Pictures from the boat tour

Group picture

Group picture

The organizers: from the left Pratap Raychaudhuri, Antonio Garcia-Garcia, Lara Benfatto, Jerome Lesueur, Andrea Caviglia

Friday 31 July 2015- Jim Valles – Summary talk

Jim starts his talk by proposing a general scheme to classify the several topics covered by the workshop:

1-LAO/STO interfaces, status and advances

2- Pursuing related issues in superconductivity

3- SIT Physics in Thin films

4- Disorder and the SIT

5- Novel experimental probes

6- Interfaces matter

1- LAO/STO interfaces, status and advances. Jim remarks his personal surprise in discovering the beautiful world of interfaces, where the beauty actually come out from plenty of different features at play: SC, magnetism, spin-orbit (SO) scattering, gate tunability. The superconductivity is not so conventional, as shown by experiments of tunneling (Mannhart), by the Hc anisotropy (Gariglio) and the critical behavior (Bergeal). In the last case the beauty is that one can scale several samples because of the tunability in Vg, not only one or two, and this makes definitively a big advantage with respect to other system. Magnetism: it can be enhanced with a layer of ETO put in between (Stornaiuolo), giving rise to an anomalous Hall effect. One can also use Nb doped STO to boost the superconductivity (Hwang). How can we use this tunability? One proposal came out of an intrinsic spin-Hall effect induced by a modulated Rashba coupling (Seibold), motivated by a general phase-separation model (Grilli) with a possible QCP.

2- Pursuing related issues in superconductivity. Jim reviews then some other fundamental issues of superconductivity, triggered or not by the LAO/STO physics. One is the effect of strong SO coupling on superconductivity (Michaeli). The mechanisms leading to a SC dome: the one shown for granular AL (Pracht) resembles something we have seen, is it the same as in other systems or not?  Can superconductivity be triggered by a Van Hove singularity in Nanotubes (Barbara)? Can we go towards artificially-generated “molecular” superconductors (Ilani)? Superconductivity can even occurs at the nanoscale (Bose): even rare-grain effects (Mason) can be relevant to the Tc .

3- SIT Physics in thin films. Here Jim shows something exciting from himself! One can pattern Bi films with holes. According to the modulation of the underlying substrate one obtains different R(T) (direct SIT or percolation). Jim then compares flat films with hilly ones. In the latter ones he sees quantum oscillations with 2e periodicity even on the insulating side of the transition, while these are not seen in the flat films. Conclusion: if one wants a Bose insulator one needs inhomogeneity. The persistence of pairing on the I side of the SIT has been seen often during the workshop. For example, in InOx samples with very different Tc’s loose the magnetoresistance peak at the same field (Shahar), real signatures of bosons surviving across the SIT. The same occurs in decorated graphene (Bouchiat), even if here strictly speaking there is a metal in between (and plateaus..).

4- Disorder and the SIT. The route is difficult (as shown by  an explicative slide by Svortsov on Monday!). Short-scale effects can affect strongly the non-universality. So while in Leridon’s talk we have seen that the fluctuations seem to suggest OD physics (due to grains?), different disorder realizations can lead to inhomogeneous currents (Castellani), they can affect the STM DOS (Misha), and they can explain the failure of the Mattis-Bardeen picture of the conductivity near the SIT (Armitage). Here again Jim goes back to his own work in NHC films and addresses a different type of disorder, i.e. the flux disorder. How does it affect the quantum critical transport at the SIT? Jim shown that this can be realized by fabricating hole arrays with varying geometrical order. The spreading in the hole size induces magnetic-field oscillations or flux disorder (it is not ‘exactly’ one flux for hole, up to 10-20 %). Such a flux disorder increases with the magnetic field B. This implies that one can identify plateaus in the resistance at several B, like if one had multiple B-induced SIT. More remarkable, one finds out that the critical resistance as a function of flux disorder increases up to the ‘universal’ Rc, so the expectation of an universal Rc is not realized.  

5- Novel experimental probes. Here we had several nice examples: scanning critical current microscopy on a nanowire (Driessen), non-Gaussian noise in NbN (Ghosh), light-induced superconductivity in cuprates (Wanzheng Hu), a novel four-probe technique for measuring resistance in ultrathin FeSe films (Jia).

