14.09.2014 Views

pdf, 9 MiB - Infoscience - EPFL

pdf, 9 MiB - Infoscience - EPFL

pdf, 9 MiB - Infoscience - EPFL

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

176 CHAPTER 7. CONCLUSION<br />

is the strong Coulomb repulsion V between nearest-neighbors.<br />

In order to analyze how the magnetism and the superconductivity depend<br />

on the geometry of the compound, we have extended further in chapter 4 the<br />

variational study to correlated electrons on the honeycomb lattice, which could<br />

give a good description of graphene single sheets. At half-filling, we have found<br />

a ground state mixing at the same time antiferromagnetism and superconducting<br />

pairing. The staggered magnetization is 66% of the classical value, which is<br />

slightly higher than the 50% obtained with exact Quantum Monte Carlo. However,<br />

the energy obtained is very close to the exact value: we find a variational<br />

Heisenberg energy per site, extrapolated to the thermodynamic limit, that is only<br />

0.3% higher than the quantum Monte Carlo result. A coexistence phase between<br />

the two order parameters is found in the range x =[0, 0.07], and superconductivity<br />

is suppressed at the van Hove singularity x =1/8. Therefore the range of the<br />

superconducting order is δ =]0, 1 [. The amplitude and the range of existence of<br />

8<br />

the superconducting parameter is four times smaller than in the square lattice.<br />

We find good agreement between the VMC calculations and an RVB MF theory<br />

in the superconducting phase, namely the same d x 2 −y 2 + id xy symmetry and a<br />

similar amplitude of the pairing order parameter. For hole doping larger than<br />

1/8, we find that a spin density wave (SDW), with pitch vector equal to one of the<br />

three possible nesting vectors, is stabilized in the range δ =[ 1 , 0.22]. The SDW<br />

8<br />

phase leads to an optimization of the kinetic energy. However, a stronger gain in<br />

kinetic energy, and also a lower variational energy, is obtained at δ =0.22 with a<br />

weak ferromagnetic polarized phase, which is polarized linearly and reaches full<br />

polarization at doping δ =0.5. Ferromagnetism disappears again at δ =0.6.<br />

We performed also measurements on a number of single sheet wrapped nanotubes,<br />

in order to investigate the limit of the quasi-1D system with variational<br />

Monte Carlo. We observe that not only the amplitude of the superconductivity,<br />

but the phase of the pairing on each nearest-neighbor link depend on the wrapping<br />

of the tube. We have measured the phase after projection of the BCS pairing<br />

in the different tubes, and we observe that the phases of the pairing observable<br />

moves from the d x 2 −y 2 + id xy symmetry in the case of the 2 dimensional lattice<br />

towards intermediate value, and converge to the d-wave symmetry in the case of<br />

the 2-leg ladder, which is also the smallest nanotube that can be wrapped with<br />

a 2-site unit-cell. We found a suppression of the magnetism when the diameter<br />

is small and reaches the limit of the 2-leg ladder, and an enhancement of the<br />

pairing order parameter in the same limit. This might be interpreted as the signature<br />

that quantum fluctuations become much larger when the tube reaches the<br />

one dimensional limit, and our variational magnetic order parameter is no longer<br />

stabilized when these fluctuations become too strong. At the same time, it is<br />

interesting to note that the pairing survives very well in the 1D limit, though<br />

we do not expect any real pairing in a quasi-1D model. Indeed, in this limit the<br />

ground state will be a Luttinger liquid.<br />

Finally, we considered the square lattice geometry, and extended the pre-

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!