1: %% Journal: J. Phys.: Condens. Matter
2: %% Title: Spin exchange in quantum rings and wires in the
3: %% Wigner-crystal limit
4: %% Authors: Michael M. Fogler and Eugene Pivovarov
5: %% Contact address: Dr Michael M. Fogler, Department of Physics,
6: %% Mail Code 0319, University of California San Diego,
7: %% 9500 Gilman Dr, La Jolla, CA 92093-0319, USA
8: %% Contact e-mail: mfogler@ucsd.edu
9:
10: \documentclass{iopart}
11: \usepackage{iopams,graphicx}
12:
13: \begin{document}
14:
15: \letter{Spin exchange in quantum rings and wires in the
16: Wigner-crystal limit}
17:
18: \author{Michael M Fogler and Eugene Pivovarov}
19:
20: \address{Department of Physics, University of
21: California San Diego, La Jolla, California 92093, USA}
22: \ead{mfogler@ucsd.edu}
23:
24: \begin{abstract}
25: We present a controlled method for computing the exchange coupling in
26: strongly correlated one-dimensional electron systems. It is based on the
27: asymptotically exact relation between the exchange constant and the
28: pair-correlation function of spinless electrons. Explicit results are
29: obtained for thin quantum rings with realistic Coulomb interactions, by
30: calculating this function via a many-body instanton approach.
31: \end{abstract}
32:
33: \pacs{71.10.Pm, 73.21.Hb, 73.22.-f}
34: %% 71.10.Pm Fermions in reduced dimensions (anyons,
35: %% composite fermions, Luttinger liquid, etc.)
36: %% 73.21.Hb Quantum wires
37: %% 73.22.-f Electronic structure of nanoscale materials:
38: %% clusters, nanoparticles, nanotubes, and nanocrystals
39: %% 75.75.+a Magnetic properties of nanostructures
40:
41: \submitto{\JPCM}
42:
43: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
44:
45: \section{Introduction}
46:
47: Much attention has been devoted to the spin degree of freedom in
48: one-dimensional (1D) conductors, both of the linear shape (quantum
49: wires~\cite{Auslaender_05}, carbon nanotubes~\cite{Jarillo-Herrero_04})
50: and of a circular one (quantum rings~\cite{Lorke_00, Fuhrer_04,
51: Bayer_03}). Physical parameters of such systems, e.g., average distance
52: between the electrons $a$, their total number $N$, their effective mass
53: $m$, dielectric constant $\epsilon$, etc., can vary over a broad
54: range or can be tuned experimentally. This creates unique opportunities
55: for studying the effect of reduced dimensionality and strong Coulomb
56: interactions on quantum magnetism. A number of new developments have
57: generated a particular interest to the physics of a 1D Wigner-crystal
58: (WC). Unlike the case in higher dimensions, in 1D the crossover to this
59: strongly-correlated regime occurs at easily achievable electron
60: densities~\cite{Egger_99}, $r_s \equiv a / 2 a_B > 4$, where $a_B =
61: \hbar^2 \epsilon / m e^2$ is the effective Bohr radius. Disorder has
62: been the only major obstacle to realizing the 1D WC
63: experimentally~\cite{Auslaender_05}. A
64: successfully solution to this problem has been apparently found, at
65: least, for the case of carbon nanotubes. Very large $r_s$ values have
66: been recently demonstrated in suspended nanotube devices without
67: appreciable intervention of disorder effects~\cite{Jarillo-Herrero_04}.
68: Because of their finite length, in the desired range $r_s > 4$ these
69: devices contained only a few electrons, $N < 25$. Such finite-size
70: systems are traditionally referred to as Wigner
71: molecules~\cite{Reimann_02}. The progress towards realizing Wigner
72: crystal (molecule) states in GaAs quantum wires has also been very
73: encouraging~\cite{Auslaender_05}, therefore, one may hope that they will
74: soon follow suit.
