

Under dark condition, the TiO2 TF-based detector produced large current in the reverse bias mode as compared to Ag NPs/TiO2 TF-based plasmonic device (Fig. 1b). But under photon illumination from xenon lamp (300 W), the TiO2 TF was almost unresponsive, whereas the plasmonic detector was highly sensitive under reverse bias. The dark current of the TiO2 TF at −3 V was 44 × 10−3 mA/cm2 and for Ag NPs/TiO2 TF was 7.6 × 10−3 mA/cm2, which deflected to 55 × 10−3 mA/cm2 and 3.98 mA/ cm2, respectively, under illumination. The lower current value of the plasmonic device compared to that of the bare TiO2 TF device was due to the presence of large trap states into the Ag NPs/TiO2 TF sample. Again, the photosensitivity of the TiO2 TF and Ag NPs/TiO2 TF devices were calculated for both the reverse bias and forward bias conditions from the ratio of light to dark current of the detectors separately. There was no significant enhancement in photosensitivity (Fig. 1c) observed for the Ag NPs/TiO2 TF device as compared to that of the TiO2 TF, under the applied potential range 0 to +8 V. But the plasmonic (Ag NPs/TiO2 TF) device exhibited enlarged photo efficiency under the reverse bias condition when compared with the forward bias. The maximum photosensitivity of the Ag NPs/TiO2 TF device was ∼700 times than that of the TiO2 TF at −3 V. The coupling of Ag NPs LSPRs and TiO2 layer enhanced the intensity and optimized the projection of the internal electromagnetic field induced by the metal LSPRs. Therefore, the overall backscattering and dissipation of light absorption over 300–700 nm were suppressed, resulting in an increase in the photocurrent. Also, the Ag NPs produced Ag–O NPs [7], introduced oxygen defect related more trap states into the band gap of TiO2 material by attracting the oxygen molecules from structurally unstable TiO2 [15]. Under forward bias, a large number of electron–hole pairs were generated, which were available to ionize (by the scattering process) the interface states at Au/TiO2 TF and increased the depletion region [16] thus preventing the increase in photosensitivity. However, in case of the reverse bias, the trap states at the metal–semiconductor interface trapped holes [17], which efficiently reduced both the depletion width and the Schottky height at the junction. In addition, the electron–hole pairs were prevented from recombination and the photo-generated electrons became available to conduct the current into the device. Again, the minimum depletion layer width and hence the maximum electron tunneling happened at −3 V applied potential, which resulted in the extreme efficiency of the plasmonic device.

The ideality factor (n) and the Schottky barrier height (SBH) of the TiO2 TF and Ag NPs/TiO2 TF-based detectors were evaluated by using the following equation based on the thermionic emission theory [18].

$$ I=A{A}^{**}{T}^2{e}^{\left(\frac{-q{\phi}_b}{kT}\right)}\left[{e}^{\left(\frac{-qV}{nkT}\right)}-1\right] $$
(1)
$$ I={I}_0\left[{e}^{\left(\frac{-qV}{nKT}\right)}-1\right]\;\mathrm{f}\mathrm{o}\mathrm{r}\kern0.5em V > \frac{3kT}{q} $$
(2)

Where A is the contact area, V is the applied voltage drop across the rectifying barrier, T is the absolute temperature in Kelvin, K is the Boltzmann constant, q is the charge of an electron, n is the ideality factor, and I 0 is the saturation current given by

$$ {I}_0=A{A}^{**}{T}^2{e}^{\left(\frac{-q{\phi}_b}{kT}\right)} $$
(3)

Where, A** (9.4 A cm−2 K−2) [19] is the effective Richardson constant and φb is the barrier height at zero bias. The barrier height can be obtained from the equation

$$ {\phi}_b=\frac{kT}{q} \ln \left(\frac{A{A}^{**}{T}^2}{I_0}\right) $$
(4)

Using the thermionic emission (TE) model, the Schottky barrier height (SBH) (φb) and ideality factor (n) of the devices were determined from log I–V (Fig. 1b) characteristic. The calculated value of фb from the I–V characteristics of TiO2 TF and Ag NPs/TiO2 TF-based plasmonic device under dark condition were 0.36 and 0.43 eV, respectively. The ideality factor n is defined as

$$ n=\frac{q}{kT}\frac{dV}{d\left( \ln (I)\right)} $$
(5)

