Aniello
Pelella
*a,
Luca
Camilli
a,
Filippo
Giubileo
b,
Alla
Zak
c,
Maurizio
Passacantando
d,
Yao
Guo
e,
Kimberly
Intonti
f,
Arun
Kumar
f and
Antonio
Di Bartolomeo
*f
aDipartimento di Fisica, Università di Roma “Tor Vergata”, Via Della Ricerca Scientifica, 00133 Rome, Italy. E-mail: aniello.pelella@uniroma2.it
bCNR-SPIN Salerno, via Giovanni Paolo II, 84084 Fisciano, Salerno, Italy
cFaculty of Sciences, Holon Institute of Technology, Holon 58102, Israel
dDepartment of Physical and Chemical Sciences, University of L'Aquila, 67100 Coppito, L'Aquila, Italy
eBeijing Institute of Technology, Haidian, Beijing, 100081, China
fDepartment of Physics “E. R. Caianiello”, University of Salerno, via Giovanni Paolo II, 84084 Fisciano, Salerno, Italy. E-mail: adibartolomeo@unisa.it
First published on 3rd December 2024
Devices based on transition metal dichalcogenide nanotubes hold great potential for electronic and optoelectronic applications. Herein, the electrical transport and photoresponse characteristics of a back-gate device with a channel made of a single tungsten disulfide (WS2) nanotube are investigated as functions of electric stress, ambient pressure, and illumination. As a transistor, the device exhibits p-type conduction, which can be transformed into ambipolar conduction at a high drain–source voltage. Increasing ambient pressure enhances the p-type behaviour, while exposure to light has the opposite effect, enhancing n-type conduction. The ability to operate the device as either a p-type or n-type transistor makes it promising for complementary metal–oxide–semiconductor (CMOS) circuit applications. Light enhances the conductivity, allowing for further control and enabling the device to function as a photodetector with a photoresponsivity of about 50 mA W−1 and a broadband response in the visible range. The combination of voltage, pressure and light control paves the way for using the WS2 nanotube transistor as a multifunctional device.
Among 2D TMDs, tungsten disulphide (WS2) has garnered attention for its exciting properties and promising performance in various applications, including electronics and optoelectronics.14 WS2 exhibits strong spin–orbit coupling, band splitting, and high nonlinear susceptibility.15–17 Structurally, WS2 is a hexagonal layered compound with strong in-plane interactions and weak out-of-plane interactions, resulting in the limited amount of dangling bonds on the layers’ edges and inert van der Waals surfaces.18,19 This structural feature contributes to its stability and unique properties. The reduced effective electron mass in WS2 has been predicted to result in the highest mobility and best performance in FETs among 2D TMDs, making it a promising candidate for advanced electronic devices.20 Additionally, WS2 has shown promise in optoelectronic applications due to its layer-tuneable bandgap and exotic valley physics, making it suitable for light–matter transduction.21,22
One-dimensional (1D) nanotubes (NTs) can offer several advantages over their 2D counterparts. They can enable transistors with smaller size and enhanced gate control through a gate-all-around configuration. More importantly, WS2 nanotubes exhibit a bulk photovoltaic effect (BPVE) which can be exploited in a new generation of self-powered photodetectors. Indeed, moving from a two-dimensional monolayer to a nanotube with polar properties, reduces the crystal symmetry beyond mere broken inversion symmetry of WS2 and greatly enhances the BPVE.23 Investigations into the potential of WS2 NTs have showcased their electromechanical properties even beyond carbon NTs.24,25 Indeed, they have been identified as a versatile class of materials with a wide range of potential applications. For instance, WS2 NTs possess excellent electrochemical performance in lithium-ion batteries.26 Additionally, WS2 NTs exhibit high performance as photodetectors for visible and near-infrared light, positioning them as promising candidates for nanoscale optoelectronics.27 Moreover, their significant bulk photovoltaic effect, particularly in the infrared region, could have implications for advanced photovoltaic applications.23,28
The synthesis of WS2 NTs has been explored using various methods.25,29,30 WS2 NTs with small diameters and few walls are crucial for understanding their distinct characteristics.31,32 Indeed, studies have explored the optical properties of WS2 NTs, revealing that a decrease in NT diameter induces a red-shift in photoluminescence, suggesting a narrowed band gap due to a curvature and strain effect.33 In another study, the cathodoluminescence of WS2 NTs with different diameters and amount of layers was investigated and compared to literature results. Here, the red shift of all types NTs vs. bulk WS2 was attributed to the compatible effect between the quantum confinement (blue shift) in c-direction due to reduction of number of layers to single layer and strain promoted red shift.34 Recent studies have also focused on the functionalization of tungsten disulfide (WS2) NTs with various materials to enhance their properties for specific applications, such as fluorescence and conductivity.35,36
