We report on a comparison of four GaAs/AlGaAs-based quantum resistance
standards using an original technique adapted from the well-known Wheatstone
bridge. This work shows that the quantized Hall resistance at Landau level
filling factor $\nu=2$ can be reproducible with a relative uncertainty of
$32\times 10^{-12}$ in the dissipationless limit of the quantum Hall effect
regime. In the presence of a very small dissipation characterized by a mean
macroscopic longitudinal resistivity $\bar{R_{xx}(B)}$ of a few $\mu\Omega$,
the discrepancy $\Delta R_{\mathrm{H}}(B)$ measured on the Hall plateau between
quantum Hall resistors turns out to follow the so-called resistivity rule
$\bar{R_{xx}(B)}=\alpha B\times d(\Delta R_{\mathrm{H}}(B))/dB$. While the
dissipation increases with the measurement current value, the coefficient
$\alpha$ stays constant in the range investigated ($40-120 \mathrm{\mu A}$).
This result enlightens the impact of the dissipation emergence in the
two-dimensional electron gas on the Hall resistance quantization, which is of
major interest for the resistance metrology. The quantum Hall effect is used to
realize a universal resistance standard only linked to the electron charge
\emph{e} and the Planck's constant \emph{h} and it is known to play a central
role in the upcoming revised \emph{Syst\`eme International} of units. There are
therefore fundamental and practical benefits in testing the reproducibility
property of the quantum Hall effect with better and better accuracy.Comment: 6 pages, 6 figure