Microscopic Origin of the 0.7-Anomaly in Quantum Point Contacts
<br>
<br>
Quantum point contacts, elementary building blocks of
semiconductor-based quantum circuits, are narrow one-dimensional
constrictions usually patterned in a two-dimensional electron system,
e.g. by applying voltages to local gates. It is one of the paradigms of
mesoscopic physics that the linear conductance of a point contact, when
measured as function of its channel width, is quantized in units of $\GQ
= 2e2/h$. However, the conductance also exhibits an unexpected shoulder
at $\simeq 0.7 \GQ$, known as the ``0.7-anomaly', whose origin is still
subject to debate. Proposed theoretical explanations have evoked
spontaneous spin polarization, ferromagnetic spin coupling, the
formation of a quasi-bound state leading to the Kondo effect, Wigner
crystallisation and various treatments of inelastic scattering. However,
explicit calculations that fully reproduce the various experimental
observations in the regime of the 0.7-anomaly, including the zero-bias
peak that typically accompanies it, are still lacking. Here we offer a
detailed microscopic explanation for both the 0.7-anomaly and the
zero-bias peak: their common origin is a smeared van Hove singularity in
the local density of states at the bottom of the lowest one-dimensional
subband of the point contact, which causes an anomalous enhancement in
the Hartree potential barrier, magnetic spin susceptibility and
inelastic scattering rate. We present theoretical calculations and
experimental results that show good qualitative agreement for the
dependence of the conductance on gate voltage, magnetic field,
temperature, source-drain voltage (including the zero-bias peak) and
interaction strength. We also clarify how the low-energy scale governing
the 0.7-anomaly depends on gate voltage and interactions. For low
energies we predict and observe Fermi-liquid behaviour similar to that
known for the Kondo effect in quantum dots. At high energies, however,
the similarities between 0.7-anomaly and Kondo effect cease.