The ideality factor n can be determined from the slope of linear region of semi-log forward I–V plots (Fig. 1b) using the above equation. The ideality factor of plasmonic device was found to be ∼9.5 (calculated from Fig. 1b) and for thin film device was ∼8 (calculated from Fig. 1b). Our results showed that the ideality factors of Au/TiO2 Schottky contacts were higher than the unity. For an ideal diode, the diode ideality factor n should be nearly equal to unity. A high ideality factor is often attributed to defect states in the semiconductor band gap [20]. In a real situation, it may increase due to the effects of series resistance, leakage current, in-homogeneities of film thickness, and presence of interface states at the junction of the diode. The presence of interface defect states at the junction or metal–semiconductor junction is responsible for tunneling process. Therefore, the current conduction in such cases is dominated by tunneling instead of diffusion, which enhances the ideality factor of the detector. In our case, the deposition of Ag NPs on TiO2 TF introduced more oxygen-related defects [7], which appeared at the interface of Au/TiO2 TF. The presence of these defects could be responsible for high tunneling process in the plasmonic device and the increase in the value of ideality factor as compared to TiO2 TF detector.

High Internal Gain

Figure 2 shows the device responsivity measured at different wavelengths under the applied reverse bias of −3 V. The maximum internal gain for the device was calculated to be ∼210 at UV region (380 nm) by using the following relation [21],

$$ G=\frac{R\times h\times c}{\lambda \times e\times \eta } $$
(6)

Where R is the responsivity of the device, h is the Planck’s constant, c is the speed of light, λ is the wavelength of incident radiation, e is the electronic charge, and η is the quantum efficiency (assuming η = 1). The internal gain was found to be much higher than those reported by Zhang et al. for TiO2 (∼142.3) [22] and Zr0.5Ti0.5O2 based solar blind ultraviolet photo detector (∼3.1) [23].

Fig. 2
figure 2

Photo responsivities of the devices

Full size image

The high internal gain of the AgNPs/TiO2 TF/ n-Si-based plasmonic detector was due to enhanced photon absorption and presence of oxygen defect related more trap states into the band gap of TiO2 material and their corresponding conduction as discussed earlier.

Junction Capacitance of the Device

The junction capacitances of the detectors were measured to determine the contribution of the trap states in device operation. The junction capacitance of the TiO2 TF/n-Si and TiO2 TF/Ag NPs/TiO2 TF/n-Si (plasmonic)-based devices were measured through Ag Schottky contact using LCR meter (Agilent E4980-A). Figure 3 shows the capacitance (C)–voltage (V) curves at different frequencies both for the TiO2 TF (Fig. 3a) and Ag NPs/TiO2 TF (Fig. 3b)-based devices. The variations of capacitance were highlighted for the inversion region of the devices.

Fig. 3
figure 3

Frequency-dependent C–V characteristics for a TiO2 TF and b Ag NPs/TiO2 TF (plasmonic) devices

Full size image

The capacitance of the TiO2 TF-based device decreased from ∼87 to ∼6 pF and for Ag NPs/TiO2 TF-based plasmonic device from ∼57 to ∼4 pF due to the increase in frequency from 100 KHz to that of 2 MHz, measured at −12 V applied potential at the top Ag contact. A large amount of variation in inversion capacitance ∼81 pF was calculated for the TiO2 TF-based device, which was larger than that of the plasmonic device ∼53 pF. The presence of trap states at the interface of Ag/TiO2 responded efficiently at lower frequencies, which resulted in the anomalous behavior of the junction capacitance of the devices. Therefore, the small numbers of traps into TiO2 TF were incapable to follow the a.c. signal at higher frequency estimated to the case of Ag NPs/TiO2 TF-based plasmonic detector. The mechanism of following the a.c. signals by the trap states present in TiO2 nanowires (NWs)-based devices and its anomalous C–V behavior were described by the authors [24, 25]. A moderate variation ∼53 pF of inversion capacitance for plasmonic-based device compared to TiO2 TF device may be due to the efficient response of traps at higher frequency.

Memory Window

The capacitance–voltage characteristics (Fig. 4a, b) of the TiO2 TF and plasmonic devices were measured separately at 1 MHz to find out the charge-trapping into the devices. The memory windows of the devices were calculated from the C–V curves in the inversion region.

