1: %\documentclass[prb,twocolumn]{revtex4}
2: \documentclass[prl,twocolumn]{revtex4}
3: \usepackage{amssymb}
4: \usepackage{graphicx}
5: \usepackage[latin1]{inputenc}
6: %\newcommand{\beq}{\begin{equation}}
7: %\newcommand{\eeq}}{\end{equation}}
8: \begin{document}
9: \title{Electromotive force and internal resistance of an electron pump}
10: \author{Miguel Rey}
11: \email{miguel.rey@uam.es}
12: \affiliation{Departamento de Física Teórica de la Materia Condensada and
13: Instituto Nicolás Cabrera,
14: Universidad Autónoma de Madrid, E-28049 Madrid, Spain}
15: \author{Fernando Sols}
16: \email{fernando.sols@uam.es}
17: %\altaffiliation{
18: %Universidad Autónoma de Madrid, E-28049 Madrid, Spain}
19: \affiliation {Departamento de Física Teórica de la Materia Condensada and
20: Instituto Nicolás Cabrera,
21: Universidad Autónoma de Madrid, E-28049 Madrid, Spain}
22: \date{\today}
23: \begin{abstract}
24: We present a scattering theory of the electromotive force and
25: internal resistance of an electron pump. The characterization of
26: the device performance in terms of only two parameters requires
27: the assumption of incoherent multiple scattering within the
28: circuit and complete thermalization among electrons moving in a
29: given direction. The electromotive force is shown to be of the
30: order of the driving frequency in natural units. In an open setup,
31: the electromotive force adds to the voltage difference between
32: reservoirs to drive the current, both facing a contact resistance
33: which is absent in the case of a closed circuit of uniform width.
34: \end{abstract}
35: \maketitle
36: Electrons pumps favor electron motion in a given direction by
37: combining nonlinear ac driving and some asymmetry in the spatial
38: structure or in the temporal signal. This rectification of
39: electron motion generates current in the absence of a net dc
40: voltage bias
41: \cite{swit99,thou83,staf96,pede98,brou98,wagn99,kim02,lehm03}.
42: Most theoretical calculations have dealt so far with the
43: calculation of the {\it pump current}, which is the current
44: flowing through the device when the electrons incident from both
45: sides are characterized by the same chemical potential. In
46: practice, however, one is likely to be interested in the
47: performance of the pump as a circuit component, something which
48: cannot be predicted from the mere knowledge of the pump current.
49: This creates the need to characterize the electron pump as a
50: battery with a certain {\it electromotive force} and {\it internal
51: resistance}. Although the electromotive force can in principle be
52: obtained from the dc bias that exactly cancels the pump current
53: \cite{swit99,wagn99}, its derivation within a unified and general
54: scheme seems desirable. On the other hand, there is no obvious
55: ansatz for the calculation of the internal resistance. A potential
56: application of this new class of devices is the generation of
57: current in small closed circuits not attached to broad wires
58: acting as electron reservoirs. Such a setup is schematically
59: depicted in Fig. 1. One may also consider a pump in series with a
60: resistor, both within a lead that couples to large reservoirs
61: through ideal contacts, as indicated in Fig. 2.
62: In this article, we derive formulae which express the battery
63: parameters in terms of the transmission and reflection
64: probabilities for electrons crossing the pump and the resistor.
65: The scattering theory here presented attempts to play a role for
66: the electromotive force and internal resistance of an electron
67: pump similar to that which the Landauer-B\"uttiker theory has
68: represented for the conductance of nanostructures \cite{land57}.
69: An important difference, however, is that the scenario which we
70: investigate requires a more coarse-grained description if we wish
71: to uniquely characterize the circuit performance of the pump in
72: terms of a small set of parameters. Such an effective
73: self-averaging of the device performance requires the assumption
74: of electron decoherence between the pump and the resistor in
75: series. Unlike in Ref. \cite{butt86a}, we assume that effective
76: phase randomization can be achieved with negligible
77: backscattering. This hypothesis is supported by the fact that a
78: minor distortion of the environment suffices to induce electron
79: dephasing, while a more continued interaction is needed to change
80: the electron energy or direction appreciably \cite{ster90}. We
81: find that reflectionless decoherence between the circuit elements
82: is still insufficient to permit the characterization in terms of
83: only {\it two} parameters, and not twice as many as available
84: transverse channels. To achieve a simple and manageable
85: description we must assume that, within the leads, electrons
86: moving in each direction are characterized by a single chemical
87: potential. The adequacy of this assumption, or its replacement by
88: a weaker one within a model of comparable tractability, deserves
89: further study.
