Facile and fast growth of high mobility nanoribbons of ZrTe5
Wang Jingyue1, Niu Jingjing1, Li Xinqi1, Ma Xiumei1, Yao Yuan2, Wu Xiaosong1, 3, 4, †
State Key Laboratory for Artificial Microstructure and Mesoscopic Physics, Peking University, Beijing 100871, China
Institute of Physics, Chinese Academy of Sciences, Beijing 100190, China
Frontiers Science Center for Nano-optoelectronics and Collaborative Innovation Center of Quantum Matter, Peking University, Beijing 100871, China
Department of Physics, South University of Science and Technology of China, Shenzhen 518055, China

 

† Corresponding author. E-mail: xswu@pku.edu.cn

Project supported by the National Key Research and Development Program of China (Grant Nos. 2016YFA0300600, 2016YFA0300802, 2013CB932904, and 2016YFA0202500) and the National Natural Science Foundation of China (Grant Nos. 11574005, 11774009, and 11234001).

Abstract

Recently, ZrTe5 has received a lot of attention as it exhibits various topological phases, such as weak and strong topological insulators, a Dirac semimetal, a three-dimensional quantum Hall state, and a quantum spin Hall insulator in the monolayer limit. While most of studies have been focused on the three-dimensional bulk material, it is highly desired to obtain nanostructured materials due to their advantages in device applications. We report the synthesis and characterizations of ZrTe5 nanoribbons. Via a silicon-assisted chemical vapor transport method, long nanoribbons with thickness as thin as 20 nm can be grown. The growth rate is over an order of magnitude faster than the previous method for the bulk crystals. Moreover, transport studies show that the nanoribbons are of low unintentional doping and high carrier mobility, over 30000 cm2/V⋅s, which enable reliable determination of the Berry phase of π in the ac plane from quantum oscillations. Our method holds great potential in growth of high quality ultra-thin nanostructures of ZrTe5.

1. Introduction

The discovery of topological insulators (TI) has brought a new insight into the classification of solid state materials and attracted enormous interest in the past few years.[1,2] Soon, the topological concept has been extended to superconductors and metals, i.e., three-dimensional (3D) topological Dirac semimetals and Weyl semimetals.[110] Among many topological materials, ZrTe5 is unique in that it exhibits various topological phases, such as weak and strong topological insulators, a Dirac semimetal, a 3D quantum Hall state, and a quantum spin Hall insulator in the monolayer limit.[1122] These phases sensitively depend on the lattice constants.[23] The change of the band structure with temperature, thickness, and pressure has been experimentally observed.[2428]

Most of studies on ZrTe5 have been carried out for 3D bulk.[15,16,18,29,30] On the other hand, nanowires/nanoribbons of topological materials can be very useful, as they not only enhance the contribution from the surface states due to a large surface-to-volume ratio, but also give rise to new properties.[3133] For instance, studies on nanostructured topological insulators have revealed interesting phases and spin related transport.[3436] The marriage between InSb nanowires and superconductors has given birth to the experimental realization of Majarona zero modes.[37] It is therefore highly desired to develop methods for growth of high quality nanostructured ZrTe5 so that new properties may be further introduced. However, such a study has so far not been reported.

Conventionally, ZrTe5 is synthesized by a chemical vapor transport method. The growth is time-consuming, in the order of weeks, which hinders efforts on improvement of the crystal quality.[15,17,38,39] In this work, we employ a silicon-assisted chemical vapor transport method to grow both ZrTe5 nanoribbons and bulk crystals. The growth time of millimeter size single crystals is reduced to less than 90 min, representing substantial improvement of the growth rate. Quantum transport measurements show that the as-grown nanoribbons are of high mobility and low unintentional doping. Well-resolved quantum oscillations confirm a topological band with a Berry phase π in the ac plane. Our method holds great potential in growth of ultra-thin nanostructures and bulk crystals of ZrTe5.