75:
76: On the theoretical side, the 1D WC is interesting because of a dramatic
77: difference between the characteristic energy scales for orbital and spin
78: dynamics. This \emph{strong\/} spin-charge separation has been recently
79: predicted to cause anomalies in many essential electron properties,
80: e.g., ballistic conductance~\cite{Matveev_04} of quantum wires and
81: persistent current of quantum rings~\cite{Reimann_02}. In view of the
82: above, obtaining a reliable estimate of the spin-related energy scales,
83: notably, the exchange-coupling $J$ of the nearest-neighbour electrons, is
84: desirable. It has been an outstanding challenge, though. As depicted in
85: \fref{fig:Instanton}(a), $J$ is determined by the acts of quantum
86: tunnelling in which any two such electrons interchange. At $r_s \gg 1$
87: the corresponding potential barrier greatly exceeds the kinetic energy
88: of the electron pair, which makes $J$ exponential small and difficult to
89: compute numerically~\cite{Reimann_02}. Attempts to derive $J$
90: analytically (for the nontrivial case $N > 2$) were based on the
91: approximation that neglects all degrees of freedom in the problem except
92: the distance between the two interchanging electrons~\cite{Matveev_04,
93: Hausler_96}. We call this a Frozen Lattice Approximation (FLA). The
94: accuracy of the FLA is unclear because it is not justified by any small
95: parameter. When a given pair does its exchange, it sets all other
96: electrons in motion, too (\fref{fig:Instanton}). To obtain the much
97: needed reliable estimate of $J$ one has to treat the spin exchange in a
98: Wigner molecule (or a WC) as a truly many-body process. This is done
99: below in this letter, where we compute $J$ to the leading order in the
100: large parameter $r_s$.
101:
102: %%%%%%%%%%%
103: % FIG. 1
104: %
105: \begin{figure}
106: \begin{center}
107: \includegraphics[height=1.75in]{wcx_1ab.eps}
108: \end{center}
109: \caption{(Colour online)
110: The instanton trajectories. (a) Schematic representation for $N = 6$
111: Wigner molecule on a ring. (b) The trajectories of $1\leq j\leq 4$
112: electrons for $N = 8$ (for notations used see the main text). Inset:
113: function $\tr\bOmega(x)$. The units of $\tau$ and $\bOmega$ are
114: $\sqrt{2} a / s_0$ and its inverse.
115: \label{fig:Instanton}}
116: \end{figure}
117: %%%%%%%%%%%
118:
119: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
120:
121: \section{Model and results}
122:
123: We assume that electrons are tightly confined in the transverse
124: dimensions on a characteristic lengthscale $R \ll a_B$. This allows us
125: to treat the problem as strictly 1D, provided we replace the Coulomb law
126: by the appropriate effective interaction that goes to a finite value at
127: distances $r \ll R$. We adopt a simple model form~\cite{Fogler_05,
128: Usukura_05} $U(r) = e^2 / \epsilon (r + R)$, which is the simplest
129: expression that correctly captures both short- and long-range behaviour
130: of the (unscreened) Coulomb interaction for \emph{any\/} realistic
131: confining potential and is similar to other forms used in the
132: literature~\cite{Friesen_80, Szafran_04}. For convenience, we focus on
133: the quantum ring geometry where $r = (N a / \pi) \left|\sin (\pi x / N
134: a)\right|$ is the chord distance and $x$ is the coordinate along the
135: circumference.
136:
137: In the Wigner molecule configuration electrons reside
138: at the corners of a regular polygon. The effective low-energy
139: Hamiltonian of such a state is given by~\cite{Reimann_02}
140: %
141: \begin{equation}
142: H = \frac{\hbar^2}{2 I} L^2
143: + J \sum_{j} \mathbf{S}_j \mathbf{S}_{j + 1}
144: + \sum_{\alpha} n_a \hbar \omega_{\alpha},
145: \label{eqn:H}
146: \end{equation}
147: %
148: where $L$ is the centre-of-mass angular momentum, $\mathbf{S}_j$ are
149: electron spins and $n_{\alpha}$ are the occupation numbers of
150: ``molecular vibrations.'' At large $r_s$, the total moment of inertia
151: $I$ and the vibrational frequencies $\omega_{\alpha}$ are easy to
152: compute because they are close to their classical values. In order to
153: calculate $J$, which is more difficult, we first show that the
154: asymptotically exact relation exists between $J$ and the
155: pair-correlation function (PCF) $g(x)$ of a \emph{spin polarized\/} 1D
156: system. For an ultrathin wire, $\mathcal{L} \equiv \ln ({a_B}/{R}) \gg
157: 1$, it is particularly compact:
158: %
159: \begin{equation}
160: J = \left( {e^2 a_B^2} / {2 \mathcal{L} \epsilon}\right)
161: g^{\prime\prime}(0),\quad r_s \gg 1 / \mathcal{L}.
