Abhinandan Patra
and
Chandra Sekhar Rout
*
Centre for Nano and Material Sciences, Jain University, Jain Global Campus, Jakkasandra, Ramanagaram, Bangalore-562112, India. E-mail: r.chandrasekhar@jainuniversity.ac.in; csrout@gmail.com
First published on 2nd October 2020
The strong in-plane anisotropy and quasi-1D electronic structures of transition-metal trichalcogenides (MX3; M = group IV or V transition metal; X = S, Se, or Te) have pronounced influence on moulding the properties of MX3 materials. In particular, the infinite trigonal MX6 prismatic chains running parallel to the b-axis are responsible for the manifestation of anisotropy in these materials. Several marvellous properties, such as inherent electronic, optical, electrical, magnetic, superconductivity, and charge density wave (CDW) transport properties, make transition-metal trichalcogenides (TMTCs) stand out from other 2D materials in the fields of nanoscience and materials science. In addition, with the assistance of pressure, temperature, and tensile strain, these materials and their exceptional properties can be tuned to a superior extent. The robust anisotropy and incommensurable properties make the MX3 family fit for accomplishing quite a lot of compelling applications in the areas of field effect transistors (FETs), solar and fuel cells, lithium-ion batteries, thermoelectricity, etc. In this review article, a precise audit of the distinctive crystal structures, static and dynamic properties, efficacious synthesis schemes, and enthralling applications of quasi-1D MX3 materials is made.
The weak van der Waals amalgamated crystal structures of MX3 architecturally and chemically constitute a very distinct family of compounds, where M (group IV, V, or VI) represents a metal atom and X is a chalcogen atom (S, Se, or Te), and the M–X bond is ionic–covalent in nature.6–9 The compact in-plane anisotropy and quasi one-dimensional properties of monolayer and few-layer MX3 contribute a huge amount to configuring every single property of these materials. The prismatic MX6 chains of MX3 are extended along the b-axis to provide strong anisotropy. Pressure, temperature, and tensile strain can tune the electronic properties, and charge density wave transportation plays a pivotal role in making metal trichalcogenides a prime material of interest in the nanoscience and materials science communities. Instead of bulk or layered MX3, monolayer metal trichalcogenides are of considerable interest in both experimental and theoretical investigations. Monolayer TiS3 and TaS3, fabricated through mechanical exfoliation, show amazing superconductivity and charge density wave phenomena. This breakthrough accounts for their pronounced, unparalleled, and plentiful applications in the field of nanoelectronics.
Wide amounts of research have been ongoing concerning the flexible fabrication of MX3 materials, and attempts to shed several layers to obtain monolayers have also been made in the past few years through many emerging top-down synthesis methods. These new-fangled methods include chemical vapour deposition (CVD), chemical vapour transport (CVT), solvothermal methods, mechanical and chemical exfoliation, etc. As far as this review is concerned, we report cumulative information regarding crystal structures, elementary properties, synthesis procedures, applications, and experimental and theoretical developments related to MX3-based materials. Firstly, emblematic crystal structures and typical electrical, optical, magnetic, and CDW transport property characteristics are thoroughly discussed. Secondly, the fabrication of these materials is looked into. Lastly, the wide variety of applications of MX3, including field emission transistors,124–126,142 solar and fuel cells,139,143–148 photo-detectors and -sensors,33,139,150,151 lithium ion batteries,134,152–155 and thermoelectricity,156–161 are discussed with proper reference to previous studies.
Group | Transition metal | S | Se | Te |
---|---|---|---|---|
IV | Ti | Monoclinic | No report | No report |
a = 4.973 Å, b = 3.433 Å, c = 8.714 Å; β = 97.74° | ||||
Diamagnetic; n-type semiconductor; band gap: 0.8–1 eV; Hall mobility: 30 cm2 V−1 s−1 (ref. 2 and 12) | ||||
Zr | Monoclinic | Monoclinic | Monoclinic | |
a = 5.06 Å, b = 3.6 Å, c = 8.95 Å; β = 98.4° | a = 5.411 Å, b = 3.749 Å, c = 9.44 Å; β = 97.45° | a = 5.863 Å, b = 3.923 Å, c = 10.089 Å; β = 97.74° | ||
Diamagnetic; p-type semiconductor; band gap: 1.8–2.5 eV; Hall mobility: 26 cm2 V−1 s−1 (ref. 12 and 52) | Diamagnetic; n-type semiconductor; band gap: 1.1 eV; Hall mobility: 3.9 cm2 V−1 s−1 (ref. 29 and 164) | Semi-metallic; TCDW = 63 K; Tc (superconductor) = 2 K (ref. 13 and 165) | ||
Hf | Monoclinic | Monoclinic | Monoclinic | |
a = 5.08 Å, b = 3.58 Å, c = 8.96 Å; β = 98.4° | a = 5.31 Å, b = 3.73 Å, c = 9.525 Å; β = 97.18° | a = 5.879 Å, b = 3.902 Å, c = 10.056 Å; β = 97.98° | ||
Diamagnetic; p-type semiconductor; band gap: 1.9–3.1 eV; Hall mobility: 26 cm2 V−1 s−1 (ref. 52 and 165) | Diamagnetic; p-type semiconductor; band gap: 1.0 eV (ref. 3 and 7) | Semi-metallic; TCDW = 93 K; Tc (superconductor) = 4.3 K (ref. 151 and 166) | ||
V | Nb | Triclinic | Monoclinic | No report |
a = 4.963 Å, b = 6.730 Å, c = 9.144 Å; β = 97.17° | a = 10.009 Å, b = 3.48 Å, c = 15.629 Å; β = 109.47° | |||
Diamagnetic; semiconductor; band gap: 0.8–1.1 eV (ref. 55 and 83) | Metal; TCDW = 145 K and 59 K (ref. 75, 79, 94 and 164) | |||
Monoclinic | ||||
a = 9.68 Å, b = 3.37 Å, c = 14.83 Å; β = 109.9° | ||||
Diamagnetic; semiconductor; band gap: 0.3–1.0 eV; Hall mobility: 10–2400 cm2 V−1 s−1 (ref. 12 and 13) | ||||
Ta | Orthorhombic | Monoclinic | No report | |
a = 36.804 Å, b = 15.173 Å, c = 3.34 Å | a = 10.402 Å, b = 3.495 Å, c = 9.829 Å; β = 106.26° | |||
Metal8 | Metal; superconductor; Tc = 2.1 K (ref. 55, 79 and 164) | |||
Monoclinic | ||||
a = 9.515 Å, b = 3.3412 Å, c = 14.912 Å; β = 109.99° | ||||
Metal166 |
In most transition metal trichalcogenides, MX6 trigonal prisms (M as the central atom and X dwelling in the prism triangular base) are stacked so as to obtain MX3 trigonal prismatic chains, and these chains (aligned parallel to the b-axis of the unit cell) are covalently bonded through van der Waals forces [Fig. 1a].6 Due to this, these chalcogenides have the tendency to show 1D characteristics along with anisotropy. Depending on the X–X bond lengths in these MX3 crystals, three types of chain arrangement are possible,2–4 which are outlined below:
Fig. 1 (a) The crystal structure of anisotropic TiS3 indicating the bond lengths between titanium and sulfur along the b-axis and a-axis, with shorter bonds along the b-axis; reproduced with permission from ref. 6, copyright: 2015, Springer Nature. (b) The calculated electronic energy band structure along the symmetry directions of the Brillouin zone (Γ–Y–C–X–Γ) and the projected density of states (PDOS) of the optimized structure of ML TiS3. The direct band gap Eg of ML TiS3 is indicated by the red arrows. The Fermi level is set to zero energy (the dashed line); reprinted with permission from ref. 8, copyright: 2019, American Chemical Society. (c) The charge distributions of the VBM and CBM states of monolayer TiS3; republished with permission from ref. 9, permission conveyed through Copyright Clearance Center, Inc. |
➢ A one-type chain arrangement is present in ZrSe3, with Se–Se bond lengths of 0.234 nm and (Se2)2− pairs forming in the trigonal base.