6- Interfaces matter. Jim concludes with a very nice comment: “interfaces are our friends”. They can help: one can design a Josephson junction to explore the properties of a topological superconductor (Brinkman), or one has to use a STO substrate to enhance the (tunneling) Tc of FeSe (Wang).

Overall, Jim talk showed that a common language is possible for scientists working apparently on different materials, with different experimental probes, and speaking different theoretical ‘slangs’.  And people like to discuss when there are the right conditions to do that. With Jim’s talk we close this exciting workshop and we agree that we should definitively plan to have something similar in few more years.

Blogged by Lara Benfatto

Friday 31 July 2015 - Mark Blamire - Superconducting tunneling through spin filter barriers

During this workshop, we have heard about Cooper pairs being exposed to a lot of weird conditions (as compared to the original BCS scenario) such as strong Rashba Spin Orbit coupling, extreme surface confinement, and above all, disorder of all kinds that one can think of. Then came Daniela Stornaiuolo, who proposed to make them coexist with a ferromagnetic order in new hetero-structures, by including a thin EuTiO3 layer within the regular LAO/STO interface. But the hard time for the Cooper pairs arrived when Mark forced them to cross a ferromagnetic barrier.

Indeed, Mark exposed us very beautiful experiments where Josephson and tunneling junctions are made of a thin GdN barrier sandwiched between two NbN superconducting ones. At low thickness, GdN is not magnetic, but for thickness beyond a few nanometers, a ferromagnetic behavior is observed, with a weak coercive field. In that case, the junction acts as a spin filter for 80 to 90% of the electrons. I-V characteristics shows Josephson coupling, weaker when GdN is magnetic as expected because of the strong depairing effect at play. More interesting, the Ic x Rn product (critical current times the normal state resistance) of the magnetic Josephson Junctions (JJ) deviates from the regular Ambegaokar-Baratoff law at low temperature. Finally, the Fraunhofer pattern, that is the modulation of Ic with an applied magnetic field, reveals the spin filtering effect. Indeed, it appears to be first highly asymmetric in field, and second, to shift according to the ramping direction of the magnetic field. The magnetization hysteresis loop of the GdN layer accounts for the latter behavior, since the total magnetic flux that goes through the JJ controls the critical current. However, the analysis of the asymmetry shows that, when the barrier becomes magnetic, the first order term in the current-phase relationship within the JJ is weakened, and that the second order one dominates. This is not completely understood yet, even if recent theoretical papers made interesting propositions to explain this behavior.

Then, Mark described experiments where one of the electrode of the junction is normal, in order to measure the density of states of a superconductor through a spin filtering barrier, in the spirit of the pioneer work of Tedrow and Meservey a long time ago. In that case, Zeeman splitting shifts the superconducting gap, and an offset in energy is observed in the tunneling conductance curve, corresponding to an internal field in the barrier of 1.5 T. The sign of the asymmetry tells that the magnetism acts more on one side of the junction. This offset has also been seen in JJ, indicating that there is an intrinsic asymmetry in the junctions, that could account for the disappearance of the first order term in the current-phase relation. Growth considerations and intermixing at the interfaces might explain this asymmetry.

But the story is not over, and Mark made a very nice teasing by showing us the low temperature tunneling conductance curves of the magnetic junctions, where a big zero-bias conductance peak builds up at zero energy. As Jim Valles mentioned, this immediately reminded me the old days of High Tc  superconductors, when I was discovering such an anomaly in YBCO/Pb junctions, which have been later on attributed to a zero-energy Andreev state at the (110) surface of a d-wave superconductor. This is obviously a different situation, but that may be a bound state of some sort.

Mark let the discussion open about this peak ... and the Cooper pairs again in a strange situation ...

Blogged by Jerome Lesueur

Friday 31 July 2015- Emilio Artacho - On the origin of the two-dimensional electron gas at the interface between insulating perovskites

Coming from the field of electronic structure and DFT, Emilio Artacho discusses “On the origin of

the 2D electron gas at the interface between insulating perovskites”. He introduces the “polar catastrophe” argument for LaAlO3/SrTiO3 – for an idealized heterostructure, the charged layers in LaAlO3 up against the neutral layers in SrTiO3 creates electrostatic boundary conditions which would be resolved with 0.5 electrons/2D unit cell for one interface, and 0.5 holes for the other.  Assuming idealized structures, and bulk stoichiometry, the electric field building up, as the LaAlO3 layers are stacked, drives charge transfer from the surface to the interface. DFT calculations (on superlattices) show this and correspond quite well with a simple parallel plate capacitor model.