162: \label{eqn:J_from_g_ultrathin}
163: \end{equation}
164: %
165: By virtue of \eref{eqn:J_from_g_ultrathin}, the calculation of $J$
166: reduces to an easier task of computing $g(x)$. Using the instanton
167: method described below we arrive at the final result
168: %
169: \begin{equation}
170: J = \frac{\kappa}{\left( 2 r_s\right) ^{5/4}} \frac{\pi}{\mathcal{L}}
171: \frac{e^2}{\epsilon a_B}
172: \exp \left( -\eta \sqrt{2 r_s}\,\right),
173: \quad r_s \gg 1.
174: \label{eqn:J}
175: \end{equation}
176: %
177: The values of $\eta$ and $\kappa$ are given in \tref{tbl:Results}. They
178: demonstrate that the FLA~\cite{Matveev_04, Hausler_96} errs
179: significantly in $\kappa$, by about $50\%$, but surprisingly little in
180: $\eta$, only by 0.7\%.
181:
182: %
183: \begin{table}
184: \caption{\label{tbl:Results}Results for Wigner molecules on a ring
185: (finite $N$) and for wires ($N=\infty$).}
186: \begin{indented}
187: \item[]\begin{tabular}{lcccccc}
188: \br
189: $N$ &3 &4 &6 &8 &$\infty$ &$\infty$-FLA\\
190: \mr
191: $\eta$ &2.8009 &2.7988(2) &2.7979(2) &2.7978(2) &2.7978(2) &2.8168 \\
192: $\kappa$ &3.0448 &3.18(6) &3.26(6) &3.32(7) &3.36(7) &2.20 \\
193: \br
194: \end{tabular}
195: \end{indented}
196: \end{table}
197: %
198:
199: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
200:
201: \section{Three electrons on a ring}
202:
203: We start with the simplest nontrivial case: $N = 3$. Let $0 \leq x_j
204: < 3 a$, $j = 0, 1, 2$, be the electron angular coordinates. We will
205: compute the exchange coupling $J$ between the $j = 0$ and the $j = 1$
206: electrons. It is convenient to go to new variables: the relative
207: distance of the pair, $x \equiv x_1 - x_0$ and the location
208: of the $j = 2$ electron with respect to the centre of mass $X_{2} \equiv
209: x_2 - x_{\mbox{cm}} - a$. The motion of the centre of mass
210: $x_{\mbox{cm}}$ can be ignored. We arrive at the problem with one
211: fast ($x$) and one slow ($X_{2}$) degree of freedom. (Classically,
212: $X_{2} = 0$.) The total potential energy $U_{\mbox{tot}}(x, X_2) = U(x)
213: + U\left[(3 / 2) (X_2 + a) - x / 2\right] + U\left[(3 / 2) (X_2 + a) + x
214: / 2\right]$ has two global minima in the fundamental domain $|x| < 3/2
215: a$, at $x = \pm a$, $X_{2} = 0$. They give rise to the two lowest-energy
216: multiplets: the spin-singlet ground state $\mathbf{S}_0 + \mathbf{S}_1 = 0$
217: with an orbital wavefunction $\Phi_s(x, X_2)$ and the triplet with
218: a wavefunction $\Phi_t$. Their energy splitting is the desired exchange
219: coupling $J$. It is given by the formula~\cite{Herring_64,
220: Landau_III}
221: %
222: \begin{equation}
223: J = (2 \hbar^2 / \mu) \int d X_2
224: \left.\Phi_1 \partial_x \Phi_1\right|_{x = 0},
225: \label{eqn:J_from_Phi_1}
226: \end{equation}
227: %
228: where the (normalized to unity) ``single-well'' wavefunction $\Phi_1(x,
229: X_2)$ is the ground-state of the Hamiltonian with a modified potential
230: $U_{\mbox{tot}} \to U_1 \equiv U_{\mbox{tot}}\left(\max\{x, 0\}, X_2\right)$
231: and $\mu = m/2$. \Eref{eqn:J_from_Phi_1} is valid to order
232: $O(J^2)$~\cite{Herring_64}; with the same accuracy, the singlet and triplet
233: wavefunctions are symmetric and antisymmetric combinations of the
234: single-well wavefunctions,
235: $\Phi_{s,t} = \left[\Phi_1(x, X_{2}) \pm \Phi_1(-x, X_{2})\right] / \sqrt{2}$.
