1: \documentclass[12pt]{article}
2: \usepackage{epsfig}
3: \newcommand{\ket}[1]{|#1\rangle}
4: \newcommand{\bra}[1]{\langle #1|}
5: \newcommand{\Tr}{\text{Tr}}
6: \title{On the alleged nonlocal and topological nature of the
7: molecular Aharonov-Bohm effect}
8: \author{Erik Sj\"{o}qvist\footnote{Electronic address:
9: erik.sjoqvist@kvac.uu.se} \\
10: Department of Quantum Chemistry, Uppsala University, \\
11: Box 518, Se-751 20 Uppsala, Sweden}
12: \begin{document}
13: \maketitle
14: \begin{abstract}
15: The nonlocal and topological nature of the molecular
16: Aharonov-Bohm (MAB) effect is examined for real electronic
17: Hamiltonians. A notion of preferred gauge for MAB is suggested.
18: The MAB effect in the linear $+$ quadratic $E\otimes \varepsilon$
19: Jahn-Teller system is shown to be essentially analogues to an
20: anisotropic Aharonov-Casher effect for an electrically neutral
21: spin$-\frac{1}{2}$ particle encircling a certain configuration
22: of lines of charge.
23: \end{abstract}
24: \vskip 0.3 cm
25: \noindent
26: {\bf Key words:} Molecular Aharonov-Bohm effect, locality, topology,
27: Berry's phase
28: \newpage
29: \section{Introduction}
30: The molecular Aharonov-Bohm (MAB) effect, first hinted
31: at by Longuet-Higgins and coworkers
32: \cite{longuet58,longuet61,herzberg63,longuet75}, is one of the
33: paradigmatic examples on early anticipations of Berry's discovery
34: \cite{berry84} of geometric phase factors accompanying cyclic
35: adiabatic changes. The importance of MAB ranges from testable
36: shifts of the vibronic energy spectrum \cite{kendrick97,vonbusch98},
37: effects on cross-sections in molecular reactions
38: \cite{kupperman93,kendrick96,adhikari00}, and effects on reduction
39: factors \cite{sjoqvist94}, to subtle symmetry
40: assignments of vibronic states in Jahn-Teller systems
41: \cite{ham87,ham90,rios96,moate96} and search for conical intersections
42: \cite{longuet75,stone76,varandas79,xantheas90,ceotto00,johansson03}.
43: However, it is probably fair to say that interest in the MAB effect
44: itself arose first after it was realized \cite{mead79,mead80} its
45: mathematical analogy with the standard Aharonov-Bohm (AB) effect
46: \cite{aharonov59}. This mathematical analogy is the independence
47: of details of the shape of the closed path in a multiply connected
48: region of nuclear configuration space.
49:
50: In a recent work \cite{sjoqvist02}, the present author has considered
51: the physical nature of this analogue in the case of the $E\otimes
52: \varepsilon$ Jahn-Teller system. The question addressed there was: does
53: MAB share all the remarkable properties of AB? The main outcome of
54: this analysis was that although MAB obeys the above mentioned
55: independence of details of the closed path in nuclear configuration
56: space, it does not share the remaining nonlocal and topological
57: properties of AB. The purpose of the present paper is to develop the
58: local and nontopological nature of MAB further. In particular, we wish
59: to extend \cite{sjoqvist02} to any molecular system with real
60: electronic Hamiltonians. We also wish to propose a notion of preferred
61: gauge for MAB in this case, in terms of a so-called gauge invariant
62: reference section \cite{pati95a,pati95b} that appears naturally in the
63: theory of the open path Berry phase
64: \cite{samuel88,mukunda93}.
65:
66: The starting point for the argumentation in \cite{sjoqvist02} is a
67: set of criteria for a nonlocal and topological phase effect, first
68: explicitly introduced by Peshkin and Lipkin \cite{peshkin95}. These
69: are:
70: \begin{itemize}
71: \item A phase effect is nonlocal if:
72: \begin{itemize}
73: \item[(N1)] the system experiences no physical field,
74: \item[(N2)] no exchange of physical quantity takes place along the
75: system's path.
76: \end{itemize}
77: \item A phase effect is topological if:
78: \begin{itemize}
79: \item[(T1)] it requires the system to be confined to a
80: multiply connected region,
81: \item[(T2)] any assignment of phase shift along the system's path
82: is necessarily gauge dependent and thus neither objective nor
83: experimentally testable.
84: \end{itemize}
85: \end{itemize}
86: One can argue that AB fulfills all these criteria for a nonlocal and
87: topological phase effect. On the other hand, in the case of MAB, where
88: the physical system consists of the nuclear configuration and a set
89: electronic variables, the situation is very different. First, MAB is
90: local in that it obeys neither (N1) nor (N2): the nuclei experiences
91: the Coulomb field from the electrons, and there must be a local
92: exchange of force to create the change in the electronic state
93: necessary for the appearance of points in nuclear configuration space,
94: across which the electronic open path Berry phase factor changes
95: sign discontinuously \cite{sjoqvist97,garcia98,englman99,englman00}.