2. Growth method

Growth was carried out in a horizontal three-zone tube furnace. The source materials are zirconium and tellurium elements. Since zirconium powder is difficult to handle as it is easily oxidized and flammable, shots of 0.6 g each on average were used instead. Considering much less surface area of shots than powder, the amount of zirconium was significantly more than the stoichiometric ratio, e.g., 5 g Zr and 0.3 g Te. Iodine of 2 mg/cm3 was employed as the transport agent. As illustrated in Fig. 1(a), the source materials, iodine, and several growth substrates were sealed in a quartz ampoule. The size of each silicon substrate is 1 cm× 1 cm× 0.5 mm. The total length of the ampoule is around 60 cm. The distance between zirconium shots and the nearest substrate is around 50 cm, the distance from tellurium powder to the nearest substrate is around 15 cm. Before sealing, the ampoule was flushed by argon and pumped to 5 Pa before the growth process. The effect of residual gas on the transport of chemical species during reaction is negligible as the partial pressure is orders of magnitude lower than that of iodine (and its gaseous compound) and tellurium. In particular, zirconium shots and iodine particles were placed in one side of the ampoule, while tellurium powder and silicon substrates were placed in the other side. The iodine vapor transport growth of ZrTe5 can be qualitatively described by the following reactions:

Due to the low vapor pressure of Zr, the first reaction was set in the highest temperature zone, so sufficient ZrI4 could be fed to the second reaction, which was held in the lowest temperature zone. The tellurium powder was placed in the third temperature zone in between the other two zones. The optimal temperatures of the three zones were found to be 540 °C, 480 °C, and 450 °C, respectively. The growth usually took place for 150 min to grow nanostructures (or 90 min for bulk at the coldest part of the quartz ampoule), and then the ampoule was allowed to cool naturally.

Fig. 1. Growth of ZrTe5 nanostructures. (a) A schematic drawing of the growth setup for the iodine vapor transport method. A two-bulb ampoule is used. Zirconium shots and iodine particles are placed in one bulb, which is held at a higher temperature, 540 °C. Tellurium powder and silicon substrates are placed in the other bulb, which is held at a lower temperature, 480 °C and 450 °C, respectively. (b)–(e) A series of SEM images of grown nanostructures with increasing magnifications. ZrTe5 nanoribbons are grown on a mattress of ZrTe3 and Te crystals. A recession of the silicon substrate due to iodine etching can be observed in (d) and (e). (f) Energy-dispersive x-ray spectroscopy of nanoribbon, indicating ZrTe5. (g) A spectrum taken at the root of the nanoribbon, indicating ZrTe3. The black squares in (e) mark the spots where the spectra were taken.
3. Results
3.1. Morphology of ZrTe5 nanoribbons

Nanoribbons were found on the silicon substrate and certain segments of the ampoule wall, as shown by the SEM image in Fig. 1. These nanostructures can be very long, e.g., 200 μm. Later we will show via scanning transmission electron microscopy (STEM) and Raman spectroscopy that these long nanostructures are ZrTe5. A substantial advantage of our method is the fast growth rate. Depending on the growth conditions, such as the ratio and the temperature of the source materials, macroscopic crystals can also been grown in similar growth time on the ampoule wall. The size of the crystals can be as large as 10 mm by 0.2 mm by 0.05 mm, rivalling reported results with several weeks of growth time.[15,38,39] A rough estimation yields a growth rate being at least 40 times faster.

Interestingly, these ZrTe5 nanostructures did not directly grow on the silicon substrate, but on a mattress of materials, as seen in Fig. 1(b). Energy-dispersive x-ray spectroscopy (EDX) measurements show that the mattress mainly consists of Zr and Te with a ratio of ∼ 1 : 3, indicating ZrTe3 as seen in Fig. 1(g). Some tellurium crystals were also found. Apparently, ZrTe3 and Te crystals grow first. They may provide a suitable substrate for subsequent growth of ZrTe5. A close look at the edge of the mattress, as shown in Figs. 1(d) and 1(e), reveals recess etching into the silicon substrate. Iodine is known to react with silicon and forms SiI4, which is in a gaseous phase at our growth temperature. We have found that silicon plays an important role in the reaction. When the silicon substrate was replaced with mica, neither ZrTe3 nor ZrTe5 was formed under the same growth condition. On the other hand, in presence of both silicon and mica substrates, growth could occur on the mica substrate, though much less effectively as shown in Figs. 2(e) and 2(f). How silicon affects the growth processes is not clear currently and deserves further study. In the following, we focus on the quality of the grown nanoribbons.