236:
237: Let us discuss the form of $\Phi_1(x, X_{2})$. Near its maximum at
238: $x = a$, $X_2 = 0$, it is a simple Gaussian in both variables,
239: characterized by an amplitude $l$ of the zero-point motion in $x$
240: and a frequency $\Omega(a)$ of the zero-point oscillations in $X_{2}$.
241: Away from its maximum $\Phi_1$ rapidly decays at
242: $|X_2| \gtrsim l \gg a$. This justifies the following Gaussian
243: approximation\footnote{The
244: Gaussian ansatz has been used previously for computing
245: spin exchange in $^3${\sc H}e crystals~\cite{Roger_83}.} in the
246: \emph{entire\/} fundamental domain of $x$:
247: %
248: \begin{equation}
249: \Phi_1 = \phi(x) \exp\left[-(M / 2 \hbar) \Omega(x) X_2^2\right] ,
250: \label{eqn:Phi_1_3}
251: \end{equation}
252: %
253: where $M = 3\mu$.
254: It is important that at $x \ll a$, where the tunnelling barrier is
255: large, $\Omega$ is a slow function of $x$. Hence, if $g(x)$ denotes the
256: PCF of a spin-polarized molecule,
257: %
258: \begin{equation}
259: g(x) \equiv 2 \int \prod _{j = 2}^{N - 1} d X_j
260: \Phi_t^2\left(x, X_2,\ldots, X_{N - 1}\right),
261: \quad |x| < {3 a / 2},
262: \label{eqn:g_def}
263: \end{equation}
264: %
265: then \eref{eqn:J_from_Phi_1} immediately entails
266: $J = \left({\hbar^2} / {4 \mu}\right)
267: \left[{\phi(0)} / {\phi^\prime(0)}\right]
268: g^{\prime\prime}(0)$.
269: Anticipating the discussion below, \eref{eqn:g_def} is written for
270: an arbitrary $N > 2$, with the notation $X_j \equiv x_j - x_{\mbox{cm}} +
271: (N - 1 - 2 j) (a / 2)$ being used; the PCF is normalized as
272: appropriate in the WC limit, $\int_{0}^{3a/2} g(x)\, dx = 1$.
273:
274: %%%%%%%%%%%
275: % FIG. 2
276: %
277: \begin{figure}
278: \begin{center}
279: \includegraphics[height=1.3in]{wcx_2.eps}
280: \end{center}
281: \caption{The PCF of a spin-polarized system (schematically). Regions I, II
282: and III are described in the main text.
283: \label{fig:PCF}}
284: \end{figure}
285: %%%%%%%%%%%
286:
287: The equations on $\phi(x)$ and $\Omega(x)$ are obtained by
288: substituting \eref{eqn:Phi_1_3} into the Schr\"odinger equation and
289: neglecting terms small in $l/a$. This results in
290: the dependence of $g(x)$ on $x$ sketched in
291: \fref{fig:PCF}. Near its $x = a$ maximum (region III) $g(x)$ is
292: a Gaussian of width $l$. In the region II the quasiclassical approximation applies.
293: Finally, in an ultrathin wire, $\mathcal{L} \gg 1$, there is also
294: region I, $x \lesssim a_B$, where the
295: quasiclassical approximation breaks down. Fortunately,
296: the equations on $\phi(x)$ and $\Omega(x)$ can be simplified there, as
297: $\Omega(x)\simeq\Omega(0)$ and $U_{\mbox{tot}}(x)\simeq U(x)+2U(3a/2)$.
298: Similar to~\cite{Fogler_05}, this leads to
299: $\phi(0)/\phi^\prime(0)\simeq a_B /\mathcal{L}$,
300: which, combined with the expression for $J$,
301: yields equations (\ref{eqn:J_from_g_ultrathin})
302: and (\ref{eqn:J}), with $\eta$ and $\kappa$ given by
303: %
304: \begin{eqnarray}
305: \eta &=& 2 \int\limits_0^a \frac{d x}{a}
306: \left[\frac{\epsilon a}{e^2}\, \Delta U_{\mbox{tot}}(x)
307: \right]^{1/2},
308: \label{eqn:eta_3}\\
309: \kappa &=& \frac{2^{5 / 4}}{\sqrt{\pi}}\, e^{\xi(0)}
310: \sqrt{\frac{\Omega(a)}{\Omega(0)}}
311: \left[
312: \frac{\epsilon a^3}{e^2}\, U^{\prime\prime}_{\mbox{tot}}(a)
313: \right]^{3/4}.