90: The total current through a two-lead multimode structure in the
91: presence of local ac driving may be written
92: \begin{equation}
93: I = \frac{e}{h}\int dE_i [ f(E_i-\mu^{\rm{in}}_R)T_{LR}(E_i) -
94: f(E_i-\mu^{\rm{in}}_L)T_{RL}(E_i)]
95: \end{equation}
96: \begin{equation}
97: T_{LR}(E_i) \equiv \sum_{a \in L} \sum_{b \in R} \int dE_f
98: T^{LR}_{ab}(E_f, E_i),
99: \end{equation}
100: $T^{LR}_{ab}(E_f, E_i)$ being the probability distribution that an
101: electron incident from the right lead in channel $b$ with energy
102: $E_i$ is transmitted into channel $a$ of the left lead with energy
103: $E_f$. For future convenience, we assume $I>0$ when current flows
104: from right to left. The chemical potentials
105: $\mu^{\rm{in}}_L,\mu^{\rm{in}}_R$ characterize the population of
106: incoming electrons.
107: The pump effect is based on the existence of an asymmetry between
108: the left-to-right and right-to-left transmissions. Thus it is
109: convenient to define:
110: \begin{eqnarray} \label{average}
111: T(E_i) &\equiv& [T_{LR}(E_i) + T_{RL}(E_i)]/2 \\
112: \label{difference} \delta T(E_i) &\equiv& T_{LR}(E_i) -
113: T_{RL}(E_i)\ .
114: \end{eqnarray}
115: If we linearize $f(E-\mu^{\rm in}_{L,R})$ around a common
116: reference chemical potential $\mu_0$, we may write the total
117: current as the sum of a bias and a pump contribution
118: %\begin{equation} \label{linear}
119: %f(E-\mu^{\rm in}_{\alpha}) \simeq f(E-\mu_0) -
120: %(\mu^{\rm{in}}_{\alpha}-\mu_0)f'(E-\mu_0),
121: %\end{equation}
122: \begin{eqnarray}
123: I&=&I_B+I_P \\
124: I_B &\equiv& \frac{e}{h} \Delta\mu^{\rm in}
125: \int dE_i [-f'(E_i-\mu_0)] T(E_i) \\
126: \label{I-pump-general} I_P &\equiv& \frac{e}{h} \int f(E_i-\mu_0)
127: \delta T(E_i) \ ,
128: \end{eqnarray}
129: with $\Delta\mu^{\rm in} \equiv \mu^{\rm{in}}_R- \mu^{\rm{in}}_L$.
130: Hereafter, we take $\mu_0 \equiv 0$, although we note that (unlike
131: $I_B$) $I_P$ does depend on $\mu_0$.
132: \begin{figure}[t]
133: %\begin{center}
134: \includegraphics[width=8cm]{figure1sc.eps}
135: \caption{Schematic representation of an electron pump in series
136: with a generic resistor within a closed circuit where current
137: flows thanks to the action of the pump}
138: %\end{center}
139: \end{figure}
140: Let us focus on the current flow in a given channel $a$ on e.g.
