† Corresponding author. E-mail:
Project supported by Guangdong Innovative and Entrepreneurial Research Team Program (Grant No. 2016ZT06D348) and Shenzhen Peacock Program (Grant No. KQTD2016022619565991).
Single-layered zirconium pentatelluride (ZrTe5) has been predicted to be a large-gap two-dimensional (2D) topological insulator, which has attracted particular attention in topological phase transitions and potential device applications. Herein, we investigated the transport properties in ZrTe5 films as a function of thickness, ranging from a few nm to several hundred nm. We determined that the temperature of the resistivity anomaly peak (Tp) tends to increase as the thickness decreases. Moreover, at a critical thickness of ∼40 nm, the dominating carriers in the films change from n-type to p-type. A comprehensive investigation of Shubnikov–de Hass (SdH) oscillations and Hall resistance at variable temperatures revealed a multi-carrier transport tendency in the thin films. We determined the carrier densities and mobilities of two majority carriers using the simplified two-carrier model. The electron carriers can be attributed to the Dirac band with a non-trivial Berry phase π, while the hole carriers may originate from surface chemical reaction or unintentional doping during the microfabrication process. It is necessary to encapsulate the ZrTe5 film in an inert or vacuum environment to potentially achieve a substantial improvement in device quality.
Topological insulators provide a unique platform for exploring novel quantum phases and phenomena, such as the quantum spin Hall (QSH) effect,[1–3] and the quantum anomalous Hall effect,[4,5] etc. ZrTe5 is an orthorhombic layered material, which is predicted to be a 2D topological insulator in a monolayer form.[6] The ZrTe5 bulk system also displays abundant electronic and exotic transport properties, and the topological nature is still unclear, i.e., strong topological insulator,[7] weak topological insulator,[8–10] and Dirac semimetal.[11–13] However, there are significantly fewer experimental studies on ZrTe5 thin films. Moreover, the conversion between three-dimensional (3D) and 2D systems is not fully understood. Since interlayer coupling in ZrTe5 is as weak as graphite, this allows us to reduce the sample’s thickness to a monolayer by mechanical exfoliation.[14] More recently, several groups have discovered that the transport properties of ZrTe5, such as the Hall reversal,[15,16] and resistivity-anomaly peak temperature (Tp),[17,18] can be changed by modifying the thickness and temperature. Strong anisotropic transport behavior is also observed in angular geometry devices.[19] The latest angle-resolved photoemission spectroscopy (ARPES) and scanning-tunneling microscopy (STM) results provide strong evidence of the topological protected metallic state at the step edge,[9] which could host the quantum spin Hall state. These phenomena require further transport investigation of the interplay between electrical field, magnetic field, temperature, and thickness, with the objective of gaining a deeper understanding of the still unknown topological nature and resistivity anomaly of thin ZrTe5 films.
In this work, we have systematically performed transport measurements in ZrTe5 thin films, with tuning thickness, temperature, electrical field, and magnetic field. The shifting of Tp and the Hall resistance (Rxy) anomaly, which is accompanied by the SdH oscillations, were observed when the thickness was reduced, demonstrating the multi-carrier behavior in the thin films. Using the simplified two-carrier model, we determined the carrier density and mobility of each band. One is a Dirac-like electron band with a high mobility of ∼ 2 × 104 cm2 · V−1 · s−1 and a low carrier density of ∼ 1016 cm−3–1017 cm−3, which hosts the nontrivial Berry phase clarified by the Landau-Fan diagram analysis of the SdH oscillations.[16] The other is a trivial hole-like band with a low mobility of 102 cm2 · V−1 · s−1 ∼ 103 cm2 · V−1 · s−1 and a high carrier density of 1018 cm−3 ∼ 1019 cm−3. Considering the decay of the sample under ambient conditions, this hole band may originate from hole doping by surface chemical reactions. Similar results have also been discussed for several other semimetals, such as Cd3As2,[20] WTe2,[21] HfTe5,[22] TaAs2,[23] and graphite.[24] These observations provide a clue to understanding the connection between 2D and 3D ZrTe5, and to potentially achieve a substantial improvement in device quality.
A single crystal of ZrTe5 was grown by chemical vapor transport (CVT).[25] The ZrTe5 thin films were prepared by standard mechanical exfoliation of the bulk crystal, then deposited onto silicon substrates with 285-nm SiO2. The samples were first identified using optical microscopy, then immediately coated with a PMMA film in a glove box filled with argon gas. The exact thickness of the sample was finally measured using Atomic Force Microscopy (model Keysight 5500) after transport measurement. The devices were fabricated by traditional e-beam lithography in a Hall bar geometry. For the thick films, a silicone elastomer polydimethylsiloxane stamp PDMS was used as a media to transfer the film onto pre-fabricated electrode patterns.[26] Magneto-transport measurements were performed using an Oxford TeslatronPT cryostat with variable temperatures from 1.5 K to 300 K and a magnetic field up to 14 T. The standard lock-in method with a low frequency (17.777 Hz) was used to measure the longitudinal and transverse resistivity. A typical measurement current was 10 nA ∼ 1 μA depending on the resistance of the sample. The back gate experiment was performed using a Keithley 2400 source meter.
Figure
To better understand the thickness-dependent transport property, the Hall resistance of samples with different thicknesses was measured at 1.5 K. As shown in Fig.
This significant deviation forces us to consider the effect of decay during the sample preparation process. The decay takes place at the surface, where the activated atoms emerge after being exfoliated from the bulk crystal. These activated atoms can easily react with oxygen, water, and organic solvents during fabrication, leading to giant hole doping at the surface.[29,30] For thick samples (> 100 nm), this effect of surface doping is limited due to the small surface-to-volume ratio. The transport is dominated by electron carriers in the bulk ZrTe5, giving rise to the negative slopes of the Hall resistance. On the contrary, this hole doping at the surface becomes strong enough in thin films to cause the movement of the Fermi level. For samples with a thickness less than 33 nm (see Fig.
Figures
To further understand the SdH oscillations, the Landau plotting and effective cyclotron mass fitting were performed. According to the Lifshitz–Kosevich formula,[16,32]
Figure
We find that the nonlinear property in the Hall resistance is coincident with the multi-carrier transport behavior. For simplification, the carriers are divided into two parts: one is the Dirac band, and the other is regarded as an equivalent hole band with the same scattering time. Thus the standard two-carrier model is applied as,[20,24]
To reveal two-carrier transport in ZrTe5 thin films, we use Kohler’s plots[33,34] with ΔRxx(B)/Rxx(0) ∼ (B/Rxx(0))2 for the classical B2 dependence of MR, as shown in Fig.
The gate-dependent longitudinal conductivity of thin films without an external magnetic field was also investigated. For thicker samples (> 30 nm), gate tuned doping is limited because of the Thomas-Fermi dielectric screening.[35,36] The gate voltage can only modulate about 5% ∼ 20% of the original resistance without doping. For thin samples, the gate effect becomes larger, and most of the samples show p-type curves instead of an ambipolar behavior. In thin films with less decay, it is possible to observe the ambipolar behavior. Figure
The field-effect mobility μFE in the linear region of the transfer characteristics can be determined using the following formula,[39]
Figure
Figure
Finally, we study the gate effect of ZrTe5 devices at different magnetic field strengths.[19] As shown in Fig.
In summary, we have systematically performed low-temperature magnetotransport measurements to investigate thickness-dependence, temperature-tuning, and gate-modulation in ZrTe5 films. We determined that Tp increases as the thickness of the ZrTe5 is reduced. However, the Tp values are irregular and exhibit a large sample dependence, especially below 40 nm. Combining the SdH oscillations and Hall measurements, we determined the carrier densities and mobilities using two-carrier model fitting: one electron band hosts a high mobility of ∼2 × 104 cm2 · V−1 · s−1 and a low carrier density of ∼ 1016 cm−3–1017 cm−3; the other hole band has a low mobility of 102 cm2 · V−1 · s−1 ∼ 103 cm2 · V−1 · s−1 and a high carrier density of 1018 cm−3 ∼ 1019 cm−3. Considering the effect of decay on the sample surface, we propose that the multi-carrier transport property is induced by the coexistence of the Dirac-like electron band with a nontrivial Berry’s phase, and the trivial hole-like band from unintended surface chemical reactions. We anticipate that these results will be helpful in attempting to understand the conversion between a bulk system to a thin film. A stricter fabrication process is required using a uniform insulating substrate (i.e., hexagonal boron-nitride) as the capping layer, which may facilitate a reliable way to study the QSH effect and topological phase transitions in ZrTe5 films.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] | |
[35] | |
[36] | |
[37] | |
[38] | |
[39] |