However, FETs based on individual WS2 NT have not been extensively explored, compared to their 2D WS2-based counterparts.14,37 Notably, the pioneering work by R. Levi et al. in 2013 introduced the first FET38 utilizing a single WS2 NT, with limited subsequent studies reported. In this context, we have recently reported an individual multi-walled WS2 NT which has been utilized as a channel of a back-gated FET, exhibiting p-type behaviour in ambient conditions and low bias, with a hole mobility μp ∼ 1.4 cm2 V−1 s−1 and a subthreshold swing SS ∼ 10 V dec−1.39
In this work, we synthesized WS2 NTs through a two-step reaction process and spread them onto a Si/SiO2 substrate.30 With the help of scanning electron microscopy (SEM), we determined the position of single NTs and contacted them using electron beam lithography. This device is a FET with p-type behaviour at low bias and ambient conditions, and conductance that can be tuned by ambient pressure. For consistency, all the data reported herein refer to the same device. However, we fabricated several devices in the same batch, which showed similar behaviour.39
Most importantly, we demonstrate that the device can switch from hole-dominant to electron-dominant conduction, i.e. it exhibits an ambipolar behaviour, at a drain–source voltage bias higher than about 1 V. Ambipolar devices, capable of functioning as both n-type and p-type transistors by adjusting the bias, offer flexibility in logic and analog circuit design.40–42 Incorporating ambipolar transistors in CMOS (complementary metal–oxide–semiconductor) circuits enables the implementation of more complex logic functions per chip, thereby enhancing the computational capabilities of integrated circuits.
Furthermore, the device can be operated as a photodetector in the visible range, with a responsivity peak around 1.9 eV, correlated with WS2 indirect band gap value. The control by electric bias, pressure and light can make WS2 NT based transistors versatile devices in several applications.
Transmission electron microscopy (TEM) was performed to check the structure of the NT and measure its radius. From the TEM image reported in Fig. 1b, we estimated the NT outer radius to be r = 17 nm, which is in the range of the average dimensions for NTs obtained through the same fabrication process (see Experimental section).39 Here the interlayer distance in NTs’ walls is of 0.63 nm, ∼1.6% larger than that of bulk WS2 (0.62 nm), which was assigned to strain in the rolled layers.34
Electrical characterization of the device was performed under dark and high vacuum conditions (p = 1 × 10−4 mbar). The Id–Vds output characteristic, i.e. the measurement of the channel current versus the applied drain–source voltage Vds, at Vgs = 0 V, displayed in Fig. 1(c), reveals a slightly asymmetric Schottky behaviour which can be due to a small difference in the junction areas between NT and source/drain contacts.43–45 The Id–Vgs transfer characteristic at Vds = 0.25 V in Fig. 1(d) reveals a hysteretic p-type conduction. The hysteretic behaviour reported in the transfer characteristic of the device is due to intrinsic defects and the presence of O2 and H2O on the NT surface that induce slow-time charge trapping, thus enlarging the difference between the forward and the reverse sweep curve.46 From the transfer characteristic, we estimate the hole mobility using the following formula:
![]() | (1) |
To complete the FET electrical characterization in high vacuum, we measured several Id–Vds characteristics at different applied Vgs (Fig. 2a). Vgs varies in the range ±100 V, with a step of 10 V. Fig. 2a shows different behaviours between forward and reverse branches. Indeed, at high positive Vds, the drain current suddenly increases for Vgs > 30 V. This is clearly shown in the plot of Fig. 2b that reports the current at Vds = ±3 V. The black curve in Fig. 2b reveals that the WS2 NT transistor has an ambipolar behaviour at Vds = +3 V, while the conduction is clearly dominated by holes at negative Vds = −3 V, for which the transfer characteristic is p-type (red curve in Fig. 2b). We note that, at Vds = +3 V, the n-branch at positive Vgs becomes higher than the p-branch at negative Vgs, resulting in a complete switching from a p-type to a dominant n-type conduction.