➢ A two-type chain arrangement is present in TaSe3, with Se–Se bond lengths of 0.258 nm and 0.291 nm (exceeding the value of ZrSe3). TiS3 crystals are an example of this type, with the arrangement: Ti4+S2−(S2)2−.
➢ A three-type chain arrangement is present in NbSe3 unit cells, with three Se–Se bonds at the base of the trigonal prismatic chains, short (III), mean (I), and long (II), having bond lengths of 0.237 nm, 0.248 nm, and 0.291 nm, respectively. The short bond length (0.237 nm) is ascribed to (Se2)2−, which supports the weakening of bonds between two Se atoms and strengthens the bonds between Nb and Se atoms. These types of arrangement create an intermediate situation with Nb4+ and Nb5+.
All members of the ZrX3 and HfX3 family take the form of monoclinic crystals, whose lattice parameters are summarized in Table 1. As discussed, the layered-type crystal structure of MX3 has the metal ion (M) in the centre of a slanted trigonal prism; stacked trigonal faces form separated columns that arrange along the b-axis, and the base triangle has a shape with two sides much longer than the third one.12–15 This criterion is satisfied by the formulation (X2−)(X22−), with the metal oxidation state being M4+ (d0), leading to empty d-block bands. Due to this fact, most MX3 compounds are reported to be semiconducting, with a band gap of ∼1–2 eV, except ZrTe3 and HfTe3, which are semi-metallic with superconducting properties at low temperatures.12–15 The crystal structure of ZrTe3 is presented in Fig. 4a and b.15
Experimental reports by Khatibi et al. showed that TiS3 emits near infrared (NIR) light centred at about 0.91 eV (1360 nm) with undeviated polarized anisotropic photoluminescence with a radiation life span of 210 ps (Fig. 2a–d).18 The dependence of emission on the excitation power and temperature demonstrated the dominant behaviour of free and bound electron–hole pairs over excitonic radiation at room and low temperature, correspondingly. TiS3 is reported to show excellent stable emission compared to other 2D materials, such as black phosphorus, MoTe2, etc.18 TiS3 is one of the appealing materials in the MX3 family due to its anisotropic optical properties. By using a force field along either the a-axis or b-axis and carrying out detection at intermediate angles between a and b, the anisotropic optical properties of 2D TiS3 are investigated.6 The anisotropic optical properties survey demonstrated strong emission anisotropy along the b-axis (25% higher emission intensity than along the a-axis). Island et al. inspected the strong in-plane anisotropy of TiS3 through angle-resolved polarization Raman spectroscopy (ARPRS).6 Compared to at room temperature, at 25 K, the in-plane conductivity in terms of anisotropy increases 2.09-fold. Fig. 2e and f6 shows Raman studies of TiS3 ribbons under different polarization conditions. Taking the b-axis as the favourable growth axis, four distinguishable Raman peaks (excluding one peak due to the silicon substrate) due to TiS3 were observed from an isolated nanoribbon on the SiO2/Si substrate. It is reported that the intensities of all the modes can be altered via changing the polarization angle, whereas the peak close to 370 cm−1 is found to be reliant on alignment between the polarization of the excitation laser and the b-axis. Furthermore, the anisotropy of the photosensitivity is verified based on the transmission of TiS3 through characteristic angles (Fig. 2h).6 The transmission reached a minimum value when the angle between the polarized excitation light source and the elongated side of the flakes was 180° (parallel to each other) in correspondence with the b-axis. The collected optical transmission data confirmed that TiS3 flakes exhibit robust linear dichroism, where the properties along the b-axis dominate those of the a-axis by as much as 30-fold, which is commendable for a 2D material.6 Compared to common semiconductors and other 2D TMDCs, TiS3 nanoribbons have a high exciton binding energy, which makes them an excellent applicant for use in optoelectronic devices and various related applications. Iyikanat et al. calculated the band gap of TiS3 using the PBE (Perdew–Burke–Ernzerhof) approximation and HSE06 (Heyd–Scuseria–Ernzerhof functional) correction to be 0.23 eV and 1.05 eV, respectively.19 The effects of S, Ti, TiS, and double S vacancies in TiS3 on its optical and electrical properties were also discussed based on DFT calculations by the above group (Fig. 3a–d).19 Of the four types of vacancies, except for S vacancies, the others lose their semiconducting properties and become metallic, resulting in a net magnetic moment. However, the low formation energy in the case of S vacancies assists in the opening of the energy gap of TiS3 monolayers. The calculated magnetic moments for the Ti, TiS, and double S vacancies of TiS3 are found to be 0.5, 0.8, and 0.3 μB per supercell, respectively.19 Kang et al. reported that the microelectronic properties of TiS3 nanoribbons strongly depend on the side, whether it is a or b.20 a-TiS3 nanoribbons are reported to have the properties of a metal, having a width-dependent band gap, whereas b-TiS3 is a semiconductor with a direct energy gap, which can be regulated by strain, is nearly self-governing, and does not hinge on the ribbon width.20
Fig. 2 The optical properties of TiS3 flakes. (a) An optical microscope image showing the a and b axis directions and infrared light emission from TiS3. (b) Photoluminescence and (c) time-resolved photoluminescence spectra and (d) the photostability of a TiS3 flake compared to black phosphorus. Reproduced with permission from ref. 18, copyright: 2019, IOP Science. (e) Raman spectra of TiS3 ribbons with horizontal excitation and detection polarization. (f) The intensity of the 370 cm−1 Raman peak of a 3 nm-thick flake (3–4 layers) as a function of the excitation polarization angle (ref. 18). (g) The transmittance of the red, green, and blue channels as a function of the excitation polarization angle. (h) Calculated absorption spectra when the field is aligned parallel to the b-axis (dashed line) and the a-axis (solid line) with the inset showing the transmittance in the a–b plane for red (1.9 eV), green (2.4 eV), and blue (2.72 eV) excitation energies. Reproduced with permission from ref. 6; copyright: 2015, Springer Nature. |
Fig. 3 Top views of relaxed monolayer TiS3 with (a) S, (b) Ti, (c) double S, and (d) TiS vacancies. The black atoms illustrate removed atoms, and the dashed circles show the initial positions of the displaced atoms. Reproduced with permission from ref. 19, copyright: 2015, American Chemical Society. (e) An optical photograph of HfS3 needles grown on the inner walls of a quartz ampoule, an optical image of ZrS3 flakes exfoliated onto SiO2 substrates with the b-axis direction shown, and an AFM image of an exfoliated TiS3 flake with a thickness of ∼5 nm. (f) A polar plot of mode III at 372 (red), 320 (black), and 321 (blue) cm−1, which corresponds to thin TiS3, ZrS3, and HfS3 flakes, respectively. Republished with permission from ref. 24, permission conveyed through Copyright Clearance Center, Inc. |
