2024 Volume 33 Issue 6
Article Contents

Fan Zhang, Xingyuan Hou, Yuxuan Jiang, Zongyuan Zhang, Yubing Tu, Xiangde Zhu, Genfu Chen, Lei Shan. Revisit of the anisotropic vortex states of 2H-NbSe2 towards the zero-field limit[J]. Chinese Physics B, 2024, 33(6): 067401. doi: 10.1088/1674-1056/ad362d
Citation: Fan Zhang, Xingyuan Hou, Yuxuan Jiang, Zongyuan Zhang, Yubing Tu, Xiangde Zhu, Genfu Chen, Lei Shan. Revisit of the anisotropic vortex states of 2H-NbSe2 towards the zero-field limit[J]. Chinese Physics B, 2024, 33(6): 067401. doi: 10.1088/1674-1056/ad362d

Revisit of the anisotropic vortex states of 2H-NbSe2 towards the zero-field limit

通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(4)

Article Metrics

Article views(170) PDF downloads(2) Cited by(0)

Access History

Revisit of the anisotropic vortex states of 2H-NbSe2 towards the zero-field limit

Abstract: We revisited the vortex states of 2H-NbSe2 towards zero fields by a low-temperature scanning tunneling microscope. Fine structures of the anisotropic vortex states were distinguished, one is a spatially non-splitting zero bias peak, and the other is an in-gap conductance anomaly resembling evolved crossing features around the center of the three nearest vortices. Both of them distribute solely along the next nearest neighboring direction of the vortex lattice and become unresolved in much higher magnetic fields, implying an important role played by the vortex–vortex interactions. To clarify these issues, we have studied the intrinsic vortex states of the isolated trapped vortex in zero fields at 0.45 K. It is concluded that the anisotropic zero bias peak is attributed to the superconducting gap anisotropy, and the spatially evolved crossing features are related to the vortex–vortex interaction. The vortex core size under the zero-field limit is determined. These results provide a paradigm for studying the inherent vortex states of type-II superconductors especially based on an isolated vortex.

1.   Introduction
  • It has been well-known that arrayed magnetic vortices carrying quantized flux ϕ0 form in a type-II superconductor between the lower and upper critical magnetic fields.[1] Centered around each vortex, the field maintained by the surrounding superfluid decays on the length scale of the penetration depth λ, and the order parameter Δ(r) recovers from zero to its asymptotic value on another length scale of the coherence length ξ.[2] It is predicted that the spatially confined quasiparticles lie in the discrete levels with the energies of μΔ2/EF (μ = ±1/2, ±3/2, …) in s-wave superconductors.[3–5] The study of vortex bound states has long been an attractive topic for the close correlation to the superconducting order parameters and the pairing mechanism. Moreover, it has been found that an additional half-integer level shift produces the robust Majorana zero modes (MZMs) in the topological vortices.[6–16] These fine electronic structures could be directly visualized by the scanning tunneling microscope/spectroscopy (STM/STS) for the ability to map the local density of states (LDOS) on the atomic scale.[17–21] By tracing the spatial dependence of the LDOS, the evolution of the vortex bound states could be unveiled, which also helps to identify the MZMs in the topological vortices.

    However, the superfluid around a vortex could overlap with each other in finite fields since the λ is much larger than ξ. Firstly, the reduction in vortex core size caused by the cancellation of the opposite supercurrent surrounding the adjacent vortices, which is defined as the vortex lattice squeezing effect, could significantly modify the distribution of js(r), and thus B(r) and Δ(r).[22,23] Secondly, as the magnetic field further increases, the bound state wavefunctions of the neighboring vortices will undergo hybridization, enabling the intervortex transfer of the quasiparticles.[24–26] These effects of vortex–vortex interactions have a giant influence on the identification of the vortex bound states. For example, the probability of MZMs in vortex cores of the candidate topological superconductors could decrease in higher fields due to the vortex–vortex interactions.[27] Most recently, a spatially non-splitting zero-bias conductance peak (ZBCP) was observed in a normal superconductor La in a low field, which was attributed to the anisotropy of its Fermi surface (FS).[28] In fact, theoretical works indicate that similar non-splitting ZBCP could also exist in the vortices of an s-wave superconductor with significantly anisotropic superconducting order parameters.[29,30] These results raised doubt on the explanation of the non-splitting ZBCP in vortex cores by MZMs in superconductors far from the quantum limit. As the most favored platform to study the trivial vortex states, 2H-NbSe2 (next abbreviated as NbSe2) possesses anisotropic order parameters as revealed by the angle-resolved photoemission spectroscopy (ARPES) measurements,[31] and exhibits spatially anisotropic vortex bound states.[32] Therefore, intrinsic vortex structures excluding vortex–vortex interactions are valuable to be studied in this archetypical system, in particular for an isolated vortex at the zero-field limit.