314: \label{eqn:kappa_3}
315: \end{eqnarray}
316:
317: Thus, for the $N = 3$ case we were able to reduce the original
318: complicated three-body eigenvalue problem to routine operations of
319: solving an ordinary differential equation on $\Omega(x)$
320: and taking two quadratures.
321: The resultant $\eta$ and $\kappa$ are listed in \tref{tbl:Results}.
322: In comparison, the FLA~\cite{Matveev_04} underestimates $\kappa$ by about
323: $50\%$. It gets $\eta$ correctly but only for $N = 3$, see more below.
324:
325: One important comment is in order. The antisymmetry of the total fermion
326: wavefunction imposes certain selection rules~\cite{Maksym_96} for the
327: allowed values of $L$ [see \eref{eqn:H}] at a given total spin $S$.
328: The lowest-energy $L$ eigenstates for the two possible $S$ values in the
329: $N = 3$ system, $S = 1/2$ and $S = 3/2$, are $|L| = 1$ and $0$,
330: respectively. Since $J \ll \hbar ^{2}/I$ at large $r_s$, the ground
331: state of the system is the $L = 0$ spin-quartet~\cite{Reimann_02,
332: Usukura_05}.
333:
334: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
335:
336: \section{$N > 3$ electrons on a ring}
337:
338: In a system of more than three electrons, the single-well function
339: $\Phi_1(x, \mathbf{X})$ can be sought in the form similar to
340: \eref{eqn:Phi_1_3}, but with the argument of the exponential replaced by
341: $(-1/2\hbar) (\Delta\mathbf{X}^\dagger \mathbf{M}^{1/2}) \bOmega(x)
342: (\mathbf{M}^{1/2}\Delta\mathbf{X})$, where $M^{-1/2}_{ij} =
343: m^{-1/2}[\delta_{ij}- (1 - \sqrt{2 / N}) / (N - 2)]$. In the language of
344: the quantum tunnelling theory, $\bOmega(x)$ is a matrix that controls
345: Gaussian fluctuations $\Delta \mathbf{X} = \mathbf{X} - \mathbf{X}^\ast$
346: around the instanton trajectory $\mathbf{X}^\ast(x)$, where $\mathbf{X}
347: = (X_2,\ldots, X_{N - 1})^{T}$. Switching to the usual
348: parametrization of the instanton by an ``imaginary-time'' $\tau$, we
349: seek $x(\tau)$ and $\mathbf{X}^\ast(\tau)$ that minimize the action
350: %
351: \begin{equation}
352: S_N = \int\limits_0^\infty \frac{d \tau}{\hbar}
353: \left[\frac{\mu}{2} (\partial_\tau x)^2 +
354: \frac12 (\partial_\tau \mathbf{X})^\dagger \mathbf{M}\,
355: \partial_\tau \mathbf{X}
356: + \Delta U_{\mbox{tot}}
357: \right] ,
358: \label{eqn:S_N}
359: \end{equation}
360: %
361: subject to the boundary conditions $x(0) = 0$, $x(\infty) = a$ and
362: $\mathbf{X}(\infty) = 0$. Henceforth $U_{\mbox{tot}}$ is always meant to
363: be evaluated on the instanton trajectory and $\Delta U_{\mbox{tot}}$
364: stands for the difference of its values at a given $\tau$ and at $\tau =
365: \infty$. Repeating the steps of the derivation for the $N = 3$ case, we
366: derive the following equations on $\phi(x)$ and $\bOmega(x)$:
367: %
368: \begin{equation}
369: \eqalign{
370: \partial_\tau \bOmega = \bOmega^2(\tau) - \bomega^2(\tau), \cr
371: \left\{(\hbar^{2} / 2\mu) \partial^2_x - U_{\mbox{tot}}(x)
372: - (\hbar / 2) \tr\,\bOmega(x) + E\right\} \phi(x) = 0,}
373: \label{eqn:Omega}
374: \end{equation}
375: %
376: where $\bomega$ is a positive-definite matrix such that
377: $\bomega ^2=\mathbf{M}^{-1/2}\bXi \mathbf{M}^{-1/2}$ and
378: $\bXi$ is the matrix of the second derivatives $\Xi_{i j} =
379: \partial_{X_i} \partial_{X_j} U_{\mbox{tot}}$.
380: The equations are mutually
381: consistent if $E = U_{\mbox{tot}}(a) + (\hbar / 2) [\tr\bomega(a) +
382: \omega_0]$, $\omega_0 \equiv \hbar / \mu l^2$ and $\bOmega(a) =
383: \bomega(a)$.
384:
385: The PCF $g(x)$ in the quasiclassical region can be written in terms of
386: the tunnelling action (\ref{eqn:S_N}) and the appropriate prefactor
387: as follows:
388: %
389: \begin{eqnarray}
390: g(x) &=& \frac{a}{l^2}
391: \left[\frac{1}{2\pi} \frac{\Omega(a)}{\Omega(x)}
392: \frac{\hbar\omega_0}{U(x)}
393: \right]^{1/2}
394: e^{\xi(x) - 2 S_N(x)},
395: \label{eqn:g_II}\\
396: \xi(x) &=& \int\limits_x^a d y \left\{
397: \frac{\omega_0 + \tr\bOmega(a) - \tr\bOmega(y)}
398: {\left[(2 / \mu) \Delta U_{\mbox{tot}}(y)\right]^{1/2}}
399: - \frac{1}{a - y}
400: \right\}.
401: \label{eqn:xi}
402: \end{eqnarray}
403: %
404: Here the action $S_N$ is defined to be the value of the integral in
405: \eref{eqn:S_N} when its lower limit is replaced by $\tau = \tau(x)$. For
406: $\eta$ we find $\eta = 2 S_N / \sqrt{2 r_s}$, while $\kappa$
407: is given by equation (\ref{eqn:kappa_3}) after the replacement $\Omega
408: \to \det\bOmega$.
409:
410:
411: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
412:
413: \section{Calculation of the instanton}
414:
415: A few properties of the instanton follow from general considerations.
416: The dimensional analysis of action (\ref{eqn:S_N}) yields $S
417: _N\propto\sqrt{r_s}$, so that $\eta$ is indeed just a constant. Also,
418: from the symmetry of the problem, $X_{N + 1 - j}(\tau) = -X_j(\tau)$.
419: Thus, in the special case of $N = 3$, the instanton trajectory is
420: trivial: $X_2 \equiv 0$, i.e., the $j = 2$ electron does not move. This
421: is why we were able to compute $S_3$ in a closed form. For $N > 3$ the
422: situation is quite different: all electrons [except $j = (N + 1) / 2$
423: for odd $N$] \emph{do\/} move. In order to investigate how important the
424: motion of electrons distant from the $j = 0, 1$-pair is let us consider
425: the $N = \infty$ (\emph{quantum wire\/}) case, where the far-field
426: effects are the largest. If $X_j$'s were small, we could expand $\Delta
427: U_{\mbox{tot}}$ in \eref{eqn:S_N} to the second order in $X_j$ to obtain
428: the harmonic action
429: %
430: \begin{equation}
431: S_h = \frac12 \frac{m}{\hbar} \int
432: \frac{d k}{2\pi}
433: \int \frac{d\omega}{2\pi} \left|u_{k \omega}\right|^2
434: \left[\omega^2 + \omega_p^2(k)\right],
435: \label{eqn:action_harmonic}
436: \end{equation}
437: %
438: where $u_{k \omega}$ is the Fourier transform of electron displacement
439: $u_j(\tau) \equiv x_j - x_j^0$ from the classical equilibrium position
440: $x_j^0 \equiv (j - 1 / 2) a$, $j \in \mathbb{Z}$, $\omega_p(k) \simeq
441: s_0 k \ln^{1/2}(4.15 / k a)$ is the plasmon dispersion in the 1D WC and
442: $s_0 \equiv (e^2 / \epsilon\mu a)^{1/2}$. Minimization of $S_h$ with the
443: specified boundary conditions yields $u_j(\tau) \propto v x_j^0
444: /\left[(x_j^0)^2 + v^2 \tau^2\right]$, where $v \simeq (s_0 / 2)
445: \ln\left\{\left[(x_j^0)^2 + s_0^2 \tau^2\right] / a^2\right\}$.