96: Secondly, MAB is nontopological in that (T2) fails since there is an
97: objective and experimentally testable open path electronic Berry phase
98: that causes the effect on the nuclear motion. Only (T1) is fulfilled
99: for MAB: the nuclei must be confined to a multiply connected region
100: for the effect to occur. We wish to discuss the above criteria in
101: relation to MAB systems of general kind.
102:
103: In the next section, we extend the argumentation of \cite{sjoqvist02}
104: to the case of arbitrary molecular systems with real electronic
105: Hamiltonians. A notion of preferred gauge for MAB in this case is
106: suggested in section III and applied in detail to the $E\otimes
107: \varepsilon$ Jahn-Teller system. The analogue between the MAB effect in
108: the $E\otimes \varepsilon$ Jahn-Teller system and the Aharonov-Casher
109: effect \cite{aharonov84}, as pointed out in \cite{sjoqvist02}, is
110: further developed in section IV, so as to treat linear and quadratic
111: coupling simultaneously. The paper ends with the conclusions.
112:
113: \section{Locality and topology}
114: The argumentation in \cite{sjoqvist02} that MAB is essentially a local
115: and nontopological effect was put forward in the special case
116: of the $E\otimes \varepsilon$ Jahn-Teller system. This raises the
117: question whether the main conclusions arrived at in
118: \cite{sjoqvist02} also apply to other molecular systems that may be
119: described accurately by real electronic Hamiltonians. In this section
120: we address this issue and argue that MAB also in the general case
121: is local and nontopological.
122:
123: For sake of clarity, we focus on the motion in some pseudorotational
124: (internal) nuclear coordinate $\theta$, as described by the vibronic
125: Hamiltonian
126: \begin{eqnarray}
127: H = \frac{1}{2} p_{\theta}^{2} + H_e (\theta) ,
128: \label{eq:vibronicham}
129: \end{eqnarray}
130: where $p_{\theta}$ is the canonical momentum corresponding to $\theta$
131: and $H_e (\theta)$ is the electronic Hamiltonian assumed to be real,
132: traceless, and fulfilling $H_e (\theta+2\pi)=H_e (\theta)$. Let $\{
133: \ket{n} \}_{n=1}^{N}$ be a fixed orthonormal basis of the electronic
134: Hilbert space of finite dimension $N$. Furthermore, let $\{ R(\theta)|
135: \theta \in [0,2\pi)$ be a one-parameter set of members of the rotation
136: group SO(N) in ${\bf R}^N$, in terms of which the orthonormal
137: instantaneous eigenvectors of $H_e (\theta)$ read $\{ \ket{n(\theta)} =
138: R(\theta) \ket{n} \}_{n=1}^{N}$. Then, we may write
139: $H_e (\theta) = R(\theta) E(\theta) R^{\textrm{T}} (\theta)$ with
140: ${\textrm{T}}$ being transpose and $E(\theta) = {\textrm{diag}} \big[
141: E_1(\theta),\ldots, E_N (\theta) \big]$, where $E_1
142: (\theta),\ldots,E_N(\theta)$ are the electronic eigenenergies
143: corresponding to $\ket{1(\theta)}, \ldots ,
144: \ket{N(\theta)}$.
145:
146: The Born-Oppenheimer regime is attained when the electronic
147: eigenenergies $\{ E_{n}(\theta) \}_{n=1}^N$ are well separated so that
148: the nuclear motion takes place on a single electronic potential energy
149: surface, let us say $E_n (\theta)$. This may be described by the
150: effective nuclear Hamiltonian
151: \begin{eqnarray}
152: H_{n} = \langle n(\theta) |H| n(\theta) \rangle =
153: \frac{1}{2} p_{\theta}^{2} + E_{n} (\theta) ,
154: \end{eqnarray}
155: where we have used that $\bra{n} R^{\textrm{T}} (\theta) p_{\theta}
156: R(\theta) \ket{n} = 0$. The issue of MAB arises when considering the
157: single-valuedness of the molecular state vector $\ket{\Psi (\theta)}$,
158: which in the Born-Oppenheimer regime is a product of the instantaneous
159: electronic energy eigenvector $\ket{n(\theta)}$ and a nuclear factor
160: $\chi(\theta)$, i.e., $\ket{\Psi (\theta)} = \chi (\theta)
161: \ket{n(\theta)}$. Here, we use the position representation of the
162: nuclear motion in the pseudorotational angle $\theta$. The
163: single-valuedness of $\ket{\Psi(\theta)}$ requires that any
164: multi-valuedness of the electronic part must be compensated for by the
165: nuclear part. Thus, if $\ket{n(\theta + 2\pi)} = -\ket{n(\theta)}$,
166: $\chi (\theta)$ must fulfill the boundary condition
167: $\chi (\theta + 2\pi) = -\chi (\theta)$.
