In a new international system of units, fundamental constants are to play the decisive role of ubiquitous reference measurands. Voltage and resistance can be traced back to the Josephson and the von Klitzing constant, respectively, whose product is, according to theoretical predictions, equal to 2/e, with –e being the charge of an electron. If these theories are correct, also the kilogram can be
traced back to electrical units and their corresponding fundamental constants by matching mechanical and electrical power in a so-called “watt balance”. It is therefore essential for the whole system of units for theoretical predictions to be verified experimentally.
For this purpose, the current leading to a voltage U ~ h/2e over a resistor R ~ h/e² is compared to a current I = eƒ where electrons are moved through a single-electron pump at the frequency ƒ. Unavoidable faults, caused by, e. g., quantum mechanical tunnel processes, however, limit the uncertainty to fractions of ppm. With the new method, such faults can, however, be detected, so that the uncertainty is decisively reduced: the current is successively pushed through two single-electron pumps, and the charge on the island located in between is monitored. In error-free operation, the charge remains constant; in the event of a fault, it varies by one electron charge. As faults rarely occur, they can be detected even at high pump frequencies of up to several 100 MHz. Even if three pumps are used, each fault can be attributed to a certain pump thanks to correlation measurements. The current generated can thus be determined much more accurately than would be the case using only one pump.
This experiment has now been realized at PTB for the first time. Single electrons were moved through an array consisting of three active pumps (P) with two detectors (D). Both for the “red” and for the “blue” detector depicted in the figure, each electron staying on its respective island triggers a pulse which does not change the signal baseline. In the event of a fault, the baseline, however, changes, which remains detectable also at higher frequencies when pulses of single electrons are no longer resolvable.
In future, it should be possible to increase the current from the attoampere range as demonstrated here at low frequencies to the application-relevant range of several 100 pA.
michael.wulf