    In this paper, we report a close inspection of the anisotropic vortex states of NbSe2 single crystals especially under the extremely low magnetic fields and the zero-field limit by using a low-temperature STM. Fine structures of electronic states were revealed with a legible spatial evolution, which could not be well distinguished at much higher magnetic fields. One is an anisotropic non-splitting ZBCP that still exists until the zero-field limit, which could be ascribed to the anisotropy of the superconducting gap, in good agreement with a simulation based on quasi-classical theory. Another is an in-gap conductance anomaly resembling evolved crossing features around the center of the three nearest vortices, which originates from vortex–vortex interactions. Furthermore, the field-dependent vortex core size including zero-field vortex core size could be determined. These results provide a paradigm for studying the intrinsic properties of type-II superconductors based on an isolated vortex.

2.   Experimental details
  • High quality NbSe2 single crystals of clean limit studied were grown by the chemical vapor transport method with slightly excess selenide as the transport agent.[33] The transition temperature was about 7.1 K determined by the resistivity measurement. STM/STS measurements were carried out on a commercial facility (USM1300, Unisoku Co., Ltd.). The samples were mechanically cleaved in an ultrahigh vacuum (UHV) environment and then transferred immediately into the STM head maintained at a low temperature. Electrochemically etched tungsten tips were used after in situ treatment and calibration on Au(111) film. The topographic images were acquired in the constant-current mode with a bias voltage applied to the sample. The tunneling spectra of dI/dV ∼ V were obtained using the standard lock-in technique at 973.137 Hz. Most spectroscopic measurements were carried out at 0.45 K unless otherwise specified.

3.   Results and discussion
  • Figure 1(a) shows an atomically resolved topography of NbSe2 with the modulations of the incommensurate charge density waves. The primary vectors of the hexagonal lattice are labeled by the two arrows a and b. At the temperature of 0.45 K far below the Tc, the system entirely enters the superconducting state with the prominent gap dominating the tunneling spectrum represented by the black curve in the inset of Fig. 1(a). After applying a magnetic field of 200 mT, a broad ZBCP feature shows up at the vortex center due to the small energy interval Δ2/EF of the formed vortex bound states, as denoted by the red curve in the inset.[3] Thus, we can identify the spatially distributed vortices by imaging the conductance map of dI/dV (r, E = 0 meV) as shown in Fig. 1(b). They arrange into a hexagonal lattice with the field-dependent period of L=(2ϕ0/(3B))1/2, with the intervortex spacing approximately 109 nm at 200 mT. The nearest neighboring direction labeled as θ1 is exactly the same as the Se–Se directions. The LDOS around each vortex center shows a star-shaped modulation, with six rays extending from the vortex center along the next nearest neighboring direction labeled as θ2, and it also confirms that the sample is at the clean limit.

    Then we studied the evolutions of the LDOS along θ1 and θ2 at 0.45 K under different magnetic fields to reveal the anisotropic properties of the vortex states. For convenience to the comparison, the trajectories are normalized by the corresponding intervortex spacing L. Along the direction of θ1 shown in the false-color images of Figs. 1(c)–1(h), the established ZBCP at the vortex center splits into symmetric dispersive branches indicated by the dashed lines, which results from the quasiparticles wave function extending into a larger radius at higher energy. Notably, the branches indicated by the grey dashed lines become visible below 200 mT and consistently extend to L/2, while the extension length of the outer branch has a relatively weak field dependence estimated from the LDOS evolutions as indicated by the orange line segments. Along the direction of θ2 shown in Figs. 1(i)–1(n), two pairs of dispersive branches are also resolved under 200 mT as indicated by the black and grey dashed lines, respectively. Besides, another two subtle in-gap features are also identified, which are absent on the spectra along θ1. The first one is the conductance enhancement near the center of the three nearest vortices indicated by the red arrows. At the low fields under 200 mT, they become much clearer to be resolved as two gradually approaching X-shaped crossings. These crossing features are rarely noticed in earlier literature. The other feature is an emergent ZBCP as indicated by the red dash line. Without splitting in our resolution, it nearly extends to the location where the X-shaped crossings appear. These two zero-bias anomalies along θ2 seem to contribute to the six rays of the star-shaped vortex states together. Whereas, a spatially non-splitting ZBCP in vortex cores is always interpreted as an essential signature of MZMs in several topological superconductor candidates.[10–16] Thus, the observation is particularly abnormal for the topologically trivial electronic structure of NbSe2, and it will be instructive to further visualize these fine anisotropic features in this archetypical system.