![]() | ||
Fig. 2 (a) Output characteristics in dark and high vacuum conditions. (b) Current values extracted at ±3 V from data in (a). |
With the aim to investigate the ambipolar conduction of the WS2 NT FET, we further studied the transistor properties at positive Vds. Fig. 3a reports the transfer characteristics of the device at different positive Vds, from 0.25 V to 3.00 V, with a step of 0.25 V. The data show that, starting from a p-type device, the increase in Vds activates electron conduction, thus enabling ambipolarity. Fig. 3b shows that, while hole mobility stays constant, the electron mobility increases with Vds. Indeed, switching from p-type conduction to dominant n-type conduction occurs at about Vds = 1.25 V, where electron mobility overcomes hole mobility.
Fig. 3c shows the transfer characteristics of the device measured at different Vgs ranges, from ±20 V to ±100 V, at a fixed Vds = 2 V. To correctly evaluate the effect of the different Vgs ranges, we define the hysteresis H as the difference of Vmings, i.e. the voltage at which Id is minimum (Imind), in reverse and forward voltage sweeps. H versus Vgs is plotted in Fig. 3d. The increase in the hysteresis with Vgs range is due to the increasing carrier trapping at the WS2 NT/SiO2 interface, due to the longer sweeping time.41,51–53 The left-shift of Vmings during the forward sweep is stronger compared to the right-shift that occurs during the reverse sweep (Fig. 3e) and indicates that hole trapping is more effective than electron trapping.54,55 Furthermore, the increasing Vgs range and measurement time also leads to an overall increase in current, as reported in Fig. 3c and highlighted in Fig. 3f, where the minimum values of the current are plotted versus Vgs range.
The observed behaviour can be explained by the band diagram model described in Fig. 4. WS2 is a p-type semiconductor, hence the Fermi level (Ef) is close to the valence band, as demonstrate also by XPS.25 Also, the work function of WS2 is around 5.1 eV and is higher than the titanium work function, which is 4.33 eV.25,56 These considerations lead to Fig. 4a, which shows the band diagram when low Vds < 1 V and Vgs = 0 V are applied. In this scenario, a very low current, only due to the holes that can overcome the barrier at the drain contact, flows through the FET channel. When Vgs increases (Fig. 4b), the bands bend downwards, further blocking the migration of holes. In contrast, when a negative Vgs is applied, the bands bend upwards favouring the flow of holes through the channel (Fig. 4c). This explains the p-type conduction of the devices observed at low drain voltage.
On the other hand, when a high Vds > 1 V is applied, the switching of Vgs from negative to positive values enables ambipolar conduction. Indeed, when Vgs = 0 V, holes can overcome the drain barrier, generating a p-type current (Fig. 4d). At positive Vgs, the bands bend downward, suppressing hole injection from the drain; however, the simultaneous thinning of the electron barrier at the source enables electron tunnelling, which is responsible for the emergence of a n-type current (Fig. 4e). Finally, at negative Vgs, hole conduction is favoured by the upward bending of the bands (Fig. 4f).