Fig. 2. (a)–(d) Aberration corrected TEM images of ZrTe5 nanoribbon. The nanoribbon is about 50 nm thick. (a) HAADF image. Inset, atomic structure model of ZrTe5 in the (110) plane. (c) HAADF image in the (010) plane. (b), (d) Electron diffraction pattern along the [110] and [010] directions. (e) Optical micrograph of a mica substrate after growth in presence of a silicon substrate. ZrTe5 ribbons can be found. (f) Optical micrograph of a mica substrate after growth without a silicon substrate. No ZrTe5 can be found.
3.2. High quality of ZrTe5 nanoribbons

EDX was carried out for the nanoribbons, which indicates that they consist of Zr and Te. The atomic ratio is about 1 : 5 (within 0.5% error). High-angle annular dark-field (HAADF) images were taken with the aberration corrected transmission electron microscope (JEOL ARM200 F). Figure 2 shows an image of a typical nanoribbon. The most prominent feature is the vertical arrow-like atomic chains. The image matches well with the expected atomic arrangement of the (110) plane of ZrTe5 depicted in the inset of Fig. 2(a), proving that these nanostructures are ZrTe5. The electron diffraction pattern looking down the [110] and [010] directions in Figs. 2(b) and 2(d) also agrees well with ZrTe5. Furthermore, we find that the lattice constants, a = 0.40 nm, b = 1.45 nm, and c = 1.34 nm, are very closed to the expected values for ZrTe5 as well.[12,15,23,24] From these images, it can be seen that the grown samples are of high crystalline quality.

All data show that the nanoribbons are along the a-axis, and the shortest dimension is along the b-axis. This growth mode is in fact expected considering the crystal structure. ZrTe5 is a layered material coupled by Van der Waals interactions, which favors 2D growth in principle. In each layer, there is a strong structural anisotropy, which leads to the dimension along the a-axis being significantly longer than the other dimension (along the c-axis).

Raman spectroscopic measurements were performed. The Raman spectra for narrow and wide nanoribbons are similar, except that the narrow ones have a weaker signal due to a smaller volume to interact with light. Four characteristic peaks, centered at 115 cm−1, 119 cm−1, 145 cm−1, and 179 cm−1, can be readily identified as the vibration modes of Te atoms.[40] The peak positions agree with those of the bulk as shown in Fig. 3(a).

Fig. 3. (a) Raman spectra of ZrTe5 nanoribbons. Four peaks at 115 cm−1, 119 cm−1, 145 cm−1, and 179 cm−1 are well resolved in the 20 μm wide ribbon, while the signal is weaker in the 0.5 μm wide ribbon. (b) Temperature dependence of resistivity for a 10 μm × 400 nm × 90 nm ZrTe5 nanoribbon. A broad maximum appears at TP = 141 K.
3.3. Transport measurements

The temperature dependence of resistivity of the nanoribbons, shown in Fig. 3(b), exhibits a maximum at 141 K. Such a nonmonotonic behavior is characteristic for ZrTe5.[29,4144] It origins from a temperature-induced Lifshitz transition.[27] The temperature of the resistivity maximum indicates a boundary between a strong topological insulator and a weak one.[45]

The quality of the nanoribbon is manifested in electrical transport. We have carried out magnetoresistance (MR) measurements under magnetic fields in different directions, as depicted in Fig. 4. The current is always along the nanoribbon, the a-axis. When the field is perpendicular to the current, MR is significant, while it is very small when the field is parallel to the current. In all three field directions, MR displays marked oscillations, the so-called Shubnikov–de Haas oscillations (SdHOs). Surprisingly, SdHOs start to appear in a field Bmin as low as 0.34 T as shown in Fig. 4(e), which suggests a high electrical mobility.[46] Due to difficulties in measuring the Hall of a nanoribbon, we estimate the mobility by the following method. Note that SdHOs stem from formation of Landau levels, which requires a condition of ωc τ > 1. Here ωc is the cyclotron frequency, and τ is the mean free time. It is easy to show that the condition is equivalent to ωc τ = μ B > 1, where μ is the mobility. Plugging Bmin = 0.34 T, we have μ > 3 × 104 cm2 ⋅V−1 ⋅s−1. The high mobility is beneficial in studying the transport properties of this topological material. Therefore, our work provides a promising growth method for high mobility nanostructured ZrTe5.