168: A nontrivial MAB effect arises if and only if there is no
169: phase transformation of the form $\ket{n(\theta)} \rightarrow
170: \ket{\widetilde{n}(\theta)} = e^{i\xi (\theta)} \ket{n(\theta)}$
171: so that the effective vector potential
172: \begin{eqnarray}
173: A_n (\theta) \equiv
174: i \bra{\widetilde{n}(\theta)} \partial_{\theta}
175: \ket{\widetilde{n}(\theta)} = - \partial_{\theta} \xi (\theta)
176: \label{eq:mabpotential}
177: \end{eqnarray}
178: vanishes everywhere and at the same time retaining single-valuedness
179: for $\chi (\theta)$ under the requirement that $\ket{\Psi(\theta)}$
180: is single-valued.
181:
182: Under the condition that there is a nontrivial MAB effect, let us
183: ask: what is its physical nature in relation to the standard AB
184: effect? Let us first address the issue of locality. As already
185: indicated, the standard AB effect, which may occur when a charged
186: particle encircles a line of magnetic flux, is nonlocal as it fulfills
187: the criteria (N1) and (N2): the effect arises although the particle
188: experiences no physical field and no exchange of physical quantity
189: takes place along the particle's path. Could the same be said
190: about the MAB effect? One way to address this question is to
191: note that since the electronic Born-Oppenheimer states are eigenstates
192: of $H_e (\theta)$, it may be tempting to replace the electronic
193: motion by the appropriate eigenvalue of $H_e (\theta )$ in
194: the Born-Oppenheimer regime so that the electronic variables can
195: be ignored, creating an illusion that the nontrivial effect of the
196: MAB vector potential on the nuclear motion is nonlocal and topological
197: in the sense of the standard AB effect. However, this argument
198: fails essentially because the electronic variables are
199: dynamical and do not commute among themselves. Thus, there always
200: exist a subset of these variables, namely those that are off-diagonal
201: in the instantaneous electronic eigenbasis $\{ \ket{n(\theta)}
202: \}_{n=1}^{N}$, whose expectation values vanish in the Born-Oppenheimer
203: limit, but whose fluctuations do not. These fluctuations are due to
204: the local interaction between the nuclear variables and the electronic
205: degrees of freedom. Thus, both (N1) and (N2) fail for MAB.
206:
207: Next, we address the issue of topology. The standard AB effect is
208: topological in that (T1) it may occur although the region of magnetic
209: field is inaccessible to the encircling charged particle and in that
210: (T2) the charged particle only feels the gauge dependent vector
211: potential along the path. In the case of MAB, recall that two types
212: of degrees of freedom are involved: those of the electrons and
213: those associated with the nuclear configuration. Now, any assignment
214: of nuclear phase for open paths in nuclear configuration space is
215: necessarily gauge dependent and therefore unphysical. On the other
216: hand, there is an objective way to relate the origin of the MAB phase
217: effect locally in nuclear configuration space using the open path Berry
218: phase $\gamma_{n}$ for the corresponding electronic Born-Oppenheimer
219: state vector $\ket{\widetilde{n} (\theta)} = e^{i\xi (\theta)}
220: \ket{n(\theta)}$. Here, we assume $\xi (\theta)$ to be differentiable
221: along the path but otherwise arbitrary. The noncyclic Berry phase is
222: defined by removing the accumulation of local phase changes from the
223: total phase and is testable in polarimetry \cite{garcia98,larsson03}
224: or in interferometry \cite{wagh98,sjoqvist01}. We obtain for
225: $\ket{\widetilde{n} (\theta)}$
226: \begin{eqnarray}
227: \gamma_{n} & = &
228: \arg \langle \widetilde{n}(\theta_0)\ket{\widetilde{n}(\theta)} +
229: i \int_{\theta_{0}}^{\theta} \langle \widetilde{n}(\theta') |
230: \frac{\partial}{\partial \theta'} |\widetilde{n} (\theta') \rangle
231: d\theta' = \arg \langle n(\theta_0)\ket{n(\theta)} ,
232: \end{eqnarray}
233: by using Eq. (\ref{eq:mabpotential}). Clearly, $\gamma_{n}$ is locally
234: gauge invariant as it is independent of $\xi (\theta)$. It corresponds
235: to phase jumps of $\pi$ at points across which the real-valued
236: quantity $\langle n(\theta_0) \ket{n(\theta)}$ goes through zero and
237: changes sign. In the case where an even number of $\pi$ phase jumps
238: occurs, the nuclear factor $\chi(\theta)$ is single-valued and there
239: is no MAB effect. On the other hand, for an odd number of such jumps,
240: there is a physically nontrivial sign change for such a loop. Thus,
241: the presence of a nontrivial MAB effect could be explained locally as
242: it requires the existence of points along the nuclear path where the
243: electronic states at $\theta_{0}$ and $\theta$ become orthogonal. This
244: assignment of electronic Berry phase shift is gauge invariant at each
245: point along the nuclear path and thus experimentally testable in
246: principle. It shows that MAB does not obey the criterion (T2) for a
247: topological phase effect.
248:
249: \section{Preferred gauge}
250: The assignment of a gauge invariant electronic Berry phase for any
251: open portion of the closed nuclear path suggests that MAB is not
252: topological and that it might be meaningful to introduce a notion of
253: preferred gauge in this context. The corresponding preferred vector
254: potential is defined as that whose line integral gives the open path
255: Berry phase. This idea can be put forward in terms of a so-called
256: gauge invariant reference section \cite{pati95a,pati95b} as follows.
