1 Introduction
Two-dimensional (2D) semiconductors have attracted attention due to their future application in field effect transistors [
1–
6], photodetectors [
7–
10], laser [
11,
12], photoelectrochemical [
13], and other devices [
14,
15]. Molybdenum disulfide (MoS
2), as a typical 2D semiconductor, has direct band gap (~ 1.8 eV) [
16–
18] and high mobility [
1,
5], and shows great potential in optoelectronic devices [
1–
3,
5,
8,
14]. Due to the broken inversion symmetry, inducing the spin–orbit coupling, there is a large splitting of ~ 160 meV over the valence band in monolayer MoS
2, giving rise to A- and B-excitons [
16,
18–
20]. Generally, A-exciton behavior dominates the emission of the MoS
2, while observing B-exciton behavior is difficult due to the forbiddance of its transition [
17,
21]. Meanwhile, achieving p-type conductivity in chemical vapor deposition (CVD) grown MoS
2 monolayer is challenging because of its insurmountable sulfur vacancy defect [
22,
23]. Furthermore, due to the large band gap, photodetectors based on MoS
2 monolayers show a limited detection range in visible light [
8]. Consequently, effectively manipulating the photoelectric properties of MoS
2 and further enriching its properties are important for both fundamental research and applications.
Foreign substitutional doping is efficacious, stable, and exhibits long-range ordering, making it a preferred method for altering the electronic band structure, conduction type, and carrier concentration of pristine materials. Previous reports suggested that vanadium (element V) is an attractive dopant for room-temperature ferromagnetism and has found widespread use in many magneto-optic devices [
24–
27]. The introduction of V element in MoS
2 offers an effective means to modulate both the optical and the electrical properties of the samples, potentially leading to new applications and functionalities. Very recently, CVD growth of V-doped MoS
2 for use in synaptic transistors and quasi-continuously tunable carrier polarity transistors have been reported [
28–
30]. However, to the best of our knowledge, the A- and B-excitons properties and the photoelectric applications of V-doped MoS
2 have not been systematically studied.
In this work, V-doped MoS2 monolayers were grown using an alkali metal-assisted CVD approach. The achievement of V doping in high-quality crystalline monolayers were confirmed by scanning transmission electron microscopy (STEM), X-ray photoelectron spectroscopy (XPS) and Raman spectra. The emergence of p-type conduction in the achieved samples further demonstrated the successful introduction of V atoms into the crystal structure. What's even more interesting is, the distinct B-exciton emission, enhanced p-type conduction and broad spectral response observed in the V-doped MoS2. Our results could offer a novel approach to modulating exciton properties in 2D semiconductors and potentially trigger numerous applications in spintronics.
2 Experimental
2.1 Materials synthesis
A one-zone furnace was used for growth of V-doped MoS2 monolayers. Firstly, one quartz boat (boat 1), loaded with sublimate sulfur powder (S), was placed at the upstream, while another (boat 2), containing a mixture of molybdenum trioxide (MoO3, Alfa Aesar, 99%), vanadium pentoxide (V2O5, Alfa Aesar, 99%), and potassium iodide (KI, Alfa Aesar, 99%), was positioned at the center of a quartz tube (25 cm in diameter and 100 cm long). A long piece of SiO2/Si (270 nm SiO2, 1 cm × 3 cm) substrate was placed above the quartz boat 2. The quartz tube containing the reaction sources and the deposited substrate was placed in the furnace, and the quartz boat 2 was ensured to be located in the center of the furnace. High purity Ar was introduced from the upstream as carrier gas and discharged from the downstream of the quartz tube. Prior to heating, high-rate Ar gas flow (1000 SCCM) was used to remove any active gases. The furnace was then heated to 750–800 ℃ in 30 min and held at that temperature for 5−10 min to grow the V-doped MoS2 monolayers. Throughout the growth process, the Ar gas flow was maintained at 50 SCCM. Subsequent to growth, the furnace was then allowed to cool naturally to room temperature.
2.2 Characterizations
Raman and photoluminescence (PL). The Raman/PL was performed using a μ-PL system (WITec, alpha-300) equipped with a 532 nm argon ion laser.
Atomic force microscope (AFM). The AFM observations were carried out using a Bruker Dimension ICON instrument.
Scanning transmission electron microscopy. The STEM observations were acquired using a JEOL ARM200F microscope.
2.3 Device fabrication and measurements
Source-drain electrodes of the back-gated V-doped MoS2 devices were fabricated using a transfer method. Au with a thickness of 50 nm was prepared on a bare SiO2/Si substrate using photolithography and thermal evaporation deposition methods and subsequently transferred onto the top of the V-doped MoS2 monolayers as source/drain. The optoelectrical properties of the V-doped MoS2 devices were measured using Lake Shore Probe Station and Agilent B1500A semiconductor analyzer.