446: Substituting this formula into harmonic action
447: (\ref{eqn:action_harmonic}), we find that the contributions of distant
448: electrons to $S_h$ rapidly decay with $|j|$. Thus, a fast convergence of
449: $\eta$ to its thermodynamic limit is expected as $N$ increases.
450: Encouraged by this conclusion, we undertook a direct numerical
451: minimization of $S$ for the set of $N$ listed in \tref{tbl:Results}
452: using standard algorithms of a popular software package \textsc{MATLAB}.
453: The optimal trajectories that we found for the case of $N = 8$ are shown
454: in \fref{fig:Instanton}(b). As one can see, electron displacements
455: reach some finite fractions of $a$ at $\tau = 0$. This collective
456: electron motion lowers the effective tunnelling barrier and causes $\eta$
457: to drop below its FLA value, although only by $0.7\%$, see
458: \tref{tbl:Results}.
459:
460: Let us now discuss the prefactor $\kappa$. In the inset of
461: \fref{fig:Instanton}(b) we plot $\tr\bOmega(x)$
462: computed by solving \eref{eqn:Omega} numerically. To reduce the
463: calculational burden, we set $\mathbf{X}^\ast(\tau) \to 0$ instead of
464: using the true instanton trajectory. The error in $\kappa$ incurred
465: thereby is $\sim 2\%$. In comparison, the FLA, where
466: $\tr\bOmega(x) = \mbox{const}$, yields $\kappa$ about
467: $50\%$ smaller than the correct result, similar to $N = 3$.
468:
469: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
470:
471: \section{Relation to current experiments}
472:
473: For carbon nanotube quantum dots~\cite{Jarillo-Herrero_04}, where the WC
474: limit has apparently been realized, our formula~\eref{eqn:J} gives $J
475: \sim 1\,{\rm K}$ at $r_s = 4$, which should be verifiable
476: experimentally. Unfortunately, the lowest measurement temperature was
477: $0.3\,{\rm K}$; therefore, the exchange correlations may have been
478: washed out. We hope that our predictions can be checked in the next
479: round of experiments. Energy-level spectroscopy of quantum
480: rings~\cite{Lorke_00, Fuhrer_04, Bayer_03} is another promising area
481: where our results may apply. In longer 1D wires, $J$ determines the
482: velocity $v_\sigma = (\pi / 2) J a / \hbar$ of spin excitations, which
483: can be measured by tunnelling~\cite{Auslaender_05},
484: photoemission~\cite{Claessen_02}, or deduced from the enhancement of the
485: spin susceptibility and electron specific heat~\cite{Fogler_05}. Our
486: result for $v_\sigma$ reads (cf. \tref{tbl:Results})
487: %
488: \begin{equation}
489: v_{\sigma}/v_{F} = 5.67 \left( \pi / \mathcal{L}\right)
490: r_s^{3/4} e^{-\eta\sqrt{2 r_s}},\quad \eta = 2.7978(2),
491: \label{eqn:v_sigma_Coulomb}
492: \end{equation}
493: %
494: where $v_F = (\pi / 2) (\hbar / m a)$ is the Fermi velocity.
495:
496: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
497:
498: %\ack
499: \bigskip\noindent\ignorespaces
500: This work is supported by the A.~P. Sloan and the C.~\&~W.
501: Hellman Foundations.
502:
503: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
504:
505: \bigskip\noindent\emph{Note added.}---After the completion of this work, we
506: learned that Klironomos \etal \cite{Klironomos_05} independently computed
507: $\eta = 2.79805(5)$, but not the prefactor $\kappa$. These authors also
508: considered a correction to $\eta$ due to a finite radius of the wire
509: $R$. We can show that as $R$ increases, the ratio $\pi / \mathcal{L}$ in
510: \eref{eqn:v_sigma_Coulomb} is replaced by a more complicated
511: expression that tends to unity at $R > a_B$.