141: lead $L$. For convenience, we define $\tilde{I}_{\alpha}\equiv
142: (h/e)I_{\alpha}$ for all future current contributions. To achieve
143: a better perspective, we momentarily abandon the assumption that
144: the chemical potential is channel independent. The total current
145: through channel $a\in L$ can then be written:
146: \begin{equation}
147: \tilde{I}_{a}= -\mu^{\rm{in}}_a + \mu^{\rm out}_a,
148: \end{equation}
149: where $\mu_a^{\rm in(out)}$ characterizes the population of
150: electrons in $L$ approaching (leaving) the pump. We note that,
151: even if the ``in'' population is rigorously thermal, the ``out''
152: population is not. However, one can always find a suitably defined
153: chemical chemical potential $\mu_a^{\rm out}$ that reproduces the
154: same current flow (and, in one dimension, the same current density
155: \cite{sols99}). Like the total current, this {\it outgoing
156: chemical potential} has a ``bias'' and a ``pump'' contribution,
157: \begin{eqnarray}
158: \mu^{\rm out}_a &=& \mu^{\rm out,B}_a+ \mu^{\rm out,P}_a \\
159: \mu^{\rm out,B}_a &=& \sum_b S_{ab}\mu^{\rm{in}}_b \\
160: \mu^{\rm out,P}_a &=& \tilde{I}_{P,a} \ ,
161: \end{eqnarray}
162: where $\tilde{I}_{P,a}$ is the pump current in channel $a$
163: ($\sum_a \tilde{I}_{P,a}=\tilde{I}_P$). Since the bias
164: contribution depends only on the symmetrized probability [see Eq.
165: (\ref{average})], we have $S_{ab}=S_{ba}$. On the other hand,
166: unitarity requires $\sum_b S_{ab} = 1$. The term $\mu_a^{\rm
167: out,P}$ accounts for the excess (or defect) of electrons generated
168: by the pump. It reflects the fact that an operating battery
169: creates a {\it population imbalance} which ultimately drives the
170: current through the circuit.
171: Assume that, in series with the pump, we introduce a resistor
172: which is also characterized in terms of its scattering
173: probabilities. The resulting circuit is schematically depicted in
174: Fig. 1. Being the resistor a passive element, its flow equations
175: do not include a pump term. We write
176: \begin{equation} \label{resistor}
177: \tilde{I}_a = m^{\rm{in}}_a - \sum_b \sigma_{ab} m^{\rm{in}}_b
178: \equiv m^{\rm{in}}_a - m^{\rm{out}}_a,
179: \end{equation}
180: where $m^{\rm{in(out)}}_a$ is the chemical potential for the
181: electrons approaching (leaving) the resistor, and
182: $\{\sigma_{ab}\}$ are the scattering probabilities by the
183: resistor, which obey $\sigma_{ab}=\sigma_{ba}$ and
184: $\sum_a\sigma_{ab}=1$. The sign convention in Eq. (\ref{resistor})
185: is different from that used in the pump equations because
186: counterclockwise current is taken to be positive (see Fig. 1). Now
187: we note that the ``out'' population of the pump is the ``in''
188: population of the resistor, and viceversa. We seal this
189: equivalence by establishing a common notation. For $a \in L$ we
190: write
191: \begin{eqnarray}
192: \mu^{\rm{in}}_a = m^{\rm{out}}_a \equiv \mu^{\uparrow}_{L,a} \\
193: \mu^{\rm{out}}_a = m^{\rm{in}}_a \equiv \mu^{\downarrow}_{L,a}\ ,
194: \end{eqnarray}
195: and similarly for $a \in R$. The vertical arrows refer to the
196: direction of movement within the convention of Fig. 1.
197: \begin{figure}[t]
198: %\begin{center}
199: \includegraphics[width=8cm]{figure2sc.eps}
200: \caption{Electron pump in series with a scattering barrier, both
201: within a multichannel wire that couples through ideal contacts to
202: large electron reservoirs. Both the action of the pump and the
203: difference between the reservoir chemical potentials contribute to
204: drive the current through the wire}
205: %\end{center}
206: \end{figure}
207: Although a solution of the flow equations that would permit us to
208: predict the total current in terms of
209: $\{I_{P,a},S_{ab},\sigma_{ab}\}$ is formally possible, our real
210: goal is the characterization of the pump in terms of two
211: parameters. To achieve this objective, we have to introduce the
212: simplifying assumption that electrons flowing in a given direction
213: within a lead are all characterized by the same chemical
214: potential. We express it as
215: \begin{equation}
216: \mu^{\uparrow \downarrow}_{L,a} = \mu^{\uparrow \downarrow}_L
217: \hspace{0.5cm} \mu^{\uparrow \downarrow}_{R,a} = \mu^{\uparrow
218: \downarrow}_R \hspace{0.5cm} \forall a \in L,R\ .