Increasing ambient pressure to standard conditions exposes the WS2 NT to molecules such as oxygen and water. These molecules act as p-dopants when adsorbed on the surface of the NT. This results in an increase in hole density, i.e. of the channel conductance, and a slight change in the shape of the output and transfer characteristics of the device. The output (Fig. 5a) and transfer characteristics (Fig. 5b) measured at different pressures, from p = 1 × 10−4 mbar to standard ambient conditions (p = 1 × 103 mbar), confirm the expected trend. Output characteristics reveal a resistance linear behaviour, which decreases with increasing pressure, as highlighted in the inset of Fig. 5a.
Electron and hole mobilities have been estimated and reported in Fig. 5c. The charge carrier mobilities are almost constant in the investigated range (that is, 7 orders of magnitude), hence they do not vary significantly with the pressure. Finally, Fig. 5d shows the enhancement of the hysteresis at high pressure, where the concentration of O2 and H2O molecules increases.
We also investigated the effect of light on the WS2 NT FET properties, in high vacuum conditions. To this purpose, we exposed the NT to light from a supercontinuum laser. The laser beam has a nominal maximum output power P = 110 mW, which, scaled by the illuminated area of the NT and the laser spot (Aspot = 25 × 10−4 cm2), results in an incident optical power Pinc = 80 nW. Fig. 6a shows the output curves in dark and under laser illumination at different applied Vgs. In particular, at negative Vds, the effect of illumination is negligible at zero and negative high Vgs, while there is an appreciable photocurrent for high positive Vgs. This is due to the dominant p-type conduction for Vds < 0, as already commented above. At positive Vds, ambipolar conduction occurs and photoresponse varies for different Vgs. Fig. 6b reports the photocurrent Iph (defined as the difference between light and dark current) peaks at different Vgs. Different photoresponse (Vds = 2 V) occurs for different Vgs. The best performances are obtained for Vgs = 0 V, for which the highest response and the lower dark current are reported.
To further test the device as a photodetector, we changed the power of the laser beam while measuring the current flowing into the channel. The results are shown in Fig. 6c, where Vds = 2 V and Vgs = 0 V, while Pinc varies from 15 to 75 nW. Iph and the responsivity R (defined as R = Iph/Pinc) are extracted from Fig. 6c and plotted against Pinc in Fig. 6d. Higher responsivity is reported at lower incident power, along with a photocurrent that increases with increasing Pinc. The increasing illumination power leads to an increase in photogenerated charge carriers, which corresponds to an increase in photocurrent. However, at high incident optical power, the enhanced carrier density leads to an increase in the scattering rate favouring electron–hole recombination and limiting the charge carrier mobility, thus resulting in a low responsivity.57,58
Moreover, transfer characteristics at different Vds have been measured in the dark and under illumination (Fig. 6e). The incidence of the laser on the NT favours the desorbing of oxygen molecules from the surface, due to local heating, thus enhancing conduction by electrons, i.e. reversing the atmospheric p-doping effect.59,60
The responsivity as a function of the light wavelength is displayed in Fig. 6f. The peak at λpeak = 643 nm (Epeak = 1.93 eV) falls within the range of energy band gaps of WS2 NTs.61,62
The WS2 nanotubes were then spread onto a Si/SiO2 (tox = 500 nm) wafer, and their locations were determined using optical microscopy. Electrodes were created using electron beam lithography and electron beam evaporation of Ti/Au (20/120 nm) followed by a lift-off process in acetone.
The TEM sample characterization was performed with an FEI Tecnai G2 F20.
For the opto-electronic characterization, the device was tested in a Lakeshore probe-station with integrated optical fiber, connected to a Keithley Parameter Analyzer 4200-SCS. The pressure chamber was controlled by two external pumps (rotative and turbomolecular) and a needle valve.
The white laser source was a NKT Photonics Supercontinuum Laser (450–2400 nm) (SuperK COMPACT), with a nominal output power P = 110 mW and spectrum in the range 450–2400 nm.
This journal is © The Royal Society of Chemistry 2025 |