ZrS3 is reported to be a p-type semiconductor, with resistivity of 15 Ω cm at room temperature, a direct optical energy gap of 1.8–2.5 eV, and two indirect optical energy gaps of 2.055 eV (when the field is along the b-axis) and 2.058 eV (when the field is normal to the b-axis) at 4.2 K.4 Angle-resolved photoemission and optical measurements of ZrX3 (X = S or Se) and HfSe3 predict that, due to spin–orbit interactions, there is splitting of the highest occupied band into two parallel bands. The split energy rises from S to Se in the above compound, and the absorption spectra are found to be highly anisotropic considering the phonon modes and electronic transitions.21 Schairer et al. evaluated the optical gap energies of ZrS3 and HfS3 to be 2.8 and 3.1 eV.12 In another report, Jandl et al. confirmed the presence of two polytypes in accordance with the recombination of phonon replicas of excitons localized in band tails.22 Pant et al. reported the indirect band gap of ZrS3 to be 1.88 eV, and angle-resolved photoluminescence (PL) spectroscopy studies of ZrS3 nanosheets demonstrated that they are highly anisotropic, which is manifested by the large PL intensity variation with polarization direction.23 Raman spectroscopy and angle-resolved studies are fast and non-destructive optical methods to probe the anisotropic nature of MX3. In this regard, Kong et al. collected Raman data from MX3 flakes (TiS3, ZrS3, and HfS3) with the longer edge (the anisotropic b-axis) aligned parallel to both the polarization direction of the laser and the polarization exposure path of the Raman spectrometer.24 To the contrary, for TiNbS3 alloy trichalcogenides, it is not possible to determine the anisotropic direction due to the loss of anisotropy and common defects present in the random distribution of quasi-1D MX6 chains.24 Wang et al. verified the anisotropic effects in ZrS3 via angle-resolved absorption and angle-resolved Raman spectroscopy, and angle-resolved photocurrent studies, which gave a vibrant impression that all the properties were boosted along the b-axis. Angle-resolved spectroscopy studies revealed dichroic ratios of 1.73 and 1.14 for the ZrS3 nanoribbons when excited by different laser source wavelengths, i.e., 450 nm and 532 nm, respectively, considering the photocurrent density.25 The change in dichroic ratio is owing to the variation of the offset angle from the b-axis. Jin et al. showed the widening and trivial shifting of peaks in the Raman spectra due to the phonon confinement effect in the cases of ZrS3 and HfS3 nanobelts.26 Likewise, TiS3 Raman spectra collected by Pawbake et al. showed the shifting of the peaks towards lower wavenumbers upon an increase in temperature (88 K to 570 K), which is attributed to thermal expansion of the lattice and anharmonic vibration.27 Apart from temperature, pressure-dependent Raman spectroscopy studies were done by Wu et al., which describe the unorthodox negative pressure dependence of the AS–Sg S–S molecular mode in contrast to the expected stiffening of other peaks. Numerous modes of TiS3 are reported to be doubly degenerate at ambient pressure, whereas at high pressure, this effect vanishes.28
Patel et al. calculated the direct and indirect energy gaps of ZrSe3 to be around 1.47 eV and 1.1 eV, respectively. They also found that by means of an escalation in temperature in the vicinity of 303–403 K, the anisotropy and resistivity could be decreased and increased, respectively.29 In agreement with quasi particle self-energy correction, Zhou et al. found the indirect band gap of ZrSe3 to be 1.63 eV.30 Felser et al. established a dependent relationship between the density of states and the shape of the Fermi surface in Te–Te interprism interactions in ZrTe3.14 Zeng et al. showed that growing strain in HfS3 resulted in a switch from an indirect to a direct band gap of 2.2 eV, whereas Tao and his group calculated the indirect and direct optical energy gaps of HfS3 nanobelts to be 1.73 and 2.19 eV, respectively.31,32 As shown by photoluminescence (PL) studies, the nanobelts displayed strong emission at 483, 540, and 600 nm in response to excitation at 400 nm.32 Likewise, Zhao et al. calculated the band gaps for Zrs3, Zrse3, Hfs3, and Hfse3 to be 1.13, 0.23, 1.08, and 0.05 eV, respectively. According to their study, the p state of the chalcogen (governing the VBM) and the d state of the transition metal (governing the CBM) combined to form the band gap of the MX3 monolayers.33
Trisulphides, for example ZrS3- and HfS3-based structures, are reported to exhibit two dozen normal modes at the centre of the Brillouin zone and belong to the C2h space group.26,34,35 Each mode is interconnected with a clutch of atomic vibrations represented by Ag, Bg, Au, and Bu. Among these, the Au and Bg vibrations are parallel to the MX6 trigonal prism chains, and the Ag and Au bands are perpendicular to the chains. Gleason et al. investigated the pressure-induced phases of ZrTe3 and also analysed the temperature-dependent Raman spectra.36 It is reported that some specific phonon modes endure a dramatic linewidth reduction near the charge density wave temperature (TCDW), indicating the strong coupling of phonons with electronic degrees of freedom concomitant with the CDW (Fig. 4). Typical Raman spectra from a ZrTe3 crystal are shown in Fig. 4c at 295 K and 6 K with normal mode displacement patterns. The lower-energy modes (ω1–ω3) and higher-energy modes (ω4–ω6) are allocated primarily to vibrations of the trigonal prismatic rods, inwardly and outwardly, respectively. Changes in the widths of the Raman bands at different temperatures (Fig. 4d and e)36 are observed, with the large reductions in the widths of ω4 and ω5 demonstrating strong coupling between these bands. Similarly, the downfall of the long-range-order (LRO) of the rods is attributed to the suppression of phonon bands allied to internal vibrations of the ZrTe3 prismatic rods at pressures above 10 kbar (Fig. 4f and g).36
Fig. 4 The crystal structure of ZrTe3: (a) quasi-one-dimensional trigonal prism packing along the b axis and (b) a quasi-two-dimensional ZrTe3 layer along the a–c plane; reprinted with permission from ref. 15, copyright: 2013, American Physical Society. (c) Raman spectra of ZrTe3 at 6 K and 295 K. (d and e) Temperature-dependent Raman spectra at ambient pressure and (f and g) pressure-dependent Raman spectra at T = 3 K. Reprinted with permission from ref. 36, copyright; 2013, American Physical Society. |
