John Mydosh: Nernst Effect in NdBa2(Cu_1-y Ni_y)3 O_7-d

In a very inspiring talk John Mydosh presented us recent experimental results about the Nernst effect in the compound NdBa_2(Cu_(1-y)Ni_y)_3 O_{7-x}. This work was published recently in Phys. Rev. B 76, 020512(R) (2007). John started by saying that in those compounds it’s very difficult to get bulk sample, which is what his student had. The system has the same anisotropy as YBCO
and is at optimally doped for x=0. John had drawn a phase diagram for this compound, which looks very similar to the one of YBCO, with a maximum Tc of 85 K, an AF phase at low doping and a ``pseudo-gap’’ phase in between. The interest of doping with Ni is that it reduces Tc without changing the amount of oxygen doping.

Then John embarked on a very nice pedagogical review on Nernst and Seebeck effects in metals. He recalled that, for one-band metals, the Sondheimer cancellation is the cause that Nersnst effect is very small, almost undetectable. This is not true, however for semi-metals, like Bi, which have some energy gaps. Ilya Vekhter asked whether this property was true for all semi-metals. John answered that generically, if you have partial gaps in the Fermi surface the Nernst effect will be big. Andrey Chubukov asked what `` big’’ meant within these units? the answer was that already a few micro Volts per Kelvin can be considered as ``big’’.
At this point, John recalled a review of the compounds for which anomalous Nernst effect has been observed with putative explanation. I reproduce it below:

CDW-NbSe_2: counterflow of electrons and holes
PrFe_4 P_12: quadrupolar ordering
URu_2Si_2: ``hidden order’’ transition
(B_12)-(BEDT-TTF)_2 : SC near a Mott transition
CeCoIn_5: co-existence of SC and AF
MgB_2 : ?

Then John recalled that for a superconductor, a non zero Nernst effect is expected, du e to the presence of the vortices. The motion of vortices generates some phase slip and creates a perpendicular response to a temperature gradient (E_y = - e_y \grad T, with e_y the Nernst coefficient). In a superconductor, one can even see the pinning force of the vortices, which reduces very strongly the Nernst signal at very low fields. This vanishing of e_y at low fields can be considered as test mark that we are in a superconducting phase.

John reviewed the results obtained by Ong for LASCO and Nd_(2-x)Ce_xCuO_4 (electron doped). He emphasized the striking difference in the Nernst signal between the eletron and hole doped compounds, with a much stronger Nernst effect in the hole doped case. He stressed that the common view by the Ong group is that the Nernst effect is due to some pre-formed pairs and that the theory of Anderson seems to work well.

Back to his own data on NdBa_2(Cu_(1-y)Ni_y)_3 O_{7-x}, John compares three dopings O_7, O_{6.9}and O_{6.8}. One can see on the curves, that generically the Nernst signal is present in the normal phase at optimal doping, in the under-doped regime it grows to then vanish for the very under-doped case. By then playing with the doping in Ni (which semsibly reduces T_c) John compared the variation of the Nernts signal wth T_c for the three O-dopings mentioned above. For the O_7 case, the temperature below which the Nernst effect is present- let’s call it T_N- follow T_c (namely T_N= T_c + 20 K). But when one under -dopes the behavior of T_N changes and becomes independent of T_c: it’s more or less stationary. It is important to note that T_N doesn’t follow the pseudo gap temperature T^* as well, which is shown to increase with Ni doping.

John concludes that the behavior of T_N remains mysterious, but there is indication that this temperature is very sensitive to the presence of impurities. He cites the work of Alloul and Albenque using the van de Graaf in Orsay, where due to irradiation of the sample one produces intrinsic disorder. This hand-made disorder creates a Nernst signal.


Discussion followed.

Zlatko Tesanovic asked what was the difference with the work of Ong, and in particular whether there is a vortex liquid phase in this compound. John answered that the beauty of his experiment is that the Ni doping enables to reduce T_c without changing other parameters ( in particular the oxygen doping). This is due to the fact that Ni has spin 1.

Claudio Castellani commented that the Nernst effect should in general depend on the superfluid density. John answered that it is the case except if there are inhomogeneous regions, which could well be the case here.