Emilio wants to convey 2 key points in his talk:

1) The 0.5 interface charge is not only what you find in the simple ionic limit – rather it is robust to the realistic generalization (including covalency, etc.). To show this, he points out that the dipole moment/unit cell is an ill-defined concept, in that it is dependent on the choice of the origin (pointed out by Richard Martin in 70’s). A “dipole-free” unit cell can be chosen and thus projecting all the charge issues to the surface. David Vanderbilt used these ideas to make the “Berry’s phase connection” (in his case for ferroelectrics) in the 90’s, such that the boundaries can be treated much as is the case for topological insulators nowadays. Another analogy is the 1D Haldane chain, in that the relevant spin degrees of freedom are at the ends of the chain. Ultimately, the 0.5 charge is the robust consequence of being between two materials with different topological index.

2) The second point he emphasizes is that rather than purely dealing with bulk idealized stoichiometry (discussed above), another relevant degree of freedom is the stoichiometry – i.e., the 2DEG can be triggered by redox processes. The overall point is that the electrostatic boundary conditions can be resolved not just by “mobile electrons”, but also by defect chemistry. Ultimately, both processes are at play and may interplay.

Final points he makes include: not all carriers at the interface are mobile; disorder is expected; and that depending on origin, we may not necessarily have Mott-Anderson behavior, free carriers moving in a smoothly, weakly disordered potential.

Discussions include possible connections to phase separation scenarios previously discussed in the conference (Grilli et al.).

Blogged by Harold Hwang

Friday 31 July 2015 – Jianming Lu- Huge upper critical field of ionic gated MoS2

Jianming Lu reported on the latest experimental studies of superconductivity in transition metal dichalcogenides (TMDs) done in Prof. Ye’s group. The key aspect of these experiments is the use of ionic liquids to create electric double layer transistors, where a sheet of charge is

induced on the surface of the TMD by applying a gate voltage. This gating technique is extremely efficient due to the close (few nanometers) spacing between the charge and the material surface, creating an extremely large gate capacitance. As a result, high carrier densities can be induced in the TMD channel by applying just a few volts.

Prof. Ye’s group used this technique to study the effect of electrical doping on a variety of materials, including superconductors (ZrNCl), metals (Au), semiconductors and semimetals (MoS2 and graphene). This talk focused on Mo-based TMDs.

Ionic gating of MoS2, MoSe2, and MoTe2 revealed superconductivity on the electron-doping side for the first two materials. No superconductivity was reported on the hole-doping side and MoTe2 did not show any superconductivity. MoS2 and MoSe2 showed similar phase diagrams, with the dome-shaped superconducting phase starting at electron density n >  0.6X10^14/cm^2, and the critical temperature increased up to about 10 K for MoS2 and 6.5 K for MoSe2.

All these TMDs were multilayer flakes. However the speaker argued that the doping was mainly affecting the top layer, effectively decoupling it from the remaining layers and making this flake a monolayer superconductor.  The supporting evidence presented was: 1) the angular dependence of the critical magnetic field, showing a cusp for the field direction parallel to the MoS2 plane and 2) the superconducting transition showing a KT tail. This point generated questions from the audience on whether the same effects could be seen if doping and superconductivity were extended to more than one layer. In support to the single-layer argument, Jianming Lu mentioned that they had measured single layer samples and also found superconductivity with similar Tc and Tc dependence on doping. However, the single-layer work is still in progress therefore the data were not included in the presentation.

A striking result was the magnitude of the critical magnetic field in the direction parallel to the flake, larger than 80 T, much larger than the critical field measured for chemical doped bulk MoS2 and far exceeding the Pauli limit. The speaker argued that this large critical field can be explained by orthogonal protection due to Zeeman spin-orbit coupling, aligning the spins in the out-of-plane direction.