Fig. 4. (a) Optical micrograph of a typical four-point measurement device of a ZrTe5 nanoribbon. Magnetoresistance for a ZrTe5 nanoribbon when (b) Ba, (c) Bb, (d) Bc. (e) Shubnikov–de Haas oscillations appear at low field when Bb. (f)–(h) Shubnikov–de Haas oscillations with a smooth background subtracted for fields along the a-, b-, and c-axis.

In Fig. 4(c), it can be seen that the quantum limit is reached at a relatively small field, 5.5 T, indicating a low carrier density. The low doping level also suggests high quality of the sample, corroborating with other measurements. After subtracting a smooth background, the oscillations are clearly resolved, as shown in Figs. 4(f)4(h). By picking the field positions of the maxima and minima of the oscillations and plotting the inverse position against the Landau level index n, we obtain the Landau plot, shown in Fig. 5(c). Here, the maxima are assigned integer indices, while the minima are assigned half integer indices. Under this convention, an intercept of γb ≈ 0.143 on the y-axis is estimated from a linear fit when Bb. The intercept of the Landau plot has been widely used to determine the non-trivial Berry phase of π for topological systems.[47] For a 2D Dirac system, like graphene, γ = 0, while for 3D materials with a topological nontrivial band, the intercept is expected to be ±1/8.[12,16,30,47] Meanwhile, the intercepts of the Landau plots in Ba and Bc are γa ≈ 0.679 and γc ≈ 0.603, respectively, which are consistent with other reports.[22,30,48] The difference in the intercept for different field directions is caused by the spin zero effect.[48] Therefore, our data are in excellent agreement with a topological nontrivial system, confirming previous studies.[9,10,48,49]

Fig. 5. Analysis of SdH oscillations. (a) Temperature dependence of the oscillation amplitudes. Symbols are experimental data, while lines are best fits to Eq. (1). Black, blue, and red are for Landau levels n = 1, n = 2, and n = 3, respectively. (b) Temperature dependence of the SdH oscillation amplitude for fields along three axes. Open circle, Ba; open diamond, Bb; asterisk, Bc. Solid lines are fits to Eq. (1), from which the cyclotron masses are estimated as ma = 0.180me, mb = 0.034me, and mc = 0.154me. (c) Landau plot of the oscillations. The solid lines are linear fits.

The slope of the linear dependence gives the oscillation frequency Bf. For the three magnetic field orientations, Bf = 30.33 T, 5.14 T, and 25.67 T, respectively. According to the Lifshitz–Onsager relation, Bf = Se/2πe, where and e are the reduced Plank constant and the elementary charge, respectively, and Se is the extremal cross-sectional area of the Fermi surface in a plane normal to the magnetic field. Adopting an ellipsoidal Fermi surface, as suggested in earlier studies,[11,50,51] the Fermi wave vectors along the three principle axes can be estimated as ka = 0.115 nm−1, kb = 0.68 nm−1, and kc = 0.135 nm−1. Furthermore, the damping of the oscillation amplitude A with temperature can provide an estimation of the band velocity v0 of the massless Dirac fermion in the system, based on the Lifshitz–Kosevich relation

where kB is the Boltzmann constant. In Fig. 5(a). A(T)/A(0) for Bb is plotted against temperature for several Landau levels, n = 1, 2, and 3. By fitting the plots to Eq. (1), we obtain the velocities along the three directions, m⋅−1, m⋅−1, m⋅−1, in agreement with the values reported by others.[16,22,51] Next by using an energy dispersion for a massive Dirac Fermion , where M is the mass induced by the gap, we obtain the Fermi level EF = 38.3 meV. The relatively low Fermi level also indicates that our sample displays a low doping level.