In quasi-1D conductors, the condensations of electrons into CDWs arises due to Fermi surface instability, and they form a deformable medium that affects their overall static and dynamic properties, giving rise to metallic stability and hysteresis. Low-dimensional materials, including anisotropic 2D MX3 layered structures, from time to time lose their LRO and orthodox symmetry, resulting in Fermi surface instability and forming a presumed charge density wave (CDW).38,39 ZrTe3 is one attention-grabbing material in the type IV–X MX3 family, which is reported to have anisotropic nature, with a CDW at 63 K and superconductivity at 2 K.14,15,40–42 Other reports on ZrTe3 revealed a CDW transition at 63–70 K due to resistivity inconsistency along the a-axis, but not along the conventional anisotropic b-axis.41,43 Canadell et al. predicted that ZrTe3 is a type-B structure with shorter X–X contacts between adjacent MX3 units. The type-B structure of ZrTe3 plays a crucial role in determining the semi-metallic nature of the ZrTe3 chains.13 Zhu et al. reported a comparison of the electrical and superconducting properties of ZrTe3 single crystals prepared at low (735 °C) and high (950 °C) temperatures.15 From resistive, colorimetric, and magnetic studies, bulk superconductivity is perceived in the HT-ZrTe3 crystals with the help of doping after a certain temperature, i.e., 4 K, but not in the LT-ZrTe3 crystals. The difference in electrical properties is attributed to the suppression of CDWs through growth-induced structural disorder at high temperature.15 Polarized Raman measurements and first principles calculations demonstrated that precise structural vibrational arrangements from longitudinal distortions of the Te(II)–Te(III) chains had a strong association with the conduction of electrons, leading to the formation of CDWs in ZrTe3.44–46 Pressure-dependent electrical properties investigations revealed that the CDW transition temperature (TCDW) of ZrTe3 was initially amplified, then diminished at 2 GPa, and quickly vanished at 5 GPa, but superconductivity was shown at a pressure level of up to 11 GPa.47 However, Hoesch et al. performed low-temperature and high-pressure single crystal X-ray diffraction studies along with ab initio DFT studies to show that the reported abrupt demise of the CDW phase is due to instability in the Fermi surface above 5 GPa.48 Zhu et al. reported the presence of different bands, including bands with flat and dispersive profiles, along with the amalgamation of chalcogen (high mobility) and metal (low mobility) derived bands in the Fermi surface of ZrTe3. Due to the suppression of long-range CDW order, superconductivity materialises in Se-doped ZrTe3.49 The superconductivity critical temperature (Tc) increased up to 4 K, but after that, additional Se doping caused a reduction in Tc and filamentary superconductivity in ZrTe3−xSex (0 ≤ x ≤ 0.1).49 Polycrystalline ZrTe3 is testified to show a superconducting transition temperature of 5.2 K and a TCDW value of ∼63 K, both co-existing at ambient pressure.50 The intercalation of Ag and Cu into ZrTe3 favoured an escalation in electrical conductivity and resulted in CDW anomalies, which is confirmed from DC magnetisation results, whereas it had no role on TCDW and Tc.50 Li et al. reported CDW formation at TCDW = 93 K with typical anisotropy.51 Conductivity measurements parallel to the a-axis (ρa) and b-axis (ρb) of a HfTe3 crystal gave a direct indication of the disorder-related superconductor fluctuations in ρb at 4.3 K. Also, a superconducting phase with Tc = 1.7 K co-exists with a lower TCDW value of 80 K in a polycrystalline sample, whose generation could be attributed to enriched disorder scattering or unintended carrier doping.51 A combined experimental and computational study by Hu et al. explored a new concept, which emphasizes that phonon–electron coupling helps in the construction of CDWs in ZrTe3. However, their study depicted that the breaching of fractional electronic gaps in the CDW state depends upon the phonon–electron momentum and coupling.44 The resistivity and Hall mobility at room temperature of ZrSe3 were calculated by Ikari et al. to be about 9 × 102 Ω cm and 0.45 cm2 V−1 s−1, respectively.52
Lai et al. reported a comparative investigation into the magnetic properties and tensile strain response of N-a(b)-TiS3 nanoribbons, where a-TiS3 and b-TiS3 are nanoribbons reviewed along either the a- or b-axis and N indicates the number of Ti atoms in the monoclinic cell of the ribbons.53 It was found that the magnetic ground state was a ferromagnetic (FM) metal when N was equal to an odd number, whereas it behaved like an antiferromagnetic (AFM) metal when N was equal to an even number for N-a-TiS3 nanoribbons. Tensile strain (6%) could be used to tune 9-a(b)-TiS3 nanoribbons from a FM metal to a half metal. Similarly, tensile strain (4%) also could cause an AFM to FM transition in 10-a-TiS3 nanoribbons.53
Material | Crystal structure a, b, c (Å); α, β, γ (°) | Synthesis conditions | Nb–Nb (Å) | Ref. no. |
---|---|---|---|---|
NbS3-I | 4.963, 6.730, 9.144; 90, 97.17, 90 | NbS2Cl2; 588 °C (source)/569 °C (sink); 48 h; slow cooling | 3.045, 3.702 | 57 |
NbS3-HP | 9.68, 3.37, 14.83; 90, 109.9, 90 | Nb + S; 700 °C at 2 GPa; 0.5 h | 3.370 | 131 |
NbS3-II | 9.9, 3.4, 18.3; 90, 97, 90 | Nb + S; 600 °C (source)/580 °C; 15 days | — | 61 and 99 |
NbS3-II | 9.1–9.6, 18.7–19.9, 3.4; 90, 97–98, 90 | Nb + S; 500 °C | — | 60 |
NbS3-III | ∼5, —, ∼9; 90, 98–99, 90 | Nb2 + S; 550 °C; 21 days; 400 °C, 48 h (48 h, air quenching) | — | 67 |
NbS3-IV | 6.7515(5), 4.9736(4), 18.1315(13); 90, 90.116(2), 90 | Nb + S, I2 transport, 70 °C (source)/570 °C (sink), 10 days | 3.0448(8), 3.7087(8) | 55 |
NbS3-V | 4.950(5), 3.358(4), 9.079(10); 90, 97.35(2), 90 | Nb + S, 10% S, I2 transport, 670 °C (source)/570 °C (sink), 10 days | 3.358(4) | 55 |
Fig. 5 The crystal structure of NbS3-I, NbS3-IV, and NbS3-V: unit cells, chain cross-sections, layers, and perspective views. Reproduced with permission from ref. 55, copyright: 2017, Application Infrastructure Provider. |
In 1975, single-crystal NbSe3 was prepared by Meerschaut et al., and single-crystal XRD studies established the monoclinic structure, consisting of inestimable selenium trigonal prismatic chains loaded on top of each other with shared triangular faces.68–70 Similar to other MX3-based 2D materials, the structural arrangement with strong chemical bonding anisotropy makes NbSe3 crystals a tempting material with novel physical and chemical properties. The Se–Se bond lengths are testified to be 2.37 Å and 2.49 Å, respectively, with CDWs at 145 K and 59 K, and the Nb–Se bonds are strongly covalent–ionic in nature.70–73 From scanning tunnelling microscopy studies of NbSe3, it is clear that there are three distinguishable chains (I, II, and III) in the NbSe3 unit cell at all temperature.74 The III chains are identified by their association with the CDW modulation vector q1, and the two remaining chains are named I and II.