257:
258: First, in general terms, let $\ket{\psi (s)}$ be a normalized Hilbert
259: space representative of the pure quantal state $\psi (s)$ tracing out
260: the path ${\cal C}: s\in [s_0,s_1] \rightarrow \psi(s)$ in projective
261: Hilbert space ${\cal P}$. Now, if $0 \neq \big| \bra{\psi (s_0)}
262: \psi (s_1)\rangle \big| \leq 1$, then the Berry phase $\gamma
263: [{\cal C}]$ associated with ${\cal C}$ is \cite{mukunda93}
264: \begin{eqnarray}
265: \gamma [{\cal C}] & = & \arg \bra{\psi (s_0)} \psi (s_1) \rangle -
266: \int_{s_0}^{s_1} \arg \bra{\psi (s)} \psi (s+ds) \rangle
267: \nonumber \\
268: & = & \arg \bra{\psi (s_0)} \psi (s_1) \rangle +
269: i \int_{s_0}^{s_1} \bra{\psi (s)} \dot{\psi} (s) \rangle \, ds ,
270: \label{eq:ncgp}
271: \end{eqnarray}
272: where the first and second term on the right-hand side contain
273: the global phase and the accumulation of local phase changes,
274: respectively. This open path Berry phase is real-valued and
275: reparametrization invariant \cite{mukunda93}. It is gauge
276: invariant and thereby measurable
277: \cite{garcia98,wagh98,sjoqvist01,larsson03} in that it
278: is independent of choice of Hilbert space representative.
279: $\gamma [{\cal C}]$ reduces to the cyclic Berry phase
280: \cite{berry84} in the particular case where
281: $\big| \bra{\psi (s_0)} \psi (s_1) \rangle \big| = 1$.
282:
283: A gauge invariant reference section $\ket{\phi (s)}$ is
284: a nonlinear functional of $\ket{\psi (s)}$ defined as
285: \cite{pati95a,pati95b}
286: \begin{eqnarray}
287: \ket{\phi (s;s_0)} =
288: \exp \Big( -i\arg \bra{\psi (s_0)} \psi (s) \rangle \Big)
289: \ket{\psi (s)} ,
290: \label{eq:refsec}
291: \end{eqnarray}
292: which has the image ${\cal C}$ in ${\cal P}$. $\ket{\phi (s;s_0)}$
293: is in one to one correspondence with ${\cal C}$ in that it is gauge
294: invariant under phase transformations of $\ket{\psi (s)}$. Furthermore,
295: by inserting Eq. (\ref{eq:refsec}) into Eq. (\ref{eq:ncgp}), we obtain
296: \begin{eqnarray}
297: \gamma [{\cal C}] = \int_{s_0}^{s_1} {\cal A} (s;s_0) ds ,
298: \end{eqnarray}
299: where the gauge function reads
300: \begin{eqnarray}
301: {\cal A} (s;s_0) = i \bra{\phi (s;s_0)} \partial_s \phi (s;s_0) \rangle .
302: \label{eq:genrefsec}
303: \end{eqnarray}
304:
305: Now, the one to one correspondence between the path ${\cal C}$
306: and $\{ \ket{\phi_n (s;s_0)} |$ $s\in [s_0,s_1] \}$, and the
307: fact that the open path Berry phase can be directly expressed in terms
308: of the vector potential ${\cal A} (s;s_0)$, suggest that the gauge
309: invariant reference section defined in Eq. (\ref{eq:refsec}) has a
310: special status: it is natural to regard $\ket{\phi_n (s;s_0)}$
311: and ${\cal A} (s;s_0)$ to constitute a `preferred gauge' for the
312: measurable open path Berry phase. In doing so, one should take notice
313: that the functional form of the preferred vector potential ${\cal A}
314: (s;s_0)$ depends in general upon $\psi (s_0)$, but is otherwise
315: unique up to an additional term whose integral over the interval
316: $[s_0,s_1]$ equals an integral multiple of $2\pi$. In particular, once
317: $\psi (s_0)$ has been chosen the corresponding phase factor is
318: gauge invariant under phase transformations of $\ket{\psi (s)}$.
319:
320: We next apply this idea and calculate the gauge invariant electronic
321: reference section relevant for MAB and its concomitant vector
322: potential in the case of real electronic Hamiltonians. Let the
323: pseudorotational angle $\theta \in [\theta_0,\theta_0+2\pi]$
324: parametrize a closed path ${\cal C}_N$ in nuclear configuration
325: space. Let $\ket{n (\theta)}=R(\theta)\ket{n}$ be a Hilbert space
326: representative of a nondegenerate electronic energy eigenstate along
327: this path. Under these assumptions, consider a finite portion $\Delta
328: {\cal C}_N$ of ${\cal C}_N$. First, suppose that $\bra{n (\theta_0)}
329: n(\theta)
330: \rangle \neq 0, \ \theta \in \Delta {\cal C}_N$. Along such a
331: $\Delta {\cal C}_N$, the gauge invariant reference section reads
332: \begin{eqnarray}
333: \ket{\phi_n (\theta;\theta_0)} =
334: \exp \Big( -i\arg \bra{n (\theta_0)} n(\theta) \rangle \Big)
335: \ket{n(\theta)} = \pm \ket{n(\theta)} ,
336: \label{eq:nonode}
337: \end{eqnarray}
338: where the sign is independent of $\theta$. This follows since
339: $\bra{n (\theta_0)} n(\theta) \rangle$ is real-valued and may
340: only change sign where it vanishes. Inserting Eq. (\ref{eq:nonode})
341: into Eq. (\ref{eq:genrefsec}), we obtain
342: ${\cal A}_n (\theta;\theta_0) = 0, \ \theta \in \Delta {\cal C}_N$.