4. Conclusions

We report growth of ZrTe5 nanoribbons by a silicon-assisted chemical vapor transport technique. Compared with the previous growth method for bulk crystals, our technique has the advantage of a fast growth rate. The grown nanoribbons are of high crystalline quality and display low unintentional doping and high mobility, suitable for study of topological properties near the charge neutrality point and beyond the quantum limit. Quantum transport experiments indicate that the ZrTe5 nanoribbon is a topological semimetal with a nontrivial Berry phase π in the ac plane. Our work is the first experiment on growth of ZrTe5 nanostructures and provides a good starting point for studies of nanostructured ZrTe5.

Reference
[1] Hasan M Z Kane C L 2010 Rev. Mod. Phys. 82 3045
[2] Qi X L Zhang S C 2011 Rev. Mod. Phys. 83 1057
[3] Kitaev A 2009 AIP Conf. Proc. 1134 22
[4] Roy R 2008 arXiv:0803.2868 [cond-mat.mes-hall]
[5] Schnyder A P Ryu S Furusaki A Ludwig A W W 2008 Phys. Rev. 78 195125
[6] Qi X L Hughes T L Raghu S Zhang S C 2009 Phys. Rev. Lett. 102 187001
[7] Wan X Turner A M Vishwanath A Savrasov S Y 2011 Phys. Rev. 83 205101
[8] Burkov A A Balents L 2011 Phys. Rev. Lett. 107 127205
[9] Song H D Sheng D Wang A Q Li J G Yu D P Liao Z M 2017 Chin. Phys. 26 037301
[10] Wang H Wang J 2018 Chin. Phys. 27 107402
[11] Zheng G Lu J Zhu X Ning W Han Y Zhang H Zhang J Xi C Yang J Du H Yang K Zhang Y Tian M 2016 Phys. Rev. 93 115414
[12] Yu W Jiang Y Yang J Dun Z L Zhou H D Jiang Z Lu P Pan W 2016 Sci. Rep. 6 35357
[13] Zhou Y Wu J Ning W Li N Du Y Chen X Zhang R Chi Z Wang X Zhu X Lu P Ji C Wan X Yang Z Sun J Yang W Tian M Zhang Y Mao H 2016 Proc. Natl. Acad. Sci. USA 113 2904
[14] Chen R Y Zhang S J Schneeloch J A Zhang C Li Q Gu G D Wang N L 2015 Phys. Rev. 92 075107
[15] Li Q Kharzeev D E Zhang C Huang Y Pletikosic I Fedorov A V Zhong R D Schneeloch J A Gu G D Valla T 2016 Nat. Phys. 12 550
[16] Yuan X Zhang C Liu Y Narayan A Song C Shen S Sui X Xu J Yu H An Z Zhao J Sanvito S Yan H Xiu F 2016 NPG Asia Mater 8 e325
[17] Wu R Ma J Z Nie S M Zhao L X Huang X Yin J X Fu B B Richard P Chen G F Fang Z Dai X Weng H M Qian T Ding H Pan S H 2016 Phys. Rev. 6 021017
[18] Pariari A Mandal P 2017 Sci. Rep. 7 40327
[19] Chen R Y Chen Z G Song X Y Schneeloch J A Gu G D Wang F Wang N L 2015 Phys. Rev. Lett. 115 176404
[20] Manzoni G Gragnaniello L Autès G Kuhn T Sterzi A Cilento F Zacchigna M Enenkel V Vobornik I Barba L Bisti F Bugnon Ph Magrez A Strocov V N Berger H Yazyev O V Fonin M Parmigiani F Crepaldi A 2016 Phys. Rev. Lett. 117 237601
[21] Li X B Huang W K Lv Y Y Zhang K W Yang C L Zhang B B Chen Y B Yao S H Zhou J Lu M H Sheng L Li S C Jia J F Xue Q K Chen Y F Xing D Y 2016 Phys. Rev. Lett. 116 176803
[22] Tang F Ren Y Wang P Zhong R Schneeloch J Yang S A Yang K Lee P A Gu G Qiao Z Zhang L 2019 Nature 569 537
[23] Weng H Dai X Fang Z 2014 Phys. Rev. 4 011002
[24] Niu J Wang J He Z Zhang C Li X Cai T Ma X Jia S Yu D Wu X 2017 Phys. Rev. 95 035420
[25] Lu J Zheng G Zhu X Ning W Zhang H Yang J Du H Yang K Lu H Zhang Y Tian M 2017 Phys. Rev. 95 125135
[26] Zhang J L Guo C Y Zhu X D Ma L Zheng G L Wang Y Q Pi L Chen Y Yuan H Q Tian M L 2017 Phys. Rev. Lett. 118 206601
[27] Zhang Y Wang C Yu L Liu G Liang A Huang J Nie S Sun X Zhang Y Shen B Liu J Weng H Zhao L Chen G Jia X Hu C Ding Y Zhao W Gao Q Li C He S Zhao L Zhang F Zhang S Yang F Wang Z Peng Q Dai X Fang Z Xu Z Chen C Zhou X J 2017 Nat. Commun. 8 15512
[28] Tang F Wang P Wang P Gan Y Wang L Zhang W Zhang L 2018 Chin. Phys. 27 087307
[29] Chi H Zhang C Gu G Kharzeev D E Dai X Li Q 2017 New J. Phys. 19 015005
[30] Liu Y Yuan X Zhang C Jin Z Narayan A Luo C Chen Z Yang L Zou J Wu X Sanvito S Xia Z Li L Wang Z Xiu F 2016 Nat. Commun. 7 12516
[31] Imura K I Okamoto M Yoshimura Y Takane Y Ohtsuki T 2012 Phys. Rev. 86 245436
[32] Qian X Liu J Fu L Li J 2014 Science 346 1344
[33] Zhou J J Feng W Liu C C Guan S Yao Y 2014 Nano Lett. 14 4767
[34] Zhang Y Vishwanath A 2010 Phys. Rev. Lett. 105 206601
[35] Peng H L Lai K J Kong D S Meister S Chen Y L Qi X L Zhang S C Shen Z X Cui Y 2010 Nat. Mater. 9 225
[36] Hong S S Zhang Y Cha J J Qi X L Cui Y 2014 Nano Lett. 14 2815
[37] Mourik V Zuo K Frolov S M Plissard S R Bakkers E P A M Kouwenhoven L P 2012 Science 336 1003
[38] Sodeck H Mikler H Komarek K 1979 Monatshefte For Chem. Chem. Mon. 110 1
[39] Levy F Berger H 1983 J. Cryst. Growth 61 61
[40] Taguchi I Grisel A Levy F 1983 Solid State Commun. 46 299
[41] Okada S Sambongi T Ido M Tazuke Y Aoki R Fujita O 1982 J. Phys. Soc. Jpn. 51 460
[42] DiSalvo F J Fleming R M Waszczak J 1981 Phys. Rev. 24 2935
[43] Tritt T M Lowhorn N D Littleton R T Pope A Feger C R Kolis J W 1999 Phys. Rev. 60 7816
[44] Manzoni G Sterzi A Crepaldi A Diego M Cilento F Zacchigna M Bugnon P Berger H Magrez A Grioni M Parmigiani F 2015 Phys. Rev. Lett. 115 207402
[45] Xu B Zhao L X Marsik P Sheveleva E Lyzwa F Dai Y M Chen G F Qiu X G Bernhard C 2018 Phys. Rev. Lett. 121 187401
[46] Wu M Zhang H Zhu X Lu J Zheng G Gao W Han Y Zhou J Ning W Tian M 2019 Chin. Phys. Lett. 36 067201
[47] Murakawa H Bahramy M S Tokunaga M Kohama Y Bell C Kaneko Y Nagaosa N Hwang H Y Tokura Y 2013 Science 342 1490
[48] Wang J Y Niu J J Yan B M Li X Q Bi R Yao Y Yu D P Wu X S 2018 Proc. Natl. Acad. Sci. USA 115 9145
[49] Chen Z G Chen R Y Zhong R D Schneeloch J Zhang C Huang Y Qu F Yu R Li Q Gu G D Wang N L 2017 Proc. Natl. Acad. Sci. USA 114 816
[50] Kamm G N Gillespie D J Ehrlich A C Wieting T J Levy F 1985 Phys. Rev. 31 7617
[51] Zheng G Zhu X Lu J Ning W Zhang H Gao W Han Y Yang J Du H Yang K Zhang Y Tian M 2017 Phys. Rev. 96 121401