3 Results and discussion
V-doped MoS
2 monolayers were synthesized using an alkali metal-assisted CVD method as depicted in Fig. 1a. As is the case in previous reports, the chemical reaction between MoO
3, V
2O
5 and the KI results in formation of volatile oxyhalide species [
31,
32]. In brief, solid-phase MoO
3 reacts with solid-phase KI, producing the gaseous-species MoO
2I
2. Similarly, the solid-phases V
2O
5 and KI react with each other to form gaseous-phase VOI
3. The gas–gas phase reactions among MoO
2I
2, VOI
3 and S could occur easily and efficiently. The possible reaction routes in the system are listed as follows:
Figure 1b displays the side- and top-views of the atomic structure of the V-doped MoS2 monolayer, wherein the yellow, blue and gray spheroids represent the S, Mo and V atoms, respectively. V atoms have substituted and occupied the sites of Mo atoms. Figure 1c shows the optical image of the V-doped MoS2 monolayers, indicating the uniform distribution on a large scale (> 100 μm). Additionally, though in a small proportion, V-doped MoS2 bilayers with deeper contrast than monolayers were observed and are marked in Fig. 1c. AFM characterization was used to confirm the thickness of the V-doped MoS2 nanosheet (Fig. 1d), which was measured to be about 1.0 nm (inset in Fig. 1d), indicating a single layer nature.
The atomic structure and chemical composition of the V-doped MoS2 monolayers were investigated by STEM, energy dispersive X-ray spectroscopy (EDS), high-resolution STEM (HRSTEM) and selected-area electron diffraction (SAED). Figure 2a displays a typical low-magnification annular dark-field (ADF) STEM image of a single-domain V-doped MoS2 sample transferred onto a Cu grid. In Fig. 2b, the EDS spectrum obtained from the white dot in Fig. 2a revealed elemental peaks corresponding to S, Mo, and V, confirming the composition of the doped nanosheets. Thus, we could deduce the V mole fraction of the V-doped MoS2 sample, indicating a composition of V0.11Mo0.89S2. Figure S1 shows the EDS spectra taken from more samples, these samples were used for Raman, PL and electrical characterizations.
Figure 2c presents the SAED pattern of the sample, displaying a single set of diffraction spots with sixfold symmetry, confirming the single-crystal nature of V-doped MoS
2. The lattice spacing of the V-doped MoS
2 was obtained using the diffraction peaks yields (100) and was measured to be 0.272 nm, which is similar with that of MoS
2 [
5]. HRSTEM was employed to characterize the atomic structure of the achieved V-doped MoS
2 (Fig. 2d), in which the different atoms could be identified using Z-contrast (Z
2Mo = 1764, Z
2V = 529 and Z
2S + Z
2S = 512). Figure 2e and f provide enlarged experimental and simulation HRSTEM images, in which V atoms are marked with yellow dotted circles. As shown in Fig. 2g, the ADF intensity profile acquired along the blue arrow in Fig. 2e and f also confirms similar results.
XPS was adopted to characterize the chemical states of elements in another V-doped MoS
2 sample. From the XPS survey spectra as shown in Fig. S2, we determined that the V mole fraction (
x) of the tested samples was 9%, indicating a composition of V
0.09Mo
0.91S
2. Figure 3a shows the element binding energy comparison between V-doped MoS
2 (upper half) and pristine MoS
2 (lower half). For MoS
2, five peaks were observed at 233.2, 230.0, 227.2, 164.0 and 162.9 eV. The first two peaks are attributed to the Mo element (Mo
4+ 3d
3/2 and Mo
4+ 3d
5/2), while the remaining peaks belong to S element (S
2− 2s, S
2− 2p
3/2 and S
2− 2p
3/2) [
33]. In the case of V-doped MoS
2, a slight down-shift of the above peaks was observed. Such down-shift indicated that the Fermi level of the achieved V-doped MoS
2 was closer to the valence band compared with that of the pristine MoS
2 monolayer, which demonstrated an increase in p-type doping concentration [
34,
35]. Moreover, two additional peaks located at 516.7 and 524.3 eV were observed, which was attributed to the presence of V atoms. These two peaks were indexed as V
4+ 2p
3/2 (516.7 eV) and V
4+ 2p
1/2 (524.3 eV) [
36].
Raman spectra were used to investigate the vibration modes of the V-doped MoS
2 monolayers (V
0.11Mo
0.89S
2), while pristine MoS
2 was chosen as a representative sample. Pristine MoS
2 exhibited two Raman peaks centered at 384 and 404 cm
−1, corresponding to the
E2g(S-Mo) mode and the
A1g(S-Mo) mode, respectively (bottom panel in Fig. 3b). However, in the Raman spectrum collected from the V-doped MoS
2 monolayer (top panel in Fig. 3b), four additional peaks located at 323, 349, 389 and 406 cm
−1 were observed, indicating the introduction of V atoms [
37,
38]. It is worth noting that, compared with pristine MoS
2, the
E2g(S-Mo) and
A1g(S-Mo) peaks taken from V-doped sample exhibited a slight down-shift of 2.82 and 2.96 cm
−1, respectively, further implying hole doping induced by V atoms [
34]. Raman mapping images of the V-doped MoS
2 monolayer (as shown in Fig. 3c) demonstrated the uniform chemical composition of the tested sample.
The optical properties of the V-doped MoS
2 monolayers were investigated with steady-state PL spectroscopy. Figure 4a shows the optical image of a selected V-doped MoS
2 monolayer with low doping concentration (V
0.05Mo
0.95S
2). The typical PL spectrum collected from pristine MoS
2 (depicted by a black line in Fig. 4d) exhibited two peaks centered at 1.85 and 2.00 eV, corresponding to A- and B-excitons, respectively. Clearly, the A-exciton emission dominated the whole emission behavior. The ratio of the emission integrated intensities between the B- and A-excitons was calculated as
ηP =
IB/
IA = 0.083, where
IA and
IB are the integrated intensities of A- and B-excitons, respectively. In contrast, the spectrum of the V-doped MoS
2 (illustrated by a red line in Fig. 4d) showed distinct B-exciton emission that did not occur in the case of pristine MoS
2. The
ηP of the V-doped MoS
2 was calculated to be 3.13, which was 38 times higher than that occurring in the case of pristine MoS
2 monolayer. In addition, noticeable red-shifts of both the emission peaks of A- (~ 46 meV) and B-excitons (~ 99 meV) were observed in the V-doped MoS
2 monolayer, primarily resulting from doping-induced bandgap narrowing [
39]. The split of A- and B-exciton bands (Δ
BA =
EB−
EA) was determined to be ~ 110 meV for V-doped MoS
2 monolayer, smaller than that of pristine MoS
2 monolayer of ~ 150 meV, indicating that the doped V atoms effectively reduced the spin–orbit splitting between A- and B-excitons behavior in MoS
2 monolayer (Fig. S3). As shown in Fig. S4, for the V
0.02Mo
0.98S
2 and V
0.05Mo
0.95S
2 monolayers (with low doping concentration), the
ηP increased with the increase of V composition. But the sample with high doping concentration (V
0.11Mo
0.89S
2) did not have PL emission. Such PL quenching is contributed to the localized exciton trapped states caused by doped V atoms (Fig. S3) [
39].
Figure 4b and c show the energy-selected PL intensity mapping images of the V-doped MoS
2 at 1.80 and 1.91 eV, respectively. The slightly non-uniform PL mapping is mainly caused by strain effects [
40]. To further explore the spectral features, we studied the excitation power dependence of A- and B-excitonic transitions. Figure 4e presents the 2D pseudo-color plots of PL spectra under illumination by different laser excitation power densities. The log–log plots of integrated PL intensities of the A- (black dots in Fig. 4f) and B-excitons (red dots in Fig. 4f), as functions of excitation power, are presented in Fig. 4f. The relationship between integrated PL intensity
I and excitation power
P can be written as:
I =
Pβ, where
β ≈ 1 implies an exciton-like transition [
41]. Temperature-dependent experiments were conducted to explore the distinct B-exciton emission. Figure 4g exhibits the PL spectra measured in the V-doped MoS
2 monolayer at different temperatures, while the corresponding 2D pseudocolor results are shown in Fig. 4h. The inset in Fig. 4g presents the integrated intensities ratio (
ηP) of A- and B-excitons at different temperatures. As can be seen,
ηP increased gradually with rising temperature, indicating the activation of B-exciton emission was at high temperature. This can be attributed to the enhancement of efficient valley scattering with the assistance of phonons (Fig. S3) [
42]. On the contrary, the abnormal behavior here was attributed to the efficient valley scattering induced by the V atoms doping, which should be largely suppressed at low temperature, subsequently quenching the B-exciton emission.
In addition, we also studied the exciton-phonon interactions of the V-doped MoS2 monolayer. As can be seen, the two emission peaks broadened and shifted towards lower energy with increasing the temperature (Fig. 4i). Such variation can be well interpreted in terms of by Varshni's semi-empirical formula:
where
α and
β are the characteristics of the given material. We found that the energies shift of A- and B-excitons of the V-doped MoS
2 matched well Eq. (4), and the results are shown in Table 1. Such observed unequal shifts of A- and B-excitons were quite similar to those of other TMDCs [
42].
Full width variation at half-maximum (FWHM) of these spectra is shown in Fig. 4i, which can be described by Bose−Einstein type expression:
where Γ
0 represents the inhomogeneous broadening term and the line width at 0 K, Γ
LO represents the exciton-longitudinal optical (LO) phonon coupling, and the
ωLO is the dominant phonon or an average phonon energy. The values of the parameters are summarized in Table 2. The temperature-dependent shifts of the energies and FWHM of the excitonic transitions are mainly due to the lattice constant variations and interactions with relevant acoustic and optical phonons [
43].
To explore the electronic properties and applications of our achieved V-doped MoS2 monolayers, we fabricated back-gated field-effect transistors. Figure 5a provides the schematic diagram of the device. Figure 5b shows the transport properties of the samples with different V composition (MoS2, V0.02Mo0.98S2, V0.05Mo0.95S2 and V0.11Mo0.89S2). It is evident that both the MoS2 and V0.02Mo0.98S2 exhibited typical n-type unipolar transfer behavior. With the introduction of V atoms, transitions from n-type first to ambipolar and then to p-type conduction occurred, revealing that the V atom doped in MoS2 monolayer could effectively implement p-type doping. Ambipolar conduction behavior was observed in V0.05Mo0.95S2 based device, while p-type unipolar conduction behavior was obtained with further increases in the V doping concentration (V0.11Mo0.89S2), which was consistent with the Raman and XPS results. The above results demonstrated that with the increase of V component, the V-doped MoS2 exhibited a transition from non-degenerate (MoS2 and low doping concentration V-doped MoS2) to degenerate (high doping concentration V-doped MoS2) semiconductor.
The photoelectric behavior of the V-doped MoS
2 was then examined and V
0.02Mo
0.98S
2 monolayer based device was selected. Figure 5c shows the output characteristic curves, exhibiting liner results. Figure S5 displays the photoconductive properties of the V-doped MoS
2 monolayer device under illumination by 450 and 633 nm lasers with different power densities. It is clear that, the channel current increased gradually with increases in the power densities (
Pin), which demonstrated that the V-doped MoS
2 monolayer has good visible light response. It is worth noting that our V-doped MoS
2 monolayer device exhibited significant light response under illumination by near infrared light at a wavelength of 980 nm; the
Ids–
Vds curves are shown in Fig. 5d. The obtained extracted photocurrents (
Iph =
Ilight −
Idark) under illumination by different laser power are shown in Fig. 5e, fitted with an equation of
Iph =
aPα. The fitted parameter of
a were 3.36 × 10
–10, 9.51 × 10
−10 and 1.57 × 10
−11; and
α were 0.65, 0.32 and 0.50 under illumination by 450, 633 and 980 nm lasers, respectively. The responsivity (
R) was calculated by the equation of
R =
Iph/(
P ×
A), where
Iph is the photocurrent,
P is the incident light power density, and
A is the effective area of the device channel (
A = 1207.44 μm). As shown in Fig. 5f, the responsivity of the device based on V-doped MoS
2 monolayer reached up to 0.23, 0.25 and 1.41 × 10
−4 A/W under illumination by 450, 633 and 980 nm lasers, respectively. Detectivity (
D*) is used to characterize the sensitivity of a photodetector, which can be estimated by
D* =
RA1/2/(2e ×
Idark)
1/2. The maximum estimated
D* values were 6.59 × 10
8, 1.31 × 10
9 and 1.07 × 10
6 Jones under illumination by 450, 633 and 980 nm lasers, respectively (Fig. 5g). The summarized result of the responsivity and detectivity is shown in Table 3. The performance comparison between the photodetectors based on V-doped MoS
2 monolayer and other reported photodetectors is shown in Table S1, which demonstrates that our photodetectors have moderate performance [
44–
52]. Figure 5h shows the switching behavior of the photodetector measured under light irradiation with visible light 450 (green line), 633 (red line) and near-infrared light 980 nm (violet line) lasers, indicating that the device could be switched effectively between high resistance and low resistance states. As shown in Fig. S3, in V-doped MoS
2 monolayers, an accept band lever caused by V doping between the conduction band and valence band can be introduced, which can facilitate the absorption of short-wavelength light, enabling near-infrared light detection. The above results indicate that the V-doped MoS
2 device exhibits significant broad spectral response.
4 Conclusion
In summary, a series of substitutional V-doped MoS2 monolayers were synthesized using an alkali metal-assisted CVD growth method. The samples were systematically characterized by XPS, Raman, STEM and electrical transport characterizations, confirming that V atoms were uniformly doped into the MoS2 samples. Interestingly, enhanced B-exciton emission was observed in the doped samples and the emission behaviors were systematically studied using steady temperature-dependent PL experiments. Electrical transport measurements indicated that enhanced p-type conduction occurred in the achieved V-doped MoS2 monolayers. Moreover, the photodetector based on V-doped MoS2 monolayer showed broad spectral response from visible to near-infrared light. The synthesized V-doped MoS2 nanosheets can provide new material platform for spintronics related fundamental research and device applications.