Among TaX3-based TMTCs, TaS3 and TaSe3 have been investigated due to their various fundamental microelectronic properties, which range from insulating to metallic conducting.65,75 TaS3 was first described by Blitz and Kocher in 1938 and later through XRD studies implemented by Jellinek in 1962.76,77 There are two commonly referenced structures of TaS3, a monoclinic state (m-TaS3) and an orthorhombic state (o-TaS3) (structure shown in Fig. 6a)104 (Table 1). Meerschaut et al. reported the complete structure of m-TaS3 with the space group P21/m, and the lattice constants are a = 9.515(2) Å, b = 3.3412(4) Å, c = 14.912(2) Å, and β = 109.99°.75 o-TaS3 forms in the space group C2221 and the lattice constants are a = 36.804 Å, b = 15.173 Å, and c = 3.340 Å.65,77 A high-pressure mhp-TaS3 phase with the identical space group to m-TaS3 has similar lattice parameters, expect there is a deviation of β by 3°.65 m-TaS3 is a low-dimensional conductor with inestimable chains of tantalum atoms along the b-axis of the primitive cell. The remarkable properties of bulk and few-layer TaS3 are directly correlated to the crystal structure, specifically the chains of tantalum atoms and the b-direction of the unit cell, which leads to quasi-1D properties. TaSe3 belongs to the V–X group of MX3 with a monoclinic crystal structure, in which the unit cell consists of stacks of Ta atoms, each of which are fused to three Se atoms above and below the b-axis.78,136 The monoclinic crystal structure of TaSe3 was first recorded by Bjerkelund and Kjekshus in 1965 with a = 10.402 Å, b = 3.495 Å, c = 9.829 Å, and β = 106.26°.79,80 The inter-planar distance (nearly 4 Å) is more than the distance between the Ta atoms (3.495 Å), which gives clear evidence for the quasi-1D properties of TaSe3.79
Fig. 6 (a) The orthorhombic structure of TaS3. (b and c) FESEM images of TaS3 nanobelts. (d) Temperature-dependent resistance changes of a single nanoribbon, with the inset showing a fabricated device. (e) Current–voltage curves and (f) differential conductance as a function of voltage measured at specific temperatures, with the inset of (f) showing the temperature dependence of the threshold voltage. Republished with permission from ref. 104, permission conveyed through Copyright Clearance Center, Inc. |
Monoclinic NbSe3 consists of three equidistantly placed metal chains with different (Se2)2− groups, and it is reported to show metallic conductivity with two charge density wave temperatures of 145 K and 59 K.69,87,88 Chaussy et al. reported two phase-transition temperatures of NbSe3 crystals at 145 K and 59 K based on electrical resistivity, magnetic susceptibility, and heat capacity measurements.87 As per their revisions, when the temperature decreased, the resistivity (ρ) decreased, showing the behaviour of a metal, and saturation was reached below 10 K. The maximum specific heat capacity was attained at 49 K and gradually decreased at T = 0 K. It was found to be diamagnetic at 4.2 K, and for fibres and powder samples of NbSe3, the magnetization values were equal to 1.35 × 10−7 emu g−1 (attributed to a parallel orientation to the magnetic field) and 5 × 10−7 emu g−1 (attributed to a random orientation to the magnetic field), respectively.87 Whiskers of NbSe3 and Fe-doped NbSe3 nanowires also showed two anomalies in resistivity that were adjunct to the CDW transitions at 140 K and 50 K.89 The successful doping of Fe atoms can be observed based on the strengthening of both the threshold fields, ET1 and ET2. Four-probe resistivity measurements of single-crystal NbSe3 nanowires showed the expected CDW transitions at T1 = 142 K and T2 = 58 K, and there was no magnetoresistance above the higher TCDW value and positive magnetoresistance below the lower TCDW value.72 Ido et al. explored the effects of pressure on CDW formation and superconductivity in NbSe3 via resistivity and diamagnetic measurements.90 It was observed that both the CDW transition temperatures T1 and T2 decreased steadily with an increase in pressure, while they later changed abruptly above 6 kbar pressure and tended to zero above Pc = 7.5 kbar (superconductivity appeared). The change in superconductivity arose from a clampdown on electron–phonon coupling.90 Latyshev et al. showed that the induction of oscillations into non-linear CDW conductivity could be attributed to columnar defects in NbSe3, which fluctuated with a magnetic field when the field was oriented parallel to the axes of the defects.73,91 Due to the CDW transitions, there were ample signs of electron–phonon scattering affecting the transport properties relating to the thermal conductivity of the lattice of NbSe3 nanowires.92 Ong et al. performed Hall measurements of NbSe3 and showed that at both TCDW values there is an increase in Hall resistivity (RH).93 But this was self-regulating below 3 K, and with an increase in the field value, it saturated at 2.3 × 10−6 m3 C−1. Surprisingly, the value of RH was 4.1 × 10−7 m3 C−1 at 2 K.93
Resistivity measurements and electron diffraction studies of m-TaS3 and o-TaS3 showed two transition temperatures (240 K and 160 K), which were interpreted as succeeding Peierls transitions on the diverse chain types of TaS3.94 For o-TaS3, as the temperature dropped below the ambient temperature, the resistance gradually rose up to 230 K and then increased sharply. However, for m-TaS3, this happened in a typical manner, i.e., there was a sequential decrease and increase in resistance at 270, 220, and 180 K.94 The Peierls transition temperature, the temperature at which the slope of R vs. T attains its maximum, was also shown by Roucau et al. Peierls transitions in quasi-1D conductors are associated with intrinsic superstructures.94 Polarized Raman scattering studies of o-TaS3 demonstrated that at the Fermi surface, the formation of a CDW gap was held responsible for the reduction in free carriers.95 Thus, the scattering intensity decreased from interband processes along with an increase in the phonon energy. Nanosized TaS3 samples showed step-like conductivity as a function of strain, demonstrating the association of the steps with the quantization of the CDW wave vector.96 In o-TaS3, the asymmetrical conductivity had no dependency below a certain temperature (2 K), which was caused by soliton transport in the CDW system.97 Nichols et al. studied the frequency and voltage dependencies of voltage-induced torsional strain in o-TaS3 and concluded that the strain is allied with a divergence in the CDW instead of the CDW current.98 A change in the length, L, (depending on the electric field and time) of TaS3 samples demonstrated that hysteresis partly corresponds with resistance.99 Thermal expansion studies of o-TaS3 crystals showed the uncharacteristic behaviour of the hysteresis loop of length L below the Peierls transition temperature and at 100 K, the elastic modulus meets the Young's modulus of the CDW wave.100 Gorlova and co-workers observed electric-field-induced torsional strain corresponding to colossal shear in TaS3 whiskers.101 The threshold and hysteresis behaviour of torsion demonstrated its link with CDW deformation. Correspondingly, Nichols et al. investigated the effects of hysteretic voltage-induced torsional strain on CDW depinning in o-TaS3 using square-wave and triangular-wave voltages of dissimilar frequencies and amplitudes.102 Inagaki et al. reported the magnetoresistance of a CDW in o-TaS3 whiskers under a magnetic field up to 4.2 K. When the field was aligned with the a-axis, the maximum amplitude of angle-dependent magnetoresistance was achieved, whereas there was zero value at the b-axis.103 Electrical transport measurements of single o-TaS3 by Farley et al. discovered the depression of the Peierls transition temperature to 205 K. Below this temperature, there was depinning of the CDW, which was accredited to broadening of the electric field and surface confinement; low dimensionality and a limited size effect caused a great enhancement in the threshold voltage for the nucleation of CDW dislocations (Fig. 6).104 Wu et al. studied the pressure-dependent vibrational properties, which imply that the one and only S‖ mode (at 54 cm−1) was held accountable for the crystalline orientation of TaS3 through angle-resolved Raman spectroscopy and high-pressure diamond anvil cell studies.105 Frequency-dependent conductivity measurements showed the Drude type behaviour of the inertial feedback of TaS3, associated with damping.106
TaSe3 is reported to be superconducting below 2.1 K with anisotropic properties under the application of a magnetic field79 and Kikkawa et al. also indicated the superconductivity of TaSe3 at T = 1.9 K.65 Thickness-dependent work function variations of TaSe3 flakes confirmed the extended screening length (24 nm) at which the work function suddenly decreased, which specified its hydrophilic nature, and the Raman modes at 128 cm−1; 141, 164, 217, and 237 cm−1; 177 cm−1; and 186 cm−1 corresponded to Bg, B2, Ag, A1g modes, respectively.107 TaSe3 is reported to have a high current-carrying capacity and high current density, and it has emerged as a possible potential candidate to act as an interconnector in electronic devices due to capping with hexagonal boron nitride (h-BN).108 Experimental observations demonstrated that quasi-1D TaSe3 nanowires transmit lower levels of normalized noise spectral density, with potential for rationalized local interdependent applications.109–111 The main properties of MX3 are summed up in Fig. 7.