    To clarify this issue, we measured the angle-dependent dI/dV spectra along a circle 40 nm away from a vortex center at 25 mT as displayed in Fig. 2(a). The angular coordinate begins from θ = 0°, which is defined as the atomic primary orientation a. Significantly, the non-splitting ZBCP appears periodically in the six-fold rotation symmetric directions of θ2. This is further verified by the periodically enhanced ZBCs in Fig. 2(b), which were extracted from the spectra in Fig. 2(a). Thus, the vortex bound states also host the six-fold symmetry but deflect 30° from the crystal structure and the vortex lattice, we can not determine the causation of the anisotropic ZBCP and the crossing features from either the vortex–vortex interactions[34] or the intrinsic FS or gap anisotropy.[29,30] These delicate features may be smeared out in finite magnetic fields or higher temperatures, which were rarely described in the previous studies on NbSe2. To further verify the key effect of extremely low magnetic fields on revealing the detailed vortex bound states, we compared typical spectra along θ2 under varied fields in Fig. 2(c). The sharp coherence peaks and the vivid in-gap features become substantially obscured when the applied field exceeds 100 mT, though the field induced pair-breaking effect is not that intense. Hence, it is reasonable to speculate that the intervortex transfer of the quasiparticles smears distinct features on the density of states, making a significant influence on the intrinsic vortex bound states.

    In this sense, intrinsic vortex structures excluding vortex–vortex interactions are strongly desired to be revealed. With the temperature constant at 0.45 K, there is a certain probability of observing an isolated vortex by applying a magnetic field along the c direction and then withdrawing the field. The trapped isolated vortex provides an excellent platform to shed light on this issue. Figure 3(a) displays the ZBC mapping of a trapped isolated vortex at 0 T, which exhibits intrinsically anisotropic characteristics with six rays extending along the directions of θ2 (shown by the red dash lines). This suggests that the anisotropic vortex bound states are rooted in the intrinsically anisotropic properties of NbSe2 but not the neighboring vortices interactions. The LDOS along the θ1 direction in Fig. 3(b) shows similar evolution to those measured in the low fields, with a broad ZBCP in the vortex center splitting into two pairs of dispersive branches within our best resolution. For the evolution along the direction of θ2 shown in Fig. 3(c), there still exists the robust ZBCP character extending more than 100 nm away from the dispersive branches as indicated by the dash lines. It should be pointed out that, we failed to observe this anisotropic ZBCP at 1.6 K either in vortices under extremely low fields or a trapped isolated vortex at zero field, lacking robustness against the thermal excitations in addition to the vortex–vortex interactions. On the other hand, the above-mentioned small X-shaped crossing features along θ2 outside the vortex core are no longer visible. These results clearly show that the X-shaped crossing features arise from the vortex–vortex interaction while the anisotropic non-splitting ZBCP is an intrinsic feature of the vortex bound state of NbSe2.

    From this point of view, it has been theoretically explained that anisotropic vortex states can result from the intrinsic anisotropy of the FS or superconducting gaps.[28–30] The latter is known to be much more pronounced than the former in NbSe2.[31,35,36] Unfortunately, the models were difficult to verify experimentally for lacking a truly isolated vortex in the past to the best of our knowledge. We simulated the LDOS of an isolated vortex with anisotropic gaps to compare with our results by using the quasi-classical Green function approach within the self-consistently obtained pair potential.[37] Figure 3(d) shows the calculated ZBC around the isolated vortex by using an anisotropic superconducting gap Δ(θ) = Δ0 [1+ (1/3)cos(6θ)], which reproduces the star-shaped vortex states. Figures 3(e) and 3(f) show the low energy LDOS along the two special directions θ1 and θ2, respectively. Besides, we extracted typical spectra from Figs. 3(b), 3(c), 3(e), and 3(f) for comparison as shown in Fig. 3(g). It can be seen that the discrete energy levels split into symmetric branches along the two characteristic directions, however, a unique anisotropic non-splitting ZBCP is formed only along the θ2 direction, showing good agreement with our experimental observations. This suggests that the anisotropic non-splitting ZBCP of NbSe2 stems from the anisotropy of the superconducting gap, while the vortex–vortex interactions also produce an essential impact on the performance of the vortex bound states. There is still a slight difference between theoretical simulation and experiment. To completely reproduce the performance of theoretical calculations, lower temperature and higher energy resolutions are required. Due to their topologically trivial electronic structure, these results call for caution when interpreting spatially non-splitting ZBCP in vortex cores as an essential signature of MZMs in superconductors with anisotropic gaps or FS far from the quantum limit.

    Since the vortex bound states are sensitive to the magnetic field, it is meaningful to determine the vortex core size ξc towards the zero-field limit. The vortex core size is defined as ξc = Δ0[dΔ(r)/dr|r→0]−1, which is usually obtained by fitting the evolution of the normalized ZBC across the vortex center to evaluate the superconducting coherence length.[38,39] An accurate model with convergence condition dΔ(r)/dr|r→L/2 = 0 for the two-band system was adopted to determine the value of ξc, which takes the periodicity of the vortex lattice into account.[40,41]

    where x = 2r/L, ην = 2ξcν/L, ν = 1, 2, and L is the nearest neighbor spacing of vortices. We have performed this analysis to extract the field-dependent ξc by fitting the ZBC profiles across the vortex center along the θ1 direction, avoiding the influence of the zero energy anomalies along θ2. Figures 4(a) and 4(b) display the double-coordinate normalized experimental ZBC data for T = 0.45 K and 1.6 K, respectively. It can be directly observed that as the magnetic fields increase, the individual vortex gradually expands in the Wigner–Seitz unit cell of the vortex lattice, resulting in enhanced interactions between each other. This is also the primary factor contributing to the formation of a hexagonal vortex lattice in the absence of a strong pinning effect. We find that the data can be well described with the same ξc for two bands as demonstrated by the corresponding best-fitting curves, which is consistent with the behavior of multiband superconductors NbSe1.8S0.2, NbS2 and CaKFe4As4.[40,41] Recently, theoretical considerations re-examine the almost identical vortex core sizes in two-band superconductors with a microscopic approach that covers a broad temperature and magnetic field regime.[42] Such a behavior which is also known as “length-scales locking” is a robust phenomenon for a moderately coupled multiband system.

    To definitively confirm this and eliminate any influence from vortex–vortex interactions, it is preferable to determine the intrinsic vortex core size based on a single trapped vortex in the absence of external fields. In this case, the intervortex spacing L used in the model should ideally be infinite. We have concluded that ξcs converge when the intervortex spacing L is taken to be sufficiently large during the fitting process for the isolated vortex. Figure 4(c) presents the ZBC profiles of the isolated single vortex with the fitted curves superposed on them for temperatures of 0.45 K and 1.6 K, respectively. Remarkably, the fittings using the same ξc still exhibit excellent agreement with the experimental data, implying a not weak interband coupling in this system.[42,43] We could finally define the vortex core sizes in the zero-field limit based on the same model. Upon initial inspection, the size of the isolated single vortex at 1.6 K is much larger than that at 0.45 K, in spite of the little superconducting gap variation deduced from the Bardeen–Cooper–Schrieffer relationship.[44] This thermal excited expansion, known as the Kramer–Pesch effect, can be attributed to the increased population of the in-gap quasiparticle states with larger wave function radii.[45]

    Moreover, the ξc values of the trapped single vortex are respectively larger than those of any other vortices in finite fields at the same temperature. In clean superconductors, the shrinkage of the vortex core with the increasing magnetic field at low temperatures can be attributed to two key factors related to vortex–vortex interactions. The first one is the vortex lattice squeezing effect, which arises from the overlap of the superfluid surrounding the vortex cores, particularly the cancellation of opposite supercurrents between the adjacent vortices. This overlap significantly alters the overall distribution of js (r), and then modifies B(r) and Δ(r).[22] The second factor is the quasiparticle delocalization effect induced by the quasiparticle transfer between the neighboring vortices due to the wavefunction hybridization of their in-gap bound states.[24–26]ξcs show a decrease at the extremely low fields and then the weak field dependence at 0.45 K as guided by the green curve. The vortex core size behaviors we obtained from the density of states are relatively consistent with that obtained from the μSR experiment by fitting B(r).[22] In the μSR experiment, this non-monotonic behavior is attributed to the different quasiparticle localization effects of the different gaps.[22,46–48] However, this behavior of the vortex core sizes is consistent with that predicted by theory for multiband systems with “length-scales locking”.[42] As a side note, such measurement is difficult to achieve through other methods in the zero-field limit, highlighting the unique role played by STM on the trapped single vortex at 0 T. Furthermore, the trapped vortices could be artificially enhanced by introducing defects or adatoms as the pinning centers into various superconducting systems.[49–54]

4.   Conclusion
  • In conclusion, we revisited the fine structure of the vortex states of 2H-NbSe2 under extremely low magnetic fields towards zero fields by low-temperature scanning tunneling microscope. Some anisotropic fine structures were revealed on the evolving spectra across the vortex center along the next nearest neighboring direction, including some crossing features and spatially extended zero bias peak. Based on the results around a trapped isolated vortex without any external magnetic field, these features were respectively identified as the effect of the vortex–vortex interactions and the intrinsic superconducting gap anisotropy. Furthermore, we showed the relationship between vortex core size and magnetic fields at different temperatures, which contains vortex core size under the zero-field limit. It provides an effective way to determine the vortex states at the zero-field limit, which could be widely used for many type-II superconductors.

Figure (4)  Reference (54)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return