Fig. 4
figure 4

C–V curves at 1 MHz under a sweeping top electrode voltage from ±6 to ±10 V for a TiO2 TF and b Ag NPs/TiO2 TF devices. c Memory windows of the devices

Full size image

The change of capacitance with respect to the applied voltage on the devices are shown for TiO2 TF (Fig. 4a) and Ag NPs/TiO2 TF (Fig. 4b) under a sweeping top electrode voltage from ±6 to ±10 V. Figure 4c displayed the memory windows for different sweeping top electrode voltage. A memory window of ∆V ≈ 0.63 V for TiO2 TF and ∆V ≈ 0.52 V for plasmonic devices were calculated under a sweeping top electrode voltage (V s) from +6 to −6 V, and then back to +6 V. The memory window increases with the increase of the sweeping top electrode voltages. When the applied sweeping top electrode voltages Vs were ±8 and ±10 V, the corresponding maximum memory windows in its C–V curves were about 1.8 and 2.1 V for TiO2 TF and about 1.9 and 4.5 V for plasmonic devices. Lee et al. have reported that the presence of Pt NPs into TiO2 TF can induce charge trap states [11]. In our case, the presence of Ag NPs into the TiO2 TF matrix acted as charge trap elements, which resulted to large memory window for the plasmonic device compared to bare TiO2 TF.

Device Impedance

Figure 5a, b displayed the frequency dependent impedance at different applied potential from −7 to +12 V, for both TiO2 TF and Ag NPs/TiO2 TF-based devices. The maximum impedance of 720 KΩ was recorded for the Ag NPs/TiO2 TF-based device at 100 KHz, which decreased to 35 KΩ at 2 MHz. In case of bare TiO2 TF, the low impedance of 384 KΩ (100 KHz) and 20 KΩ (2 MHz) was observed as compared to Ag NPs/TiO2 TF-based device. Due to the introduction of Ag NPs into the TiO2 matrix, the oxygen-related defect states were increased (described in the above sections), which enhanced the trap states at the metal semiconductor interface.

Fig. 5
figure 5

Frequency dependent impedance for a TiO2 TF and b Ag NPs/TiO2 TF-based plasmonic devices

Full size image

The presence of trap states into the material held the electrons and decreased the mobility of the device and therefore showed the high impedance of the Ag NPs/TiO2 TF device. The experiment reversely supported the reason of getting low dark current from the TiO2 TF/Ag NPs/TiO2 TF/n-Si (plasmonic) based device as compared to TiO2 TF/n-Si (shown in Fig. 1b) using Au Schottky contact. Therefore, we have observed the same role of trap states in device operation, both for the cases of Ag and Au top electrodes. Finally, it can be manifested that the introduction of defect states due to the presence of Ag NPs into TiO2 band gap, assisted large photon absorption and therefore the photoconduction. Again, the presence of trap states reduced the dark conductivity and charge-trapping capability of the plasmonic device, which can be proposed for the fabrication of nonvolatile memory devices. The enhanced photoconduction suggested that the Ag NPs/TiO2 TF system is capable for efficient photo energy conversion.

Conclusion

The enlarged photocurrent under photon illumination of the Ag NPs/TiO2 TF (plasmonic) device was observed at reverse bias condition than that at forward bias. Under xenon lamp excitation, with top Au electrode the plasmonic device produced maximum ∼700 times enlarged current at applied voltage −3 V, due to coupling of Ag NPs LSPRs and TiO2 medium together with the hole trapping process. A large ideality factor of 9.5 was calculated for the plasmonic device as compared with 8 for bare TiO2 TF. The presence of large numbers of oxygen-related defect states at the interface of metal and semiconductor induced tunnel current into the device, which increased the ideality factor. The maximum internal gain of ∼210 was calculated for the plasmonic detector in the UV region at 380 nm. The capacitances of the TiO2 TF and Ag NPs/TiO2 TF-based plasmonic devices were measured with different biasing voltages under the variation of frequency from 200 KHz to that of the 2 MHz. The continuous decrease in inversion capacitance with increase in frequency was observed with moderate variation (∼53 pF) in case of plasmonic device as compared to TiO2 TF (∼81 pF), which might be due to the presence of huge numbers of traps in plasmonic devices. The maximum memory window of 4.5 V was determined for the Ag NPs/TiO2 TF-based device as compared to bare TiO2 TF of 2.1 V under sweeping top electrode voltage of ±10 V. The presence of high memory window in case of Ag NPs/TiO2 TF was due to the large trap densities and corresponding charge trapping. Due to the introduction of Ag NPs into the TiO2 matrix, the oxygen-related defect states were increased, which enhanced the trap states at the metal semiconductor interface and the impedance of the Ag NPs/TiO2 TF device. Finally, the presence of memory window suggests that the plasmonic devices also can be used as charge-trapping memory device and large photoconduction property for efficient energy harvesting.