219: \label{equiv-chem-pot}
220: \end{equation}
221: Hereafter we differentiate between reflection and transmission
222: probabilities: $ S_{ab} \rightarrow R_{ab}, T_{ab}$ and
223: $\sigma_{ab} \rightarrow \rho_{ab}, \tau_{ab}$. We introduce the
224: notation
225: \begin{eqnarray} \label{RaTa}
226: R_a \equiv \sum_b R_{ab} \hspace{0.5cm} T_a \equiv \sum_b T_{ab}
227: \hspace{0.5cm} (R_a + T_a = 1)\ , \\
228: \label{RT} R \equiv \sum_a R_a
229: \hspace{0.5cm} T \equiv \sum_a T_a \hspace{0.5cm} (R + T = N)\ ,
230: \end{eqnarray}
231: $N$ being the number of transverse channels. The resistor
232: parameters $\rho_a,\tau_a,\rho,\tau$ are defined analogously. We
233: introduce an average ``pump chemical potential'' $\mu_P \equiv
234: \sum_a \mu^{\rm out, P}_{a}/N=\tilde{I}_P/N$.
235: The total current can now be written
236: \begin{equation} \label{total-I}
237: \tilde{I} = N(\mu^{\downarrow}_L - \mu^{\uparrow}_L) =
238: N(\mu^{\uparrow}_R - \mu^{\downarrow}_R)\ .
239: \end{equation}
240: These four chemical potential are not independent but are rather
241: related by the flow equations
242: \begin{eqnarray} \label{L-down}
243: N\mu^{\downarrow}_L &=& R \mu^{\uparrow}_L + T \mu^{\uparrow}_R + N \mu_P \\
244: \label{L-up} N\mu^{\uparrow}_L &=& \rho \mu^{\downarrow}_L + \tau
245: \mu^{\downarrow}_R \\
246: \label{R-down}
247: N\mu^{\downarrow}_R &=& T \mu^{\uparrow}_L + R \mu^{\uparrow}_R - N \mu_P \\
248: \label{R-up} N\mu^{\uparrow}_R &=& \tau \mu^{\downarrow}_L + \rho
249: \mu^{\downarrow}_R \ ,
250: \end{eqnarray}
251: which are physically transparent. The different sign carried by
252: the pump contribution $N\mu_{P}$ in Eqs. (\ref{L-down}) and
253: (\ref{R-down}) expresses the fact that, when $\mu_P>0$, there is
254: an excess of outgoing electrons on the left of the pump and a
255: corresponding defect of outgoing electrons on the right. When Eqs.
256: (\ref{L-down}) -- (\ref{R-up}) are introduced into Eq.
257: (\ref{total-I}), we obtain for the total current
258: \begin{eqnarray} \label{I-T-up}
259: \tilde{I} &=& -T \Delta \mu^{\uparrow} + N \mu_P \\
260: \label{I-tau-down} &=& \tau \Delta \mu^{\downarrow}\ ,
261: \end{eqnarray}
262: where the chemical potential differences $\Delta \mu^{\uparrow
263: \downarrow} \equiv \mu^{\uparrow \downarrow}_L - \mu^{\uparrow
264: \downarrow}_R$ satisfy the relations
265: \begin{eqnarray} \label{dif-down}
266: N \Delta \mu^{\downarrow} &=& (R-T) \Delta \mu^{\uparrow} + 2
267: N\mu_P
268: \\ \label{dif-up}
269: N \Delta \mu^{\uparrow} &=& (\rho-\tau) \Delta \mu^{\downarrow} \
270: ..
271: \end{eqnarray}
272: We may solve for $\Delta \mu^{\uparrow \downarrow}$ in Eqs.
273: (\ref{dif-down}) and (\ref{dif-up}) and introduce the solutions in
274: either (\ref{I-T-up}) or (\ref{I-tau-down}) to obtain
275: \begin{equation} \label{I-scattering}
276: \tilde{I} = \frac{ (N/T)I_P } {
277: %\frac{\rho}{\tau}
278: \rho/\tau+ R/T } \ .