512:
513: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
514:
515: \section*{References}
516: \begin{thebibliography}{99}
517:
518: \bibitem{Auslaender_05} Auslaender O~M, %\etal
519: Steinberg~H, Yacoby~A, Tserkovnyak~Y, Halperin B~I,
520: Baldwin K~W, Pfeiffer L~N and West K~W
521: 2005 \textit{Science} \textbf{308} 88
522: and references therein
523: \item[] Field S~B, Kastner M~A, Meirav~U,
524: Scott-Thomas J~H~F, Antoniadis D~A, Smith H~I and Wind S~J
525: 1990 \PR B \textbf{42} 3523
526:
527: \bibitem{Jarillo-Herrero_04} Jarillo-Herrero~P, %\etal
528: Sapmaz~S, Dekker~C, Kouwenhoven L~P and van der Zant H~S~J
529: 2004 \textit{Nature} \textbf{429} 389
530:
531: \bibitem{Lorke_00} Lorke~A, %\etal
532: Luyken R~J, Govorov A~O, Kotthaus J~P,
533: Garcia J~M and Petroff P~M
534: 2000 \PRL \textbf{84} 2223
535: \item[] Warburton R~J, %\etal
536: Sch\"aflein~C, Haft~D, Bickel~F, Lorke~A,
537: Karrai~K, Garcia J~M, Schoenfeld~W and Petroff P~M
538: 2000 \textit{Nature} \textbf{405} 926
539:
540: \bibitem{Fuhrer_04} Fuhrer~A, Ihn~T, Ensslin~K, Wegscheider~W
541: and Bichler M
542: 2004 \PRL \textbf{93} 176803
543:
544: \bibitem{Bayer_03} Bayer~M, %\etal
545: Korkusinski~M, Hawrylak~P, Gutbrod~T, Michel~M
546: and Forchel~A
547: 2003 \PRL \textbf{90} 186801
548:
549: \bibitem{Egger_99} Egger R, H\"ausler W, Mak C~H and Grabert~H
550: 1999 \PRL \textbf{82} 3320
551:
552: \bibitem{Reimann_02} Reimann S~M and Manninen~M
553: 2002 \RMP \textbf{74} 1283
554: \item[] Viefers~S, Koskinen~P, Singha Deo~P and Manninen~M
555: 2004 \textit{Physica} E \textbf{21} 1
556:
557: \bibitem{Matveev_04} Matveev K~A
558: 2004 \PR B \textbf{70} 245319
559:
560: \bibitem{Hausler_96} H\"ausler~W
561: 1996 \ZP B \textbf{99} 551
562:
563: \bibitem{Fogler_05} Fogler M~M
564: 2005 \PR B \textbf{71} 161304(R)
565:
566: \bibitem{Usukura_05} Usukura~J, Saiga~Y and Hirashima D~S
567: 2005 \JPSJ \textbf{74} 1231
568: and references therein
569:
570: \bibitem{Friesen_80} Friesen W~I and Bergerson B
571: 1980 \JPC \textbf{13} 6627
572:
573: \bibitem{Szafran_04} Szafran~B, Peeters F~M,
574: Bednarek~S, Chwiej~T and Adamowski~J
575: 2004 \PR B \textbf{70} 035401
576:
577: \bibitem{Herring_64} Herring~C
578: 1964 \PR \textbf{134} A362
579:
580: \bibitem{Landau_III} Landau L~D and Lifshitz E~M
581: 1977 \textit{Quantum Mechanics} (Oxford: Pergamon) sec 81
582:
583: \bibitem{Roger_83} Roger~M, Hetherington J~H and Delrieu J~M
584: 1983 \RMP \textbf{55} 1
585:
586: \bibitem{Maksym_96} Maksym P~A
587: 1996 \PR B \textbf{53} 10871
588: \item[] Koskinen~P, Koskinen~M and Manninen~M
589: 2002 \textit{Eur. Phys. J.} B \textbf{28} 483
590:
591: \bibitem{Claessen_02} Claessen R, %\etal
592: Sing~M, Schwingenschl\"ogl~U, Blaha~P, Dressel~M
593: and Jacobsen C~S
594: 2002 \PRL \textbf{88} 096402
595:
596: \bibitem{Klironomos_05} Klironomos A~D,
597: Ramazashvili R~R and Matveev K~A
598: 2005 \textit{Preprint} cond-mat/0504118
599:
600: \end{thebibliography}
601:
602: \end{document}
603: