Resistive phase transition of the superconducting Si(111)-()-In surface
© Uchihashi et al.; licensee Springer. 2013
Received: 19 February 2013
Accepted: 21 March 2013
Published: 11 April 2013
Skip to main content
© Uchihashi et al.; licensee Springer. 2013
Received: 19 February 2013
Accepted: 21 March 2013
Published: 11 April 2013
Recently, superconductivity was found on semiconductor surface reconstructions induced by metal adatoms, promising a new field of research where superconductors can be studied from the atomic level. Here we measure the electron transport properties of the Si(111)-()-In surface near the resistive phase transition and analyze the data in terms of theories of two-dimensional (2D) superconductors. In the normal state, the sheet resistances (2D resistivities) R□ of the samples decrease significantly between 20 and 5 K, suggesting the importance of the electron-electron scattering in electron transport phenomena. The decrease in R□ is progressively accelerated just above the transition temperature (T c ) due to the direct (Aslamazov-Larkin term) and the indirect (Maki-Thompson term) superconducting fluctuation effects. A minute but finite resistance tail is found below T c down to the lowest temperature of 1.8 K, which may be ascribed to a dissipation due to free vortex flow. The present study lays the ground for a future research aiming to find new superconductors in this class of materials.
Semiconductor surface reconstructions induced by metal adatoms constitute a class of two-dimensional (2D) materials with an immense variety [1, 2]. They are considered one form of atomic layer materials which can possess novel electronic properties and device applications [3, 4]. Recently, superconductivity was measured by scanning tunneling microscopy (STM) for atomically thin Pb films [5, 6] and three kinds of Si(111) surface reconstructions: SIC-Pb, ()-Pb, and ()-In . This discovery was followed by a demonstration of macroscopic superconducting currents on Si(111)-()-In by direct electron transport measurements . These findings are important because they enable us to create superconductors from the atomic level using state-of-the-art nanotechnology. In addition, the space inversion symmetry breaking due to the presence of surface naturally leads to the Rashba spin splitting [9, 10] and may consequently help realize exotic superconductors .
In reference, we have unambiguously clarified the presence of Si(111)-()-In (referred to as ()-In here) superconductivity. However, systematic analysis on electron transport properties above and below the transition temperature (T c ) is still lacking. For example, 2D superconductors are known to exhibit the precursor of phase transition due to the thermal fluctuation effects just above T c [12–14]. Superconductivity is established below T c , but vortices can be thermally excited in a 2D system. Their possible motions can cause the phase fluctuation and limit the ideal superconducting property of perfect zero resistance . These fundamental properties should be revealed before one proceeds to search for new superconductors in this class of 2D materials.
In this paper, the resistive phase transition of the ()-In surface is studied in detail for a series of samples. In the normal state, the sheet resistances (2D resistivities) R□ of the samples decrease significantly between 20 and 5 K, which amounts to 5% to 15% of the residual resistivity Rn,res. Their characteristic temperature dependence suggests the importance of electron-electron scattering in electron transport phenomena, which are generally marginal for conventional metal thin films. T c is determined to be 2.64 to 2.99 K and is found to poorly correlate with Rn,res. The decrease in R□ is progressively accelerated just above T c due to the superconducting fluctuation effects. Quantitative analysis indicates the parallel contributions of fluctuating Cooper pairs due to the direct (Aslamazov-Larkin term) and the indirect (Maki-Thompson term) effects. A minute but finite resistance tail is found below T c down to the lowest temperature of 1.8 K, which may be ascribed to a dissipation due to free vortex flow.
We note that, although the nominal coverage of the evaporated In is more than several monolayers (ML), post annealing removes surplus In layers and establishes the ()-In surface. The In coverage of this surface reconstruction was originally proposed to be 1 ML for the ‘hexagonal’ phase (()-In-hex) and 1.2 ML for the ‘rectangular’ phase (()-In-rect) , where 1 ML corresponds to the areal density of the top-layer Si atoms of the ideal Si(111) surface. However, recent theoretical studies point to the coverages of 1.2 ML for the ()-In-hex and of 2.4 ML for the ()-In-rect [21, 22]. For our experiments, the dominant phase is likely to be the ()-In-hex judging from the resemblance of the obtained STM images (Figure 1b) to the simulated image of the ()-In-hex (Figure two, panel b in ). The relation between the surface structure and the superconducting properties is intriguing and will be the subject of future work.
In the previous study, van der Pauw’s measurement was adopted to check the anisotropy of electron conduction and to exclude the possibility of spurious supercurrents. In this setup, however, transport characteristics should be analyzed with care because the spatial distribution of bias current is not uniform. To circumvent this problem, in the present study, we adopted a configuration with a linear current path between the voltage terminals (Figure 1d). The black regions represent the area sputtered by Ar + ions through the shadow mask. The figure also shows the current density distribution calculated by the finite element method in color scale, which confirms that it is homogeneous between the voltage probes. This allows us to determine the sheet resistance R□ of the sample in a more straightforward way: R□=(V/I)×(W/L), where V is the measured voltage, I is the bias current, W=0.3 mm is the width of the current path, and L=1.2 mm is the distance between the voltage probes. Figure 1e shows the optical microscope image of a sample, confirming the clear boundary between the shadow-masked and sputtered regions. Although the sputtering was very light, the resulting atomic-scale surface roughening was enough to make an optical contrast between the two regions.
Following the sample preparation, four Au-coated spring probes were brought into contact with the current/voltage terminal patterns in a UHV-compatible cryostat. Four-terminal zero bias sheet resistance R□ was measured with a DC bias current I=1 µA, and the offset voltage was removed by inverting the bias polarity. To access the electron conduction only through the ()-In surface at low temperatures, Si(111) substrates without intentional doping (resistivity R>1,000 Ω cm) were used. Leak currents through the substrate and the Ar +-sputtered surface region were undetectably small below 20 K, which allowed precise measurements in this temperature region.
Summary of the fitting analysis on the resistive transition of the ()-In surface
T c (K)
The determined T c ranges from 2.64 to 2.99 K. This is in reasonable agreement with the previously determined value of T c =2.8 K, but there are noticeable variations among the samples. The normal residual resistance Rn,res also shows significant variations, ranging from 108 to 394 Ω. These two quantities, T c and Rn,res, could be correlated because a strong impurity electron scattering might cause interference-driven electron localization and suppress T c . However, they are poorly correlated, as shown in the inset of Figure 2. This is ascribed to possible different impurity scattering mechanisms determining Rn,res and T c as explained in the following. Electron scattering should be strong at the atomic steps because the surface layer of ()-In is severed there. Therefore, they contribute to most of the observed resistance [8, 24]. However, the interference between scatterings at the atomic steps can be negligibly weak if the average separation between the atomic steps dav is much larger than the phase relaxation length L ϕ . This is likely to be the case because dav≈400 nm for our samples, and L ϕ is several tens of nanometer for typical surfaces . In this case, electron localization and resultant suppression of T c are dominated by other weaker scattering sources within the size of L ϕ , not by the atomic steps that determine Rn,res.
where C′ is a proportional constant, εF is the Fermi energy, and kB is the Boltzmann constant. The log term in Equation 4 results in a weaker temperature dependence than that in a 3D Fermi liquid (∝T2). Fitting the data with Equation 4 instead of the C T a term in Equation 1 gives εF≈0.1 eV, although the uncertainty is quite large.
We note that a decrease in resistance in a conventional metal film is usually very small in this temperature range. For example, it is less than 1% within the range of 2<T<20 K for 2-nm-thick single-crystal Nb films, although R□=122 Ω of the film is comparable to the observed Rn,res in the present experiment . For a metal thin film with a large resistance, R□ even increases slightly with decreasing T as a consequence of the electron localization . In clear contrast, a decrease in R□ between 20 and 5 K in our samples, Δ R□, amounts to as much as 5% to 15% of Rn,res (see Figure 2 and Table 1). In this sense, the observed temperature dependence is rather unusual. The ()-In surface studied here has an atomic-scale dimension in the normal direction and may thus have an enhanced electron-electron interaction because of insufficient electrostatic screening. In comparison, the contribution from the electron-phonon interaction can be smaller because it decreases rapidly at low temperatures as Re-ph∝T5.
where ξ is the Ginzburg-Landau coherence length, R□,n is the normal sheet resistance of the sample, B is the magnetic field perpendicular to the suface plane, and Φ0=h/2e is the fluxoid quantum. A crude estimation using ξ=49 nm,R□,n=290 Ω, and B=3×10−5 T gives R□,v=6.3×10−2Ω, which is in the same order of magnitude as the observed value of approximately 2×10−2Ω. We note that ξ=49 nm was adopted from the value for the Si(111)-SI-Pb surface , and ξ is likely to be smaller here considering the difference in T c for the two surfaces. The present picture of free vortex flow at the lowest temperature indicates that strong pinning centers are absent in this surface superconductor. This is in clear contrast to the 2D single-crystal Nb film , where the zero bias sheet resistance was undetectably small at sufficiently low temperatures. In accordance with it, the presence of strong vortex pinning was concluded from the observation of vortex creep in . This can be attributed to likely variations in local thickness of the epitaxial Nb film at the lateral scale of vortex size . The absence of ‘local thickness’ variation in the present surface system may be the origin of the observed free vortex flow phenomenon.
where and R□,n are identified with Rn,res=293 Ω of sample S1 here. However, Equation 7 is derived from the assumption of the dirty-limit BCS superconductor, which is not applicable to the ()-In surface with high crystallinity. Unfortunately, the present experimental setup does not allow us to observe the expected temperature dependence of Equation 6 down to T K because of the presence of the stray magnetic field. Furthermore, the predicted I-V characteristics V∝I a where the exponent a jumps from 1 to 3 at T K should be examined to conclude the observation of the KT transition [31, 32]. Construction of a UHV-compatible cryostat with an effective magnetic shield and a lower achievable temperature will be indispensible for such future studies.
We have studied the resistive phase transition of the ()-In surface in detail for a series of samples. In the normal state, the sheet resistances R□ of the samples decrease significantly between 20 and 5 K, which amounts to 5% to 15% of the residual resistivity Rres. Their characteristic temperature dependence suggests the importance of electron-electron scattering in electron transport phenomena. The poor correlation between the variations in T c and Rres indicate different mechanisms for determining these quantities. The decrease in R□ was progressively accelerated just above T c due to the superconducting fluctuation effects. Quantitative analysis indicates the parallel contributions of fluctuating Cooper pairs corresponding to the AL and MT terms. A minute but finite resistance tail was found below T c down to the lowest temperature of 1.8 K, which may be ascribed to a dissipation due to free vortex flow. The interpretation of the data based on the KT transition was proposed, but further experiments with an improved cryostat are required for the conclusion.
This work was partly supported by World Premier International Research Center (WPI) Initiative on Materials Nanoarchitectonics, MEXT, Japan, and by the Grant-in-Aid for JSPS Fellows. The authors thank M. Aono at MANA, NIMS, for his stimulous discussions.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.