279: \end{equation}
280: Calculating the electromotive force ${\cal V}_{\rm emf}$ and the
281: internal resistance ${\cal R}_{i}$ amounts to finding a relation
282: \begin{equation} \label{I-circuit}
283: I = \frac{ {\cal V}_{\rm{emf}}}{{\cal R} + {\cal R}_{i}}\ ,
284: \end{equation}
285: where ${\cal R}$ is a suitably defined resistance for the
286: resistor. Comparison of Eqs. (\ref{I-scattering}) and
287: (\ref{I-circuit}) uniquely leads to the result
288: \begin{eqnarray} \label{final-V-emf}
289: {\cal V}_{\rm{emf}} &=& \frac{h}{e^2} \frac{I_P}{T}
290: %= \frac{h}{e^2}
291: %\frac{\sum_a I_a^P}{\sum_{ab} T_{ab}}
292: \\
293: \label{final-R-i} {\cal R}_{i} &=& \frac{h}{Ne^2} \frac{R}{T}
294: %= \frac{h}{Ne^2} \frac{\sum_{ab} R_{ab}}{\sum_{ab} T_{ab}},
295: \ ,
296: \end{eqnarray}
297: provided that
298: \begin{equation} \label{final-R}
299: {\cal R} = \frac{h}{Ne^2} \frac{\rho}{\tau}
300: %= \frac{h}{Ne^2}
301: %\frac{\sum_{ab} \rho_{ab}}{\sum_{ab} \tau_{ab}}.
302: \ .
303: \end{equation}
304: The prefactors have been chosen to make ${\cal V}_{\rm emf}$ an
305: intensive quantity while ${\cal R} ,{\cal R}_{i}\sim N^{-1}$ as $N
306: \rightarrow \infty$.
307: We may apply our results for ${\cal V}_{\rm emf}$ and ${\cal R}_i$
308: to the analytically solvable pipeline model, which assumes that
309: transmission takes place only within a single pair of channels
310: \cite{wagn99}. It can be expressed as:
311: \begin{equation}
312: T_{ab}^{LR}(E_f, E_i) = J\delta_{ab}\delta(E^z_f- E_2)
313: \delta(E^z_i- E_1) \ .
314: \end{equation}
315: Here $E_{\alpha}^z$ ($\alpha=i,f$) is the energy in the direction
316: perpendicular to the planar structure and $(E_2-E_1)/\hbar=\omega
317: >0 $ is the driving frequency. The other scattering probabilities
318: are determined by time-reversal symmetry in the presence of
319: coherent ac driving [$T_{ab}^{LR}(E, E')=T_{ba}^{RL}(E',E)$] and
320: unitarity. For three dimensions, the single pipeline model yields
321: \begin{equation}
322: T = DJ
323: \end{equation}
324: where $D = Am/2\pi\hbar^2$ is the two-dimensional transverse
325: density of states, $A$ being the interface area. Preservation of
326: unitarity requires $DJ<N$. The total pump current is
327: \begin{equation}
328: I_P = eDJ\omega/2\pi \ ,
329: \end{equation}
330: so we interpret $eDJ$ as the pumped charge per cycle. For the
331: circuit parameters we obtain
332: \begin{eqnarray} \label{emf-pump}
333: {\cal V}_{\rm{emf}} &=& \hbar \omega/e
334: %\frac{\hbar \omega}{e}
335: %\ , %\hspace{0.5cm} \forall J
336: \\ \label{int-resist-pump}
337: {\cal R}_{i} &=& \frac{h}{Ne^2} \frac{N-DJ}{DJ}
338: \end{eqnarray}
339: The result that the electromotive force is just $\hbar \omega/e$,
340: independently of the transmittivity $J$, is remarkable if one
341: looks at the general structure of Eqs. (\ref{I-pump-general}),
342: (\ref{RaTa}), (\ref{RT}), and (\ref{final-V-emf}), but could have
343: been expected from the notion that the pipeline model allows only
344: for an energy gain $\hbar \omega$ as the electron is pumped from
345: right to left, regardless of the total electron flow. We readily
346: conclude that, in a more general pump structure, ${\cal
347: V}_{\rm{emf}} \sim \hbar \omega/e$, in agreement with Refs.
348: \cite{swit99,wagn99}. By contrast, the internal resistance is very
349: sensitive to the transmittivity of the pump. In particular, we
350: note that ${\cal R}_{i}\rightarrow \infty$ as $J\rightarrow 0$.
