Absence of skewness in the voltage fluctuations of a tunnel junction in the quantum regime
Clovis Farley, Bertrand Reulet · 2025
Read Absence of skewness in the voltage fluctuations of a tunnel junction in the quantum regime by Clovis Farley, Bertrand Reulet (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-15352.txt.
Introduction
Clovis Farley and Bertrand Reulet study small electric noise in a tunnel junction. Such a junction lets electrons cross a thin barrier, and its current flickers. At low voltage, part of that flicker looks like vacuum noise, which is the faint motion left even in empty space. They ask whether this vacuum noise can be uneven, or skewed, as well.
The pair work out the math for a junction in any electric setting at zero temperature. They show the third moment of the voltage noise is always zero above a set frequency. Tests on real junctions match the result. The paper settles a basic question about how quantum devices behave.
Catalog record
| CREATOR | Clovis Farley, Bertrand Reulet |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 15 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-15352
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Absence of skewness in the voltage fluctuations of a tunnel junction in the quantum regime Clovis Farley, Bertrand Reulet Abstract ICNF-2025 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx June 17-20, 2025 – Taormina, Italy Abstract: Current fluctuations in a tunnel junction have a remarkable property: On the one hand, their variance corresponds to vacuum fluctuations at low voltage bias V , when the electron energy eV is smaller than the photon detection energy hf . On the other hand, their skewness, i.e. their third moment, is frequency independent, equal to e^{2}I as if electron transport were simply Poissonian. We address the following question: Could it be that at low voltage, the vacuum fluctuations generated by the junction have a finite skewness, i.e. that the junction generates skewed vacuum ? To answer this question we calculate the effect of an arbitrary electromagnetic environment at zero temperature and show that the bispectrum of third moment of voltage fluctuations of any quantum conductor is always zero at frequencies larger than the voltage. We also show experimental data on tunnel junctions in the quantum regime that agree with our calculation. Keywords: Quantum Noise, electron transport, skewness, electromagnetic environment, high frequency noise Introduction. The third moment of current/voltage fluctuations in an electrical device is known to be dependent on the electromagnetic environment of the sample, i.e. the circuit to which it is connected [ 1 , 2 , 3 ] . This occurs as soon as the variance of current/voltage fluctuations depends on the bias voltage of the device, i.e. for all components but a linear resistor. This comes from two factors: i) external voltage fluctuations modulate the noise of the device, and ii) current fluctuations generated by the device generate voltage fluctuations on the device itself through the external impedance to which the device is connected. These contributions add to the intrinsic skewness of the component. A general problem of quantum noise of an electronic conductor is to understand how electronic fluctuations imprint their statistical properties in the electromagnetic radiation they generate. In the case of the variance of fluctuations, the link between current fluctuations and electromagnetic power detected at frequency f is well understood and depends on the detection scheme (photon detector, voltage detector, etc.) [ 4 ] . In the case of a classical detector, the averaged square voltage measured at frequency f is proportional to n(f)+\frac{1}{2} with n(f) the number of photons per second emitted in the detection bandwidth. The 1/2 background corresponds to vacuum fluctuations. If electrons are given an energy eV<hf they cannot emit photons so n(f)=0 and the detected noise is that of vacuum fluctuations. Here we want to probe what happens to the bispectrum \langle v(f_{1})v(f_{2})v(f_{3})\rangle in the limit where all frequencies f_{i} are greater than eV/h , i.e. in the full quantum regime where the variance of the noise is that of vacuum at all frequencies f_{1} , f_{2} and f_{3} . Could it be that the vacuum generated by a quantum conductor at low bias is skewed ? Or is it that the environmental contributions always conspire to make the skewness vanish in this regime ? Theory. We consider the general case depicted in Fig. 1: a quantum conductor of resistance R is connected to a perfect amplifier via an arbitrary scattering matrix S . The perfect amplifier is made of a circulator with a load Z_{0} at zero temperature, followed by a real amplifier. With such a scheme, the fluctuations generated by both the conductor and the vacuum fluctuations generated by the circulator load are partially reflected by the environment and partially transmitted to the amplifier. All parasitic capacitances, inductances, delay lines, etc. are incorporated in the S -matrix of the environment. The S -matrix is chosen with port 1 (the sample) having an impedance R and port 2 (the amplifier) an impedance Z_{0} . S_{ij} represents the scattering amplitude from port j to port i . Figure 1 : Detection scheme. Green arrows represent the vacuum fluctuations generated by the load Z_{0} . Red arrows represent the noise generated by the sample at frequencies f_{1} and f_{2} . According to the circuit of Fig. 1 , the voltage detected by the amplifier is given by: v(f)=\frac{1}{2}S_{22}(f)Z_{0}i_{e}(f)+\frac{1}{2}S_{21}(f)Ri_{s}(f) (1) Here i_{e} and i_{s} represent the current noise generated by the load and the sample respectively. The voltage fluctuations across the sample are given by: v_{s}(f)=\frac{1}{2}S_{12}(f)Z_{0}i_{e}(f)+\frac{1}{2}\big(1+S_{11}(f)\big)Ri_{s}(f) (2) Note that the reflection coefficient \Gamma=(R-Z_{0})/(R+Z_{0}) on the sample from an impedance Z_{0} does not appear thanks to the choice of the port 1 of the matrix S being R , not Z_{0} . We start by calculating the power per unit bandwidth detected at frequency f : P_{0}=\frac{1}{Z_{0}}\langle|v(f)|^{2}\rangle=\frac{1}{4}|S_{22}(f)|^{2}Z_{0}N_{e}(f)+\frac{1}{4}|S_{21}(f)|^{2}\frac{R^{2}}{Z_{0}}N_{s}(f) (3) Here N_{e}(f)=\langle i_{e}(f)i_{e}(-f)\rangle and N_{s}(f)=\langle i_{s}(f)i_{s}(-f)\rangle are respectively the spectral densities of the current noise generated by the load and the sample. At T=0 , the load generates vacuum fluctuations, N_{e}(f)=h|f|/Z_{0} . For the sample, we first focus on a tunnel junction, which when biased at voltage V generates current fluctuations of spectral density: N_{s}(f)=\frac{|eV-hf|+|eV+hf|}{2R} (4) For e|V|<h|f| , N_{s}(f)=h|f|/R (this is valid for any sample, not just the tunnel junction). In this limit, one has: P_{0}=\frac{hf}{4}\Big(|S_{22}(f)|^{2}+|S_{21}(f)|^{2}\frac{R}{Z_{0}}\Big) (5) Thanks to the unitarity of the S matrix, the quantity within parenthesis is 1, so P_{0} corresponds to vacuum fluctuations, part of which coming from the sample, the rest from the load. The factor R/Z_{0} comes from the choice of the impedance of the ports of the scattering matrix S [ 5 ] . We now consider the skewness \langle v(f_{1})v(f_{2})v(f_{3})\rangle . Since the frequencies must obey f_{1}+f_{2}+f_{3}=0 in the absence of ac excitation, we take 0<f_{1}\leq f_{2}<-f_{3}=f_{1}+f_{2} . The current fluctuations of the sample and the load are uncorrelated, however the noise of the sample can be correlated to the voltage across itself, so correlators involved in the skewness are: \langle v(f_{1})v(f_{2})v(f_{3})\rangle=\langle j_{s,1}j_{s,2}j_{s,3}\rangle+\langle j_{s,1}j_{s,2}j_{e,3}\rangle+\langle j_{s,1}j_{e,2}j_{s,3}\rangle+\langle j_{e,1}j_{s,2}j_{s,3}\rangle (6) with j_{s,k}=(R/2)S_{21}(f_{k})i_{s}(f_{k}) and j_{e,k}=(Z_{0}/2)S_{22}(f_{k})i_{e}(f_{k}) . Terms of the form \langle j_{s}^{2}j_{e}\rangle correspond to the modulation of the sample noise by the external vacuum fluctuations and are calculated using: \langle i_{s}(f_{i})i_{s}(f_{j})i_{e}(f_{k})\rangle=\frac{d\langle i_{s}(f_{i})i_{s}(f_{j})\rangle}{dv_{s}(-f_{k})}\langle i_{e}(f_{k})v_{s}(-f_{k})\rangle (7) with \langle i_{e}(f)v_{s}(-f)\rangle=\frac{1}{2}S_{12}(-f)Z_{0}N_{e}(f) . The quantity: \chi_{ijk}=\frac{d\langle i_{s}(f_{i})i_{s}(f_{j})\rangle}{dv_{s}(-f_{k})} (8) is the noise susceptibility, which measures the linear response of the noise to a voltage excitation across the sample at frequency -f_{k} . It has been studied theoretically and experimentally [ 6 , 7 , 8 , 9 , 10 ] . At zero temperatures, it reduces, for a tunnel junction, to: \chi_{ijk}=\frac{e}{2h|f_{k}|R}[|hf_{i}+eV|-|hf_{i}-eV|+|hf_{j}+eV|-|hf_{j}-eV|] (9) In the low voltage limit e|V|<hf_{1} , \chi_{231}=\chi_{312}=0 (10) i.e., excitations at frequencies f_{1} or f_{2} do not modulate the noise at frequency f_{3} . \chi_{123}=\frac{2e^{2}I}{h|f_{3}|} (11) Thus the contribution of the environmental noise on the skewness bispectrum \langle v(f_{1})v(f_{2})v(f_{3})\rangle at low voltage bias is: \frac{R^{2}Z_{0}}{8}S_{21}(f_{1})S_{21}(f_{2})S_{22}(f_{3})S_{12}(-f_{3})e^{2}I (12) This contribution stems from voltage fluctuations across the sample due to external noise. The noise of the sample itself contributes also to these fluctuations, leading to the feedback term, which is similar with \langle i_{s}(f)v_{s}(-f)\rangle=\frac{1}{2}\big(1+S_{11}(-f)\big)RN_{s}(f) , see Eq.( 2 ). These terms contribute to the skewness as: \frac{R^{3}}{8}S_{21}(f_{1})S_{21}(f_{2})S_{21}(f_{3})\big(1+S_{11}(-f_{3})\big)e^{2}I (13) Finally the intrinsic electronic skewness contributes to the same quantity by: -\frac{R^{3}}{8}S_{21}(f_{1})S_{21}(f_{2})S_{21}(f_{3})e^{2}I (14) Adding the three contributions leads to the prefactor: S_{21}(f_{3})S_{11}(-f_{3})+S_{22}(f_{3})S_{12}(-f_{3})\frac{Z_{0}}{R}=0 (15) The unitarity of the S matrix ensures that this vanishes, taking into account the