Fig. 7 A schematic diagram showing the properties of quasi-1D MX3. Resistivity figure reprinted with permission from ref. 78, copyright: 2019, American Chemical Society; in-plane anisotropy figure republished with permission from ref. 9, permission conveyed through Copyright Clearance Center, Inc. (2016); superconductivity figure reproduced with permission from ref. 124, copyright: 2015, Royal Society of Chemistry; optical properties figure reproduced with permission from ref. 6, copyright: 2019, IOP Science; carrier mobility figure reprinted with permission from ref. 78, copyright: 2019, American Chemical Society; CDW properties figure republished with permission from ref. 104, permission conveyed through Copyright Clearance Center, Inc.; band gap figure reproduced with permission from ref. 6, copyright: 2019, IOP Science; magnetic properties figure reproduced with permission from ref. 9, copyright: 2015, American Chemical Society. |
Material | Starting precursors | Growth conditions | Heating temperature; substrate | Size | Ref. no. |
---|---|---|---|---|---|
TiS3 crystals | Ti sheets, S powder | Vacuum-sealed (10−6 torr) ampoule, 5 days | 520 °C; wall of quartz tube | Bulk crystals | 24 |
TiS3 thin film | TiCl4, t-butyl disulfide | Vacuum heating (10−1 torr) | 260 °C; glass, and Ti and Al foil | Large-area thin film | 127 |
TiS3 nanowhiskers | Ti metal foil, S powder | Vacuum-sealed (10−5 torr) ampoule, 3 days | 500 °C; Ti foil and wall of quartz ampoule | l = 100 μm, w = few μm, t = 0.5 μm | 28 and 124 |
TiS3 nanoribbons | Ti disc, S powder | Vacuum-sealed (10−3 torr) ampoule, 20 h | 520 °C; wall of quartz tube | l = 100 μm, w = 1–5 μm, t = 5–200 nm | 144 |
ZrS3 flakes | Zr metal sheets, S pellet | Vacuum-sealed (10−6 torr) ampoule, 5 days | 650 °C; wall of quartz tube | Bulk needles and mechanically exfoliated flakes | 25 and 144 |
ZrSe3 crystals and nanobelts | Zr powder, Se powder | Vacuum-sealed (10−2 torr) ampoule, 24 h | 650 °C; wall of quartz tube | l = tens of μm, w = 20–2600 nm | 129 |
ZrTe3 crystals | Zr powder, Te powder | Vacuum-sealed ampoule | 950 °C (source)/850 °C (sink) and 735 °C (source)/660 °C (sink); wall of quartz tube | l = 5 mm, w = 0.8 mm, t = 0.2 mm | 15 |
HfS3 crystals and flakes | Zr metal sheets, S pellet | Vacuum-sealed (10−6 torr) ampoule, 5 days | 650 °C; wall of quartz tube | Bulk needles and mechanically exfoliated flakes | 24 |
HfSe3 crystals and flakes | Hf powder, Se powder | Vacuum-sealed (10−2 torr) ampoule, 24 h | 650 °C; wall of quartz tube | l = tens of μm, w = 20–2600 nm | 129 |
HfTe3 crystals and flakes | Hf powder, Te powder | Vacuum-sealed, CVT-iodine carrier | 500–540 °C; wall of quartz tube | l = 0.3 mm, w = 0.3 mm, t = 0.1 mm | 51 |
NbS3 whiskers and flakes | Nb powder, S powder | Vacuum sealed (10−5 torr) ampoule, 7 days | 550 °C (source)/470 °C (sink); wall of quartz tube, mechanical exfoliation | Whisker-like crystals | 84 |
NbSe3 crystals | Nb powder, Se powder | Reaction under vacuum, 15 days | 700 °C; wall of quartz tube | l = 7.0 mm, w = 0.05 mm, t = 0.01 mm | 87 |
NbSe3 nanoribbons | Nb powder, Se powder | Reaction under vacuum, >15 days | 700 °C; wall of quartz tube | l = >10 μm to few mm, w = t = 20–700 mm | 72 |
TaS3 crystals and flakes | Ta powder, S powder | Vacuum sealed, CVT-iodine carrier, 48 h | 760 °C; wall of quartz tube | l = several cm, w = 40–900 nm, t = 20–50 mm | 132 |
TaSe3 crystals and flakes | Ta powder, Se powder | Vacuum sealed, CVT-iodine carrier, 10 days | 700–600 °C; wall of quartz tube | l = >10 mm, w = t = ∼200 μm | 107 and 108 |
TaSe3 nanowires | TaCl5, Se powder | CVD growth under vacuum, Ar + H2 gas, 15 min | 400 °C; wall of quartz tube and substrate | l = >120 nm | 78 |
Fig. 8 The synthesis of MX3. Optical photographs of the ampoule (a) before and (b) after the growth of TiS3 whiskers on Ti foil and the surface of quartz. (c and d) SEM images of the grown TiS3 whiskers with arrow marks indicating samples collected from the respective positions. Reproduced with permission from ref. 124, copyright: 2015, Royal Society of Chemistry. (e) The synthesis of ZrTe3 via a chemical vapour deposition approach. (f) FESEM and TEM images of ZrTe3. Republished with permission from ref. 130, permission conveyed through Copyright Clearance Center, Inc. |
Fig. 9 (a) The crystallographic structure of TaSe3. (b) A schematic diagram of the CVD set-up for the growth of TaSe3 nanowires. (c) An optical image and (d and e) SEM images of a population of TaSe3 nanowires. Reprinted with permission from ref. 78, copyright: 2019, American Chemical Society. (f) A schematic illustration of the mechanical exfoliation of TaSe3 flakes from bulk TaSe3. (g) A 3D representation of a quasi-1D TaSe3 nanoribbon monolayer on a SiO2/Si substrate. Reproduced with permission from ref. 107, copyright: 2019, Multidisciplinary Digital Publishing Institute. (h) Source-drain current vs. voltage for a 11.6 nm TaSe3 nanowire in a 2-electrode configuration, with the inset showing an SEM image of the device. (i) Resistivity as a function of bundle width of CVD-grown TaSe3 nanowires and a comparison with bulk Cu and exfoliated TaSe3. (j) The current density response of a 7 × 7 nm2 nanowire as a function of Vds with failure at a current density of 108 A cm−2. Reprinted with permission from ref. 78, copyright: 2019, American Chemical Society. |
Mechanical exfoliation is a widely used method to isolate single- to few-layered MX3 nanosheets from bulk material.6 On the other hand, liquid phase exfoliation is an effective method for the large-scale production of MX3 nanosheets.139 Few-layer TiS3 material was prepared by Island et al. at a temperature of 400 °C with sheet-like morphology, and this was later mechanically exfoliated to few-layer form.6 Xie et al. reported the liquid phase exfoliation of bulk ZrS3 in n-propylamine for 30 min followed by heating at 120 °C for 3 days in a Teflon-lined autoclave.139 The obtained product was washed and ultrasonicated in 1-cyclohexyl-2-pyrrolidinone to achieve ZrS3 nanosheets. NbS3 and NbSe3 nanoparticle colloidal solutions were prepared via a top-down approach through the ultrasonication of powder in different solutions (DMF, acetone, acetonitrile, ethanol, a water–ethanol mixture, etc.) for 3 h. Li-ion intercalation is proposed as a suitable method to prepare MX3 nanoribbons, in which three lithium atoms are incorporated into an MX3 unit to yield Li3MX3.140 A mechanical exfoliation approach is employed to prepare few-layered flakes from TaSe3 crystals (Fig. 9f and g).107 A schematic diagram of the different steps used in this process is shown in Fig. 9, in which wafer dicing tape is used and stuck onto the crystal, which is then removed and adhered to the SiO2/Si substrate.
Fig. 10 A schematic diagram showing the various applications of MX3. Solar and fuel cell figure republished with permission from ref. 33, permission conveyed through Copyright Clearance Center, Inc.; lithium-ion batteries figure reprinted with permission from ref. 134, copyright: 2015, American Chemical Society; sensors figure reprinted with permission from ref. 133, copyright: 2018, American Chemical Society; photodetectors figure republished with permission from ref. 129, permission conveyed through Copyright Clearance Center, Inc.; FETs figure reproduced with permission from ref. 124, copyright: 2015, Royal Society of Chemistry; thermoelectricity figure republished with permission from ref. 157, permission conveyed through Copyright Clearance Center, Inc.; central figure reproduced with permission from ref. 6, copyright: 2015, Springer Nature. |
Fig. 11 Few-layered TiS3 FETs. (a) A schematic diagram and (b) an SEM image of a typical FET. (c) Conductivity vs. gate voltage dependencies of four different fabricated FETs, with the inset showing drain source current vs. drain source voltage at different gate voltages. Reproduced with permission from ref. 124, copyright: 2015, Royal Society of Chemistry. |
Fig. 12 The hydrogen photogeneration properties of MX3 nanostructures. (a) An SEM image and (b) the conduction and valence band energy levels on potential (V vs. NHE) and energy (eV vs. vacuum) scales, with the redox potentials for the water-splitting half reactions at pH = 9.0 vs. NHE, of TiS3 nanoribbons. The hydrogen evolution flow of TiS3 nanoribbons (c) at 0.0 V and (d) at different bias potentials. Republished with permission from ref. 144, permission conveyed through Copyright Clearance Center, Inc. |
The use of 2D materials and their heterojunctions in optoelectronics and solar-correlated devices is utterly controlled by the superiority of the heterojunction formed and the band alignment to tune carriers at interfaces.33 Zhao et al. employed DFT to calculate the band superstructures and heterostructures of IV–VIA monolayers of MX3 (M = Zr, Hf; X= S, Se) and VIIB–VIA monolayer MX2 (M = Tc, Re; X = S, Se). The calculations indicated that for MX3, the valence bands are dependent on the p-states of chalcogens, whereas the d-states of the transition metals control the conduction bands. For MX2 monolayers, both the valence and conduction bands depend on the d-states of the transition metals. Considering standard water redox potentials (−4.44 eV and −5.67 eV for reduction (H+/H2) and oxidation (O2/H2O), respectively), the combination of MX3 and MX2 monolayers and their band alignment to create efficient heterostructures have been reported.33 Fig. 1333 shows the band-offset components and a contour map of power conversion efficiency (PCE) values. From calculations, it is predicted that a ZrS3/HfS3 bilayer thin-film device could achieve 16–18% efficiency, which is much higher than other reported 2D heterojunction solar cell devices. Similarly, Ahammed et al. reported that the PCEs in ZrS3/MoS2, ZrS3/WS2, ZrS3/MoSeTe, ZrS3/WSTe, and ZrS3/WSeTe heterostructured bilayers are as high as ∼12%, 8%, 16%, 14%, and 14%, respectively.146 Recently, anisotropic ZrS3 has been reported to act as an active material for perovskite light-emitting diodes and P–N junction diodes, and the results are found to be promising, which opens up pathways for further research into the area of MX3 heterojunctions for optoelectronics and solar-cell devices.147,148
Fig. 13 The application of MX3 heterostructures to solar cells. (a) Band offsets of ZrS3/HfS3 and TcS2/ReS2 with the vacuum level as zero reference. (b) PCE contours obtained as a function of the donor band gap and conduction band offset. (c) A schematic illustration of thin-film solar cells with the associated mechanism. Republished with permission from ref. 33, permission conveyed through Copyright Clearance Center, Inc. |
Fig. 14 (a) A schematic illustration of a ZrSe3 and HfSe3 single nanobelt photodetector. SEM images of (b) ZrSe3 and (c) the HfSe3 photodetector. I–V characteristics of (d) ZrSe3 and (e) the HfSe3 nanobelt photodetector. Republished with permission from ref. 129, permission conveyed through Copyright Clearance Center, Inc. |
Fig. 15 (a) The crystal structure of NbS3. (b) Optical microscopy and (c) SEM images of a NbS3 crystal. Schematic diagrams of (d) a NbS3 photodetector and (e) bending conditions. (f) ON–OFF photovoltage curves at room temperature and (g) resistance, response time, and photovoltage data under different bending conditions. Reprinted with permission from ref. 151, copyright: 2020, American Chemical Society. |
Fig. 16 (a) Schematic representations of a FET, NO gas sensor, and photodetector based on TaS3 nanofibers. (b) An SEM image of the TaS3 nanofibers. (c) The impedance phase responses of gas sensors fabricated on different substrates (polyester, parafilm, paper) as functions of NO concentrations. (d) Selectivity studies of the sensors fabricated on different substrates. (e) An optical image of a FET based on TaS3 fibers. (f) FET transistor characteristic curves at different gate voltages (pink: −1 V, blue: 0 V, green: 1 V, red: 2 V, black: 3 V). (g) Dark current (black) and responsivity (red) curves. Reprinted with permission from ref. 133, copyright: 2018, American Chemical Society. |
Fig. 17 SEM images of (a) NbSe3 and (b) NbSe3@rGO. (c) A TEM image of NbSe3@rGO. The electrochemical performance of NbSe3 and NbSe3@rGO in the voltage range of 0.005–3 V vs. Li/Li+. Discharge and charge curves of (d) NbSe3 and (e) nanobelts. (f) The cycling performance of NbSe3@rGO and NbSe3 nanobelts at a current density of 100 mA g−1. (g) The rate performances of NbSe3@rGO and NbSe3 nanobelts. Reprinted from ref. 155, copyright: 2017, with permission from Elsevier. |
Fig. 18 (a) An SEM image of TaS3 nanowires. (b) Voltage profiles of a lithium-ion battery based on TaS3 nanowires cycled between 0.001 and 3 V vs. Li+/Li at a cycling rate of 0.1C. (c) Capacity and coulombic efficiency as a function of cycle number for a TaS3 nanowire electrode at a cycling rate of 0.1C. (d) An SEM image of nanowires after long-term cycling tests. (e) Rate performance and coulombic efficiency as a function of cycle number for TaS3 electrodes and as a function of discharge rate (0.1–10C). (f–h) A schematic diagram of a fabricated flexible lithium-ion battery with photographs showing a glowing LED powered by flat and bent states. Reprinted with permission from ref. 134, copyright: 2015, American Chemical Society. |