351: We now turn our attention to an open setup where the pump and
352: resistor stay in series within a lead coupled through ideal
353: contacts to broad electron reservoirs. As indicated in Fig. 2, the
354: chemical potentials in the reservoirs characterize the population
355: of the incoming electrons. Hence, we refer to them also as
356: $\mu_{L,R}^{\rm in}$. One may perform an analysis similar to that
357: described for the closed geometry of Fig. 1. After some algebra,
358: one obtains
359: \begin{equation} \label{I-open} I=(e/h)T'(
360: e{\cal V}_{\rm{emf}}+\Delta\mu^{\rm in}) \ ,
361: \end{equation}
362: where
363: %$\Delta \mu^{\rm in}\equiv \mu_{R}^{\rm in} - \mu_{L}^{\rm
364: %in}$,
365: ${\cal V}_{\rm{emf}}$ is given by Eq. (\ref{final-V-emf}), and
366: $T'\equiv (T\tau/N)/(1-R\rho/N^2)$ is the the average transmission
367: through the compound structure formed by the pump plus the
368: resistor.
369: Interestingly, Eq. (\ref{I-open}) can also be written as
370: \begin{equation} \label{I-open-contact}
371: I=\frac{{\cal V}_{\rm emf}+\Delta \mu^{\rm in}/e}{ (h/Ne^2)+{\cal
372: R}+{\cal R}_{i} } \ ,
373: \end{equation}
374: where the resistances ${\cal R}$ and ${\cal R}_i$ are given by
375: Eqs. (\ref{final-R}) and (\ref{final-R-i}) respectively. Thus we
376: see that, within an open geometry, the pump electromotive force
377: adds to the voltage bias generated by the potential difference
378: between the two electron reservoirs. This confirms the intuitive
379: expectation that ${\cal V}_{\rm{emf}}$ can be obtained from the
380: the voltage difference $\Delta\mu^{\rm in}$ needed to cancel the
381: pump current \cite{swit99,wagn99}.
382: A striking difference between Eqs. (\ref{I-circuit}) and
383: (\ref{I-open-contact}) is the role of the contact resistance
384: $h/Ne^2$, which is absent in the case of a closed structure.
385: Comparison of the underlying models suggests that the contact
386: resistance disappears under the assumption that the flow of
387: outgoing electrons on the left of the resistor-pump structure of
388: Fig. 2 is identified with the the flow of incoming electrons from
389: the right, and equivalently for electrons moving in the opposite
390: direction. We conclude that, within a closed circuit of uniform
391: width, there is no natural lower limit to the resistance that the
392: electron current must face as it is generated by the electron
393: pump. This result appears reasonable if one notes that contact
394: resistances along the circuit are generated by narrow-wide
395: contacts where the width of the wire (and thus the number of
396: available electron channels) changes \cite{imry86}.
397: The denominators of Eqs. (\ref{I-circuit}) and
398: (\ref{I-open-contact}) suggest that the resistances which we have
399: introduced should be additive. Unfortunately, the ratios $R/T$ and
400: $\rho/\tau$ cannot be guaranteed to be additive in general. That
401: is possible only in one dimension, where the ration $R/T$ is known
402: to be additive for barriers compounded incoherently \cite{datt95},
403: or, for multichannel wires, in the particular case where the
404: scattering probabilities are independent of the channel index
405: ($R_{ab}=R/N^2$, $T_{ab}=T/N^2$, and similarly for
406: $\rho_{ab},\tau_{ab}$). In such a case, the assumption of a common
407: chemical potential for all incoming electrons ($\mu_{a}^{\rm
408: in}=\mu_{L,R}^{\rm in}$ for all $a \in L,R $) automatically
409: guarantees an outgoing population with a common chemical for all
410: outgoing electrons ($\mu_{b}^{\rm out}=\mu_{L,R}^{\rm out}$ for
411: all $b \in L,R$). Within that scheme, the assumption of a common
412: chemical potential for all electrons moving in a given direction
413: [see Eq. (\ref{equiv-chem-pot})] is internally consistent in the
414: sense that a scenario may be conceived where the outgoing
415: population from a barrier or pump is guaranteed to be a suitable
416: incoming population for the following obstacle. Interestingly, it
417: is also in the channel-independent scattering limit where the
418: resistance defined in Ref. \cite{butt85} becomes additive and
419: equivalent to the resistance defined in Eq. (\ref{final-R}).
420: On closer inspection, one realizes that the assumption of
421: channel-independent scattering is hard to justify within an
422: independent electron picture, where no naturally additive
423: resistance can be defined for multichannel wires without invoking
424: impurity averaging. In particular, such a hypothesis is not
425: satisfied by the pipeline model invoked above, since its
426: transmission depends on the perpendicular energy. We conclude that
427: the question of the definition of an electron pump internal
428: resistance is directly connected to the discussion on the
429: additivity of resistances in multichannel wires. As long as this
430: fundamental issue is not satisfactorily resolved, the transport
431: equations (\ref{I-circuit}) and (\ref{I-open-contact}) [as
432: complemented by (\ref{final-V-emf}), (\ref{final-R-i}), and
433: (\ref{final-R})] which we have derived will have to be viewed as
434: approximations obtained from a reasonable and appealing scheme.
435: This research has been supported by the MCyT (Spain) under Grant
436: No. BFM2001-0172, the EU RTN Programme under Contract No.
437: HPRN-CT-2000-00144, and the Ram\'on Areces Foundation.
438: \begin{thebibliography}{99}
439: \bibitem{swit99} M. Switkes, C. M. Marcus, K. Campman,
440: and A. C. Gossard, Science \textbf{283}, 1905
441: (1999).
442: \bibitem{thou83} D. J. Thouless, Phys. Rev. B {\bf 27}, 6083
443: (1983); Q. Niu, Phys. Rev. Lett. {\bf 64}, 1812 (1990).
444: \bibitem{staf96} C.A. Stafford and Ned S. Wingreen, Phys. Rev. Lett.
445: \textbf{76}, 1916 (1996).
446: \bibitem{pede98} M.H. Pedersen and M. Büttiker, Phys. Rev. B \textbf{58},
447: 12993 (1998).
448: \bibitem{brou98} P. Brouwer, Phys. Rev. B \textbf{58}, R10135 (1998).
449: \bibitem{wagn99} M. Wagner and F. Sols, Phys. Rev. Lett. \textbf{83}, 4377
450: 1999); F. Sols and M. Wagner, Ann. Phys. (Leipzig) \textbf{9}, 776
451: (2000).
452: \bibitem{kim02} S. W. Kim, Phys. Rev. B \textbf{66}, 235304 (2002).
453: \bibitem{lehm03} J. Lehmann, S. Kohler, P. Hänggi, and A. Nitzan, J. Chem.
454: Phys. \textbf{118}, 3283 (2003).
455: \bibitem{land57} R. Landauer, IBM J. Res. Dev. {\bf 1}, 223 (1957); Phil.
456: Mag. {\bf 21}, 863 (1970); M. B\"uttiker, Phys. Rev. Lett. {\bf
457: 57}, 1761 (1986).
458: \bibitem{butt86a} M. B\"uttiker, Phys. Rev. B {\bf 33}, 3020 (1986).
459: \bibitem{ster90} A. Stern, Y. Imry, Y. Aharonov, Phys. Rev. A
460: {\bf 41}, 3436 (1990).
461: \bibitem{sols99} F. Sols and J. Sánchez-Cañizares, Superlatt. and
462: Microstruct. \textbf{25}, 627 (1999).
463: \bibitem{imry86} Y. Imry, in {\it Directions in Condensed Matter},
464: G. Grinstein and E. Mazenko, eds. (World Scientific, Singapore,
465: 1986).
466: \bibitem{datt95} S. Datta, \textit{Electronic Transport in Mesoscopic
467: Systems} (North-Holland, Amsterdam, 1995).
468: \bibitem{butt85} M. Büttiker, Y. Imry, R. Landauer, and S. Pinhas, Phys.
469: Rev. B \textbf{31}, 6207 (1985).
470: \end{thebibliography}
471: %\begin{figure*}
472: %\begin{center}
473: %\includegraphics{figure11.eps}
474: %\caption{HERE GOES THE FIGURE CAPTION}
475: %\end{center}
476: %\end{figure*}
477: \end{document}
478: