Unified derivation of Johnson and shot noise expressions Luca Callegaro Istituto Nazionale di Ricerca Metrologica (INRIM), Strada delle Cacce, 91-10135 Torino, Italy 共Received 31 October 2005; accepted 20 January 2006兲 Shot noise and Johnson noise in electrical circuits are usually introduced by referring to completely separate physical models and derivations. We derive Johnson and shot noise expressions from the same physical model, an ideal tunnel junction, to show the deep connection between the two types of noise. The derivation uses concepts of quantum mechanics, thermodynamics, and signal processing. © 2006 American Association of Physics Teachers. 关DOI: 10.1119/1.2174034兴 I. INTRODUCTION The sources of noise observed in electrical circuits are usually attributed to Johnson noise and shot noise. Johnson noise 共from the names of those who did the first exhaustive experimental observations1 and theoretical explanation2 of the phenomenon兲 is the random voltage 共or current兲 caused by thermal fluctuations of the electric charge in conductors. See Ref. 3 for a pedagogical review. Shot noise is caused by the discreteness 共quantization兲 of charge carriers and can be observed when a detector measures a current. Shot noise was first described by Schottky.4,5 These disparate descriptions of Johnson noise and shot noise 共see Refs. 6 and 7 for a typical introduction兲 suggest separate explanations and models of the two noise processes. The derived quantitative expressions are independent and the corresponding noise power densities are additive. This independence has been questioned for a long time and in the last few decades the treatment of electrical conductance as a quantum-mechanical transmission phenomenon has shown that both types of noise have the same explanation8 and unified descriptions are now commonplace 共see Refs. 9 and 10 for reviews兲. A unified picture has not yet appeared in general physics and engineering courses, possibly because of the difficulties involved in a rigorous derivation. The aim of this paper is to show that if rigor is not an issue, expressions for the thermal and shot noise can be derived from the same physical model. The derivation invokes several assumptions from various branches of physics and signal analysis and is a good review problem for physics or electronic courses. II. THE MODEL Consider an ideal model of a physics device, the tunnel junction. Figure 1 shows a tunnel junction J with a small insulating gap g 共vacuum, dielectric兲 dividing two metal contacts A and B connected to an ideal voltage generator G with constant voltage V. The current carriers have an elementary charge q; for convenience, we will call them electrons. We let I共t兲 equal the current flowing in the electrical circuit. The system is isothermal at temperature T. Because electrons in a circuit are quantum-mechanical particles, there are finite probabilities per unit time PAB and PBA that an electron crosses by tunneling through the junction in the A → B or A ← B direction, respectively. The charge transfer is considered instantaneous. If we know the nature of the junction, quantum mechanics10 lets us calculate the actual values of PAB and PBA. However, we will assume phe438 Am. J. Phys. 74 共5兲, May 2006 http://aapt.org/ajp nomenological values for PAB and PBA dependent on the junction and its environment, in particular V and T. We assume that current is not observed continuously, but in discrete chunks of duration ; for each chunk only its average value Ī = 兰 I共t兲dt is recorded. We can think of this observation as given by an integrating analog-to-digital converter with a sampling frequency f s = 1 / . From Nyquist’s sampling theorem,11 the equivalent bandwidth B for the observation is half the sampling frequency, B= fs 1 = . 2 2 共1兲 We assume that the sampling time is sufficiently rapid that only the following events can be observed. 共a兲 Ī = + q / . An electron has crossed the junction in the A → B direction. Such an event has a probability PAB. 共b兲 Ī = −q / . An electron has crossed the junction in the A ← B direction. Such an event has a probability PBA. 共c兲 Ī = 0. No electron has crossed the tunnel barrier. Such event has a probability 1 − 共PAB + PBA兲. Thus, at most only one electron passes the barrier in either direction. III. CURRENT The passage of an electron is a probabilistic event and therefore Ī is a random variable. The probability distribution p¯I of Ī is discrete 共only three realizations of Ī are allowed兲. Because the probability of each realization is given, p¯I is determined and all moments can be calculated. The average 具I典 is given by 1 具I典 = 兺 Īp¯I k=−1 = 冉 冊 冉 冊 −q +q 关PBA兴 + 共0兲关1 − 共PAB + PBA兲兴 + 关PAB兴 = q共PAB − PBA兲. 共2兲 The mean square value of the current is 具I2典 = 冉 冊 −q 2 PBA + 共0兲 + 冉 冊 +q 2 PAB = q2 共P + PBA兲. 共3兲 AB Therefore, knowledge of PAB and PBA permits us to derive the average current and current noise power. © 2006 American Association of Physics Teachers 438 For T = 0 and V ⫽ 0, Eq. 共10兲 gives the pure shot noise expression, 兩具I2典兩T=0 = 2q具I典B. 共11兲 For T ⬎ 0 and V → 0, Eq. 共10兲 gives the Johnson noise expression, lim 具I2典 = V→0 4kT B. R 共12兲 Fig. 1. Schematic of the model. J is a tunnel junction where contacts A and B are separated by a gap g, connected to a generator G that supplies a constant voltage V. The current I共t兲 flows in the junction and T is the temperature of the circuit. VI. COMMENTS IV. TRANSITION PROBABILITIES AND MACROSCOPIC PHYSICS The average current 具I典 can be written in terms of V and the macroscopic resistance R of the junction, 具I典 = V . R 共4兲 We can think of the tunneling event as a transition between two states: electron in A or electron in B. The states A and B have occupation numbers nA and nB. The average number of transition events per unit time in the A → B direction is PABnA; such events decrease nA and increase nB. In the same unit time, the number of transition events in the A ← B direction is PBAnB. Because the system is in a steady state, that is, on the average nA and nB are constant, the detailed balance condition is PABnA = PBAnB . 共5兲 The states A and B have energies EA and EB; the energy difference is EA − EB = qV. Their occupation numbers satisfy the Boltzmann distribution and, hence, nA = e−qV/kT . nB We have shown that the same model, the tunnel junction, gives an expression for the noise whose limits for V = 0 and T = 0 correspond to Johnson and shot noise, respectively. If V ⫽ 0 and T ⬎ 0, Eq. 共10兲 shows that Johnson noise and shot noise are not additive even for the ideal model we have considered. The main result, Eq. 共10兲, is essentially correct, although its derivation is far from rigorous.12 A review with a full quantum-mechanical derivation can be found in Ref. 10. Nyquist has used a simple thermodynamical argument to show that Eq. 共12兲 is valid for any electric dipole in thermodynamic equilibrium.2 Equation 共12兲 has been generalized beyond electric circuits by the fluctuation-dissipation theorem.13 Instead, the shot noise expression Eq. 共11兲 共giving the Poissonian shot noise value兲 is strongly related to our assumptions: mainly that charge carriers must be localized and independent particles. Typically 共although not strictly necessarily14兲 localization occurs because of a potential barrier. Correlation among charge carriers can generate subPoissonian or super-Poissonian shot noises 共see, for example, Ref. 15兲. 共6兲 ACKNOWLEDGMENTS V. NOISE If we combine Eqs. 共5兲 and 共6兲, we obtain the relation between PAB and PBA, PBA = e−qV/kT , PAB 1 共7兲 and, hence, 具I典 = V = q共PAB − PBA兲 = qPAB关1 − e−qV/kT兴, R 共8兲 q2 q2 共PAB + PBA兲 = PAB关1 + e−qV/kT兴. 共9兲 具I2典 = If we substitute Eq. 共1兲 into 共9兲, we obtain 具I2典 = 2q 439 V 共1 + e−qV/kT兲 B. R 共1 − e−qV/kT兲 Am. J. Phys., Vol. 74, No. 5, May 2006 The author warmly thanks W. Bich, G. Mana, G. Brida, and M. Pisani for fruitful discussions and for reviewing the manuscript. 共10兲 J. B. Johnson, “Thermal agitation of electricity in conductors,” Phys. Rev. 32, 97–109 共1928兲. 2 H. Nyquist, “Thermal agitation of electric charge in conductors,” Phys. Rev. 32, 110–113 共1928兲. 3 B. Abbott, D. R. Davis, N. J. Phillips, and K. Eshraghian, “Simple derivation of the thermal noise formula using window-limited Fourier transforms and other conundrums,” IEEE Trans. Educ. 39, 1–13 共1996兲. 4 W. Schottky, “Über spontane stromschwankungen in verschiedenen elektrizittsleitern,” Ann. Phys. 23, 541–567 共1918兲. 5 W. Schottky, “Small-shot effect and flicker effect,” Phys. Rev. 28, 74– 103 共1926兲. 6 C. Kittel, Elementary Statistical Physics 共Dover, New York, 1967兲, Chap. 29, pp. 141–146. 7 S. Engelberg and Y. Bendelac, “Measurement of physical constants using noise,” IEEE Instrum. Meas. Mag. 6, 49–52 共2003兲. 8 R. Landauer, “Solid-state shot noise,” Phys. Rev. B 47, 16427–16432 共1993兲. Luca Callegaro 439 9 Y. Imry and R. Landauer, “Conductance viewed as transmission,” Rev. Mod. Phys. 71, S306–S312 共1999兲. 10 Y. M. Blanter and M. Büttiker, “Shot noise in mesoscopic conductors,” Phys. Rep. 336, 1–166 共2000兲. 11 H. Nyquist, “Certain factors affecting telegraph speed,” Bell Syst. Tech. J. 32, 324–352 共1924兲. 12 Some of the simplifications are the following: The sampling procedure assumes only three possible events; the noise power is identified with the noncentral moment 具I2典 instead of the variance and hence includes the dc power; the electron populations obey a Boltzmann instead of a Fermi distribution; more generally, thermodynamic equilibrium distributions and detailed balance are assumed for a steady-state dissipative system 共see Refs. 16 and 17兲. These simplifications nevertheless give a correct 440 Am. J. Phys., Vol. 74, No. 5, May 2006 result in the low-frequency limit, that is, for f ⬍ 100 GHz at room temperature 共Ref. 3兲. 13 H. B. Callen and T. A. Welton, “Irreversibility and generalized noise,” Phys. Rev. 83, 34–40 共1951兲. 14 G. Gomila, C. Pennetta, L. Reggiani, M. Sampietro, G. Ferrari, and G. Bertuccio, “Shot noise in linear macroscopic resistors,” Phys. Rev. Lett. 92, 226601-1–226601-3 共2004兲. 15 Y. M. Blanter and M. Büttiker, “Transition from sub-Poissonian to superPoissonian shot noise in resonant quantum wells,” Phys. Rev. B 59, 10217–10226 共1999兲. 16 M. J. Klein, “Principle of detailed balance,” Phys. Rev. 97, 1446–1447 共1955兲. 17 R. M. L. Evans, “Detailed balance has a counterpart in non-equilibrium steady states,” J. Phys. A 38, 293–313 共2005兲. Luca Callegaro 440