Weak van der Waals forces of attraction between layers and anisotropy along the chain axis (b-axis) provide MX3 materials with the advantages of both structural (2D layered material and quasi-1D) and electrical (CDW phenomena) properties. In MX3, the chalcogen atoms (S, Se, Te) are considered to be electron reservoirs, thus providing exceptional electronic properties, arising from bulk and nanostructured materials. The M atom presents at the centres of all prismatic chains of MX6, which are linked to each other in an infinite chain manner and run parallel to the b-axis. At the base of the prism of MX3, one bond is shorter compared to the others, so more research needs to be carried out in this direction utilizing the advantages of this property, so that new applications of these materials can be explored. Even TiS3 shows strong dichroism in agreement with systematic anisotropy. Similarly, the creation of different types of vacancies in the TiS3 system leads to the induction of a definite magnetic moment. Spin–orbit interactions in the case of ZrX3 materials result in an increase in the energy split from S to Se atoms. Certain phonon modes undergo a dramatic linewidth reduction near TCDW, representing the strong coupling of phonons with electronic degrees of freedom associated with the CDW for ZrTe3. However, huge attention and focus should be given to the interesting property of MX3, i.e., the CDW, so that the shift in the CDW transition can be well-understood and numerous unexplored applications can be instigated in this direction. High-end techniques and better qualitative practical methods, like solid-state NMR, photon correlation spectroscopy, X-ray diffraction topography (XRT), in situ and ex situ characterization (XRD, SEM, and TEM), muon-spectroscopy, terahertz spectroscopy, etc., can be engaged to study and visualize the LRO pyramid in few-layer or monolayer MX3. The properties are quite poor along the a-axis, and they can be tuned via the application of external pressure, strain, etc. More and more focus should be given to studying the anisotropy in the a-axis direction, which is somewhat present in some metal tri-tellurides and will be helpful for outstanding applications in the field of nanotechnology. Therefore, these properties open up the endless possibility for spectacular applications. In contrast, 2D layered material characteristics are somehow not properly shown by the materials due to the quasi-1D nature.
Synthesis approaches for MX3 materials are quite cost-effective and environmentally friendly and less time-consuming in comparison to other 2D material families, like MXenes. These methods include direct chemical reactions, chemical vapour transport (CVT), chemical vapour deposition (CVD), mechanical and chemical exfoliation, etc. Subsequently, other atoms can be intercalated into the layers of MX3 in line with the feeble van der Waals forces. Characterization techniques such as X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), field-emission scanning electron microscopy (FESEM), electron-dispersive X-ray analysis (EDAX), atomic force microscopy (AFM), scanning tunnelling microscopy (STM), transmission electron microscopy (TEM), Raman spectroscopy, UV-visible spectroscopy, and optical microscopy are typical methods used to appraise the properties of MX3 materials. Mostly, CVD and CVT methods are used for producing bulk as well as nanostructured materials with layers of few mm to nm thickness. As monolayer materials show interesting and research-driven properties, innovative and scalable exfoliation techniques can be used to form thin monolayers.
Most materials are fabricated via taking both precursors at a particular stoichiometry, placing them into a vacuum-sealed ampule, and heating them at a certain temperature for a precise amount of time. Later, after the system cools down, the prepared samples are removed. Via these methods, nanowhiskers, nanoribbons, nanosheets, etc. of MX3 (TiS3, ZrX3, HfX3, NbX3) are prepared, and their characterization is carried out via the above-stated techniques. However, these methods are exceedingly time consuming, and some experiments take 15 days. Therefore, more effective strategies should be implemented for the growth of bulk as well as monolayered MX3. Upgraded CVD and CVT systems with high-end programming for the skillful flow of inert gases (Ar, H2, N2, or a mixture of any two) and proper temperature maintaining tools should be prearranged to control the number of layers and their thickness.
MX3 materials show a wide range of applications, e.g., FETs, solar and fuel cells, lithium-ion batteries, photodetectors, photosensors, and thermoelectrics. The higher mobilities and greater ON/OFF ratios of MX3-based FETs are worth noting in the present day. MX3 materials possess the advantages of having a low band gap to absorb direct solar energy, and being low-cost, plentiful, and appropriate for photoelectrochemical water splitting, making them suitable candidates for use in solar and fuel cells. The maximum efficiency reached is about 16–20%. The high photoresponse, fast switching time, and high cut-off frequency make MX3 the best material for use in sensors and photodetectors. MX3 can also be used in converting heat into electrical power due to its transport properties. Despite all the merits possessed by MX3, these materials are well behind in terms of real-world applications, like in supercapacitors, sodium-ion batteries, etc. These materials can be fused with carbon-based materials (single-walled and multiwalled carbon nanotubes and rGO) for supercapacitor applications due to their high surface areas. The efficiencies of solar cells can be improved via the incorporation and doping of active elements, which can allow more focus on the real-world applications of MX3. Work should be done to increase the capacities of lithium-ion batteries based on MX3 via the strong intercalation of other atoms into the layers of MX3. The photoresponse can be enhanced along with the switching time in photodetectors. MX3 materials can be used for building biosensors, as they are non-toxic and eco-friendly in nature.
This review article highlights the significance of the in-plane anisotropy and quasi-1D nature of MX3 materials, with a complete study of the atomic structures, physical and chemical properties, intrinsic modulations in CDW states, cost- and time-effective synthesis methodologies, and related plausible applications. We contemplate optimistically that this fragment of work on quasi-1D MX3 materials, the evolution of their primary properties, and their related applications will have contemporary significance, leading to novel and fresh research.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0ra07160a |
This journal is © The Royal Society of Chemistry 2020 |