These new exciting experimental results show once again that ionic gating is a very powerful technique to uncover the rich physics of low-dimensional materials in the large carrier density regime.

Blogged by Paola Barbara

Thursday 30 July 2015- Karen Michaeli - Superconductivity in the presence of spin-orbit coupling: old dog, new trick

Some superconducting films display an increase of Tc in the parallel magnetic critical field. This strengthening of superconductivity calls for an explanation  involving both spin-orbit coupling (SOC) and magnetic field. With this motivation Karen introduces a continuous free-electron model in the presence of  a Rashba SOC. When a magnetic field B along x is turned on, the Zeeman field generically shifts the chiral bands in opposite directions along y. However, for small momenta q=2mu_0B/v_F the lower Rashba band no longer depends linearly on B and therefore identifies a circular Fermi surface centered at zero momentum, while the second chiral band is shifted by q giving  a SC order parameter Delta(r ) =Delta exp(iqr) similar to the FFLO case. Since the pairs in the first band don't depend on B, the decoherence effect of B only arises from the pairs in the second band.  Disorder has a non trivial effect on this finite-momentum SC state: at low disorder, pairs scattered in the smaller B-dependent branch of the Fermi surface stay there for long time and suffer a strong pair breaking. This leads to rapid decrease of the critical field Bc on the disorder scattering. Increasing disorder pairs in the B-dependent branch scatter more frequently in the B-independent branch and suffer less pair-breaking leading to a strengthening of SC and a recovery of the critical field with disorder. This justifies the choice of a model where disorder is assumed to  kill triplet SC, while the singlet finite-momentum pairing is mildly affected.

The SC state in the presence of B is also characterized by a finite magnetization which enters the free energy via the SOC. According to the Edelstein magnetoelectric effect, the supercurrent is accompanied by a transverse magnetic moment, which also acquires a monopole structure when a supercurrent vortex is present. 

The (Gibbs) free energy is then transformed passing to a lattice  XY model having additional terms arising from the Rashba-like magnetoelectric terms. Once the magnetic degrees of freedom are integrated out one obtains a free energy for a classical spin model with nearest-neighbor ferromagnetic coupling (favoring uniform SC) and a frustrating term proportional to the Rashba SOC leading to an helical magnetic solution corresponding to finite-momentum SC. Finally one sees that the presence of the external magnetic field enhances the superfluid density and it extends the region with helical magnetization. This indicates that the increased stability of SC under parallel B might be related to an exotic finite-momentum superconducting state.

Blogged by Marco Grilli

Thursday 30 July 2015 – Shahal Ilani - Attraction by Repulsion: Pairing electrons by electrons

Shahal Ilani addresses a highly fundamental question in a very novel and elegant approach. The question is : « Can we make 2 electrons attract only via repulsive interaction ? ».

Of course this problem is particularly relevant in the framework of superconductors and in the search for room-temperature superconductivity, since the lighter is glue, the stronger is the coupling to be expected. The approach is here to specifically design a minimal building block to obtain attraction between 2 electrons.

Starting from ideas proposed by Little where a 1D conducting molecule is coupled to a specific medium acting as an electronic polarizer and having negative dielectric constant, Ilana and his group have designed a specific system made of  a very clean carbon nanotube (CNT) containing two quantum wells and electrically connected to a measurement setup.  The electronic polarizer is also a carbon nanotube with two quantum wells. When the electronic polarizer (EP) is sufficiently far away, measurement of the voltages in the CNT shows that the electrons in the two quantum wells repulse each other, as expected, whereas when the EP is close to the CNT, the electrons are shown to attract each other i.e. there is an observable quantum state resulting of superposition of two simultaneously occupied and vacant quantum wells.  The mechanism is that when the electronic polarizer is approached to the CNT, the presence of an electron in one of the quantum wells of the CNT will provoke the hopping of the electron that is in the closest quantum well of the EP to the other (further) quantum well, therefore creating an electric field of opposite sign. In order to share the cost for the electrical field that is created, two electrons will have a tendency to occupy/disoccupy simultaneously the nearest quantum wells on the CNT.

The authors have inspected the detuning dependence of the phenomenon and are currently exploring transport in these devices.

Blogged by Brigitte Leridon