Elementary charge-transfer processes in mesoscopic conductors

Abstract

We determine charge-transfer statistics in a quantum conductor driven by a time-dependent voltage and identify the elementary transport processes. At zero temperature unidirectional and bidirectional single-charge transfers occur. The unidirectional processes involve electrons injected from the source terminal due to excess dc bias voltage. The bidirectional processes involve electron-hole pairs created by time-dependent voltage bias. This interpretation is further supported by the charge-transfer statistics in a multiterminal beam-splitter geometry in which injected electrons and holes can be partitioned into different outgoing terminals. The probabilities of elementary processes can be probed by noise measurements: the unidirectional processes set the dc noise level, while bidirectional ones give rise to the excess noise. For ac voltage drive, the noise oscillates with increasing the driving amplitude. The decomposition of the noise into the contributions of elementary processes reveals the origin of these oscillations: the number of electron-hole pairs generated per cycle increases with increasing the amplitude. The decomposition of the noise into elementary processes is studied for different time-dependent voltages. The method we use is also suitable for systematic calculation of higher-order current correlators at finite temperature. We obtain current noise power and the third cumulant in the presence of time-dependent voltage drive. The charge-transfer statistics at finite temperature can be interpreted in terms of multiple-charge transfers with probabilities which depend on energy and temperature.

Integer and fractional charge Lorentzian voltage pulses analyzed in the framework of photon-assisted shot noise

“…The Anderson physics reflects in a marked minimum of neutral excitations for integer charge per period as theoretically observed in Ref. [28].…”

“…We compute the PASN as a function of an arbitrary dc voltage bias which provides a direct measure of the photoabsorption probabilities [28,41,45,48,54]. In particular the Lorentzian pulses show an asymmetric PASN versus dc bias characteristic of the absence of hole type excitations while sine and square waves lead to symmetric PASN.…”

“…The remarkable result of ref. [14,15] has triggered several relevant theoretical contributions [9,17,[27][28][29][30] in which the property and potential use of the Lorentzian voltage pulses are discussed. An experimental implementation has been recently done [31].…”

“…Using the separation of the voltage pulse into its mean value, or d.c. part < V (t) >= V dc and its ac component V ac (t) = V (t) − V dc as in ref. [28], the physics of voltage pulses can be conveniently discussed in the framework of Photo-Assisted Shot Noise (PASN) where both theoretical [41][42][43] and experimental results [45,48] are already available. At zero temperature, the excess noise ∆S I characterizing the extra excitations is then the difference between the PASN noise S P ASN I due to V (t) = V ac + V dc and the (transport) shot noise (TSN) S T SN I that we would have with only the dc voltage V (t) = V dc [35-40, 49, 50].…”

“…[32]. While Lorentzian shape pulses with n charges per period correspond to exactly n excitations [14,15], pulses of arbitrary shape contain more excitations [28] whose number is calculated for sine, square and rectangular wave shape. We also address the case of non-integer charge pulses which was shown in ref.…”