343:
344: Now, suppose there is a single isolated point $\theta_k \in \Delta
345: {\cal C}_N$ across which $\bra{n (\theta_0)} n(\theta_k) \rangle$ goes
346: through zero and changes sign. For such a point, it follows that
347: \begin{eqnarray}
348: \ket{\phi_n (\theta;\theta_0)} = \pm
349: \exp \Big( i\pi h_s (\theta - \theta_k) \Big) \ket{n(\theta)} , \
350: \theta \in \Delta {\cal C}_N ,
351: \end{eqnarray}
352: where $h_s$ is the unit step function. This yields
353: \begin{eqnarray}
354: {\cal A}_n (\theta;\theta_0) = -\pi
355: \delta (\theta - \theta_k), \ \theta \in \Delta {\cal C}_N ,
356: \end{eqnarray}
357: where we have used that $\partial_{\theta} h_s(\theta - \theta_k) =
358: \delta (\theta - \theta_k)$.
359:
360: Extending this to the whole closed path ${\cal C}_N$, we may assume
361: it contains $K$ angles $\theta_1, \ldots , \theta_K$, all across
362: which $\bra{n(\theta_0)} n(\theta) \rangle$ goes through zero and
363: changes sign. Then, the gauge invariant reference section reads
364: \begin{eqnarray}
365: \ket{\phi_n (\theta;\theta_0)} =
366: \exp \left( i\pi \sum_{k=1}^{K} h_s (\theta - \theta_k) \right)
367: \ket{n (\theta)}, \ \theta \in {\cal C}_N
368: \end{eqnarray}
369: with the corresponding preferred vector potential
370: \begin{eqnarray}
371: {\cal A}_n (\theta;\theta_0) = -\pi \sum_{k=1}^{K}
372: \delta (\theta - \theta_k) , \ \theta \in {\cal C}_N .
373: \label{eq:kpotential}
374: \end{eqnarray}
375: Notice that ${\cal A}_n (\theta;\theta_0)$ is a local quantity
376: as it only depends upon the angles $\theta_1, \ldots, \theta_K$,
377: whose location is gauge invariant once the `initial' electronic
378: eigenstate $n(\theta_0)$ has been chosen.
379:
380: There is a nontrivial MAB effect if and only if $K$ is odd, since
381: \begin{equation}
382: \gamma_n [{\cal C}_N] = \oint_{{\cal C}_N} {\cal A}_n d\theta = -K\pi .
383: \end{equation}
384: It is important to notice that this criterion for a nontrivial MAB is
385: only fulfilled if the effective nuclear motion has physical access to
386: the whole closed path ${\cal C}_N$. To see this, let $\langle n(\theta_0)
387: \ket{n(\theta)}$ change sign at the angles $\theta_1,\ldots,\theta_K$
388: along ${\cal C}_N$. In the preferred gauge the effective Hamiltonian
389: operator for the nuclear motion in the $\theta$ direction reads
390: \begin{eqnarray}
391: H_{n} =
392: \bra{\phi_n (\theta;\theta_0)} H \ket{\phi_n (\theta;\theta_0)} =
393: \frac{1}{2} \Big[ p_{\theta} + {\cal A}_n (\theta;\theta_0) \Big]^2
394: + E_n (\theta) .
395: \label{eq:prefham}
396: \end{eqnarray}
397: Suppose $E_n (\theta)$ comprises an infinite potential
398: barrier over the angular range $\theta \in [\vartheta,\vartheta +
399: \Delta \vartheta]$, $0<\Delta \vartheta < 2\pi$, creating an inaccessible
400: part ${\cal C}_N (\Delta \vartheta)$ of ${\cal C}_N$ for the nuclear
401: motion. Then, it is consistent with the boundary condition
402: $\chi(\vartheta) = \chi (\vartheta + \Delta \vartheta) = 0$ to absorb
403: the vector potential ${\cal A}_n$ into the phase of the nuclear factor.
404:
405: \begin{figure}[ht!]
406: \begin{center}
407: \includegraphics[width=8 cm]{longuet.eps}
408: \end{center}
409: \caption{Single line ${\cal C}^{\perp}: s \in [0,1] \rightarrow q_N (s)$
410: in nuclear configuration space for which ${\cal A} (q_N (s);q_N (s_0)) =
411: -\pi \delta (q_N - q_N(s))$, for $0 \leq s < s' <1$, and ${\cal A}
412: (q_N (s);q_N (s_0)) =0$, for $s' < s \leq 1$. For any closed path C that
413: cross ${\cal C}^{\perp}$, the electronic eigenvector picks up a sign
414: implying a degeneracy on any surface that has C as boundary. For any
415: closed path C$'$ that does not cross ${\cal C}^{\perp}$, no such sign
416: change occur. From Longuet-Higgins theorem \cite{longuet75} follows
417: that $q_N(s')$ is a point of electronic degeneracy.}
418: \label{fig:degeneracy}
419: \end{figure}
420:
421: Now, what is the origin of a nontrivial gauge invariant reference
422: section? In part this can be answered in terms of the following
423: relationship between the gauge invariant reference section and
424: degeneracy points where two or more electronic potential energy
425: surfaces cross (for an analysis of the related connection between
426: degeneracies and singularities of the electronic eigenvectors, see
427: \cite{liyin90}). Let $q_N$ be the internal nuclear coordinates and
428: consider a line ${\cal C}^{\perp}: s \in [0,1] \rightarrow q_N (s)$ in
429: nuclear configuration space for which ${\cal A} (q_N (s);q_N (s_0)) =
430: -\pi \delta (q_N - q_N(s))$, for $0 \leq s < s' <1$, and ${\cal A}
431: (q_N (s);q_N (s_0)) =0$, for $s' < s \leq 1$, see
432: Fig. \ref{fig:degeneracy}. Now, any path C in the Born-Oppenheimer
433: regime that starts and ends at the reference point $q_N (s_0)$ and
434: cross ${\cal C}^{\perp}$ once, must be associated with a sign change
435: of the electronic eigenvector and thus must enclose at least one point
436: of degeneracy on any surface that has C as boundary, as implied by
437: the Longuet-Higgins theorem \cite{longuet75}. On the other hand,
438: there is no such sign change originating from ${\cal C}^{\perp}$
439: for a closed path C$'$ that does not cross ${\cal C}^{\perp}$,
440: which implies that there must be one less degeneracy enclosed by
441: such a path than by C. Thus, $q_N (s')$ is a degeneracy point.
442:
443: To illustrate the gauge invariant reference section for MAB, let us
444: revisit the linear $+$ quadratic $E\otimes \varepsilon$ Jahn-Teller
445: effect, which is known to exhibit a nontrivial MAB structure. There,
446: the symmetry induced degeneracy of two electronic states ($E$) is
447: lifted by their interaction with a doubly degenerate vibrational mode
448: ($\varepsilon$). In the vicinity of the degeneracy point at the symmetric
449: nuclear configuration, this may be modeled by the vibronic Hamiltonian
450: \cite{zwanziger87}
451: \begin{eqnarray}
452: H = \frac{1}{2} p_{r}^{2} + \frac{1}{2r^{2}}
453: p_{\theta}^{2} + \frac{1}{2} r^{2} + \Delta {\cal E} (r,\theta)
454: \big( \cos \alpha (r,\theta) \sigma_x + \sin \alpha (r,\theta)
455: \sigma_z \big) .
456: \label{eq:Exeham}
457: \end{eqnarray}
458: Here, $(r,\theta)$ are polar coordinates of the vibrational
459: mode, $(p_{r},p_{\theta})$ the corresponding canonical
460: momenta, $k\geq 0$ and $g\geq 0$ are the linear and quadratic
461: vibronic coupling strength, respectively. $\Delta {\cal E}$
462: and $\alpha$ are given by
463: \begin{eqnarray}
464: \Delta {\cal E} (r,\theta) & = & \sqrt{k^2r^2 + kgr^3\cos 3\theta +
465: \frac{1}{4}g^2r^4} ,
466: \nonumber \\
467: \Delta {\cal E} (r,\theta) e^{i\alpha (r,\theta)} & = & kr \cos \theta +
468: \frac{1}{2} gr^2 \cos 2\theta + i \big( kr \sin \theta -
469: \frac{1}{2} gr^2 \sin 2\theta \big) .
470: \end{eqnarray}
471: The electronic degrees of freedom are described by Pauli operators
472: defined in terms of the diabatic electronic states
473: $|0\rangle$ and $|1\rangle$ as $\sigma_{x} = |0 \rangle \langle 1| +
474: |1 \rangle \langle 0|$, $\sigma_{y} = -i|0 \rangle \langle 1| + i|1
475: \rangle \langle 0|$, and $\sigma_{z} = |0 \rangle \langle 0| - |1
476: \rangle \langle 1|$. Diagonalizing the electronic part of $H$
477: yields the electronic eigenvectors
478: \begin{eqnarray}
479: \ket{+(\alpha)} & = & \cos \frac{\alpha}{2} \ket{0} +
480: \sin \frac{\alpha}{2} \ket{1} ,
481: \nonumber \\
482: \ket{-(\alpha)} & = & -\sin \frac{\alpha}{2} \ket{0} +
483: \cos \frac{\alpha}{2} \ket{1}
484: \end{eqnarray}
485: with corresponding energies $E_{\pm} = \frac{1}{2} r^2 \pm \Delta
486: {\cal E} (r,\theta)$ and where we have put $\alpha \equiv \alpha
487: (r,\theta)$ for brevity. The phase $\alpha$ is undefined only if
488: $\Delta {\cal E} (r,\theta) = 0$, corresponding to the degeneracy
489: points at $r=0$ and $r=2k/g$ for $\theta=\pi/3,\pi,5\pi/3$.
490:
491: \begin{figure}[ht!]
492: \begin{center}
493: \includegraphics[width=8 cm]{jt.eps}
494: \end{center}
495: \caption{Lines of sign change and degeneracy points in the linear $+$
496: quadratic $E\otimes \varepsilon$ Jahn-Teller model for any of the two
497: Born-Oppenheimer states. Degeneracies are indicated by small circles;
498: they occur at the origin and at $r = 2k/g$ for angles $\theta =
499: \pi /3, \ \pi, \ 5\pi/3$. The electronic reference state is chosen
500: at an arbitrary point along the dashed-dotted line ($\theta = 0$).
501: The thick continuous lines correspond to points where $\bra{\pm
502: (\alpha_0)} \pm (\alpha) \rangle$ goes through zero and changes sign.
503: These lines of sign change end either at infinity or at a degeneracy.
504: For a circular closed path with $r>2k/g$, there is an even number of
505: jumps for this path, and consequently no MAB effect for the
506: corresponding nuclear motion along the path. On the other hand, for
507: any circular closed path with $r < 2k/g$ there is a single node at
508: $\theta = \pi$, assigning a nontrivial MAB effect.}
509: \label{fig:nodes}
510: \end{figure}
511:
512: To calculate the gauge invariant reference section, we may choose
513: $\theta_0 =0$ so that $\alpha_0 \equiv \alpha (r_0,\theta_0) =0$,
514: independent of $r_0$. In this case, we obtain
515: \begin{eqnarray}
516: \bra{\pm (\alpha_0)} \pm (\alpha) \rangle =
517: \cos \frac{\alpha}{2} .
518: \end{eqnarray}
519: This vanishes for $\alpha = \pi$ corresponding to
520: \begin{eqnarray}
521: kr \sin \theta - \frac{1}{2} gr^2 \sin 2\theta & = & 0 ,
522: \nonumber \\
523: kr \cos \theta + \frac{1}{2} gr^2 \cos 2\theta & < & 0 ,
524: \end{eqnarray}
525: which have the solutions $\theta' = \pi$ for $r < 2k/g$ and $\theta' =
526: \pm \arccos [k/(gr)]$ for $r>2k/g$. Since the solutions for $r<2k/g$
527: are independent of $r$, they constitute a radial line whose
528: end-points are the electronic degeneracies at the origin and at
529: $(r,\theta') = (2k/g,\pi)$, see Fig. \ref{fig:nodes}. If $2k/g <r
530: \rightarrow \infty$, we obtain the limit angles $\theta' = \pm \pi /2$
531: at the infinity. On the other hand, if $r \rightarrow 2k/g^{+}$,
532: then the lines of sign change terminate at $\theta' = \pm \pi /3$,
533: which are the remaining electronic degeneracies, see Fig. \ref{fig:nodes}.
534:
535: Next, let us consider the gauge invariant reference section and
536: the corresponding vector potential. For $r<2k/g$, we have
537: \begin{eqnarray}
538: \ket{\phi_{\pm}} & = &
539: \exp \Big( i\pi h_s (\theta -\pi) \Big) \ket{\pm (\alpha)} ,
540: \nonumber \\
541: {\cal A}_{\pm} (\theta;0) & = & -\pi \delta (\theta - \pi) ,
542: \end{eqnarray}
543: while for $r>2k/g$, we have
544: \begin{eqnarray}
545: \ket{\phi_{\pm}} & = &
546: \exp \Big( i\pi h_s (\theta -\arccos [k/(gr)])
547: \nonumber \\
548: & & +
549: i\pi h_s (\theta +\arccos [k/(gr)]) \Big)
550: \ket{\pm (\alpha)} ,
551: \nonumber \\
552: {\cal A}_{\pm} (\theta;0) & = &
553: -\pi \delta (\theta -\arccos [k/(gr)])
554: -\pi \delta (\theta +\arccos [k/(gr)]).
555: \end{eqnarray}
556:
557: \section{Aharonov-Casher analogue}
558: It was demonstrated in \cite{sjoqvist02} that for the linear or
559: the quadratic $E\otimes \varepsilon$ Jahn-Teller system, there is a
560: close analogy with the Aharonov-Casher (AC) phase effect \cite{aharonov84}
561: for an electrically neutral spin$-\frac{1}{2}$ particle encircling a
562: line of charge. From this perspective, one may regard the presence
563: (absence) of MAB effect in the linear (quadratic) case as a nontrivial
564: (trivial) $\pi$ ($2\pi$) AC phase shift. Here, we extend this idea to
565: the linear $+$ quadratic case.
566:
567: In brief, the AC effect may occur when an electrically
568: neutral spin$-\frac{1}{2}$ particle carrying a magnetic dipole moment
569: $\mu$ encircles a straight line of charge. Under the condition that
570: the spin is parallel to the charged line, the particle acquires along
571: the closed path $\partial S$ the phase shift ($\hbar = 1$)
572: \begin{eqnarray}
573: \gamma_{{\textrm{\footnotesize AC}}} =
574: \frac{\mu}{c^2} \oint_{\partial S}
575: \big( \hat{{\bf n}} \times {\bf E} \big) \cdot d{\bf r} =
576: \frac{\mu}{c^2} \int \! \! \! \int_S \nabla \cdot {\bf E} \ dS =
577: \frac{\mu \lambda}{c^2\epsilon_0}
578: \end{eqnarray}
579: with $c$ the speed of light, $\hat{{\bf n}}$ the direction of the
580: dipole, ${\bf E}$ the electric field, $d{\bf r}$ a line element along
581: this path, $S$ any surface with $\partial S$ as boundary, $\lambda$
582: the enclosed charge density, and $\epsilon_0$ the electric vacuum
583: permittivity.
584:
585: To demonstrate the promised analogy between the MAB effect in the
586: $E\otimes \varepsilon$ Jahn-Teller system and the AC effect, we may
587: use the apparent spin analogy in terms of which we may notice that the
588: electronic Hamiltonian in Eq. (\ref{eq:Exeham}) describes a
589: spin$-\frac{1}{2}$ under influence of an effective magnetic field that
590: rotates around the $\sigma_y$ axis by the angle $\alpha
591: (r,\theta)$. We expect the electronic Hamiltonian be fixed (possibly
592: up to an unimportant $r$ and $\theta$ dependent scale factor) in an
593: internal molecular frame that co-moves with this rotation. The
594: essential point here is that, contrary to the usual case of a
595: spin$-\frac{1}{2}$ in a rotating external magnetic field, the rotation
596: angle $\alpha$ of this Jahn-Teller system depends upon the internal
597: variables $r$ and $\theta$. This has the consequence that the
598: vibronic Hamiltonian in the co-moving frame, which reads (omitting the
599: inessential $\frac{1}{2}p_r^2$ and $\frac{1}{2} r^2$ terms)
600: \begin{eqnarray}
601: H' = U^{\dagger} H U =
602: \frac{1}{2r^{2}} \left[ p_{\theta} - \frac{1}{2} \partial_{\theta}
603: \alpha (r,\theta) \sigma_{y} \right]^{2} +
604: \Delta {\cal E} (r,\theta) \sigma_{z}
605: \end{eqnarray}
606: with $U = \exp [ -i \alpha (r,\theta) \sigma_{y}/2 ]$ the unitary
607: spin rotation operator, contains a nontrivial modification of the
608: nuclear kinetic energy operator. Indeed, by using $H'$ and the
609: Heisenberg picture, this modification in turn affects the equations
610: of motion for the electronic variables, which read
611: \begin{eqnarray}
612: \dot{\sigma}_{x} & = & -\partial_{\theta} \alpha (r,\theta)
613: \dot{\theta} \ \sigma_z - 2\Delta {\cal E} (r,\theta) \ \sigma_y ,
614: \nonumber \\
615: \dot{\sigma}_{y} & = & 2\Delta {\cal E} (r,\theta) \ \sigma_x ,
616: \nonumber \\
617: \dot{\sigma}_{z} & = & \partial_{\theta} \alpha (r,\theta)
618: \dot{\theta} \ \sigma_x ,
619: \label{eq:eqm}
620: \end{eqnarray}
621: where $r^{2} \dot{\theta} = p_{\theta} - \frac{1}{2}
622: \partial_{\theta} \alpha (r,\theta) \sigma_{y}$. The
623: Born-Oppenheimer regime is characterized by the condition
624: $\partial_{\theta} \alpha (r,\theta) |\dot{\theta}| \ll \Delta
625: {\cal E}(r,\theta)$ that apparently breaks down when
626: $\Delta {\cal E}(r,\theta)$ is very small, which happens close to
627: the electronic degeneracies. From Eq. (\ref{eq:eqm}), it follows
628: that the electronic motion describes the local torque due to an
629: effective magnetic field ${\bf B}_{\textrm{\footnotesize eff}} =
630: -\partial_{\theta} \alpha (r,\theta) \dot{\theta} \, {\bf e}_y +
631: 2\Delta {\cal E} (r,\theta) \, {\bf e}_z$ seen by the electronic
632: variables in the rotating frame. The large $z$ component of
633: ${\bf B}_{\textrm{\footnotesize eff}}$ depends only on the energy
634: difference $\Delta {\cal E}(r,\theta)$ between the two electronic
635: states and is thus irrelevant to MAB. On the other hand, the small
636: $y$ component corresponds exactly to the MAB effect and gives rise
637: to a ${\mbox{\boldmath $\sigma$}} \cdot ({\bf v} \times
638: {\bf E}_{\textrm{\footnotesize eff}})$ term for the
639: electronic variables in the co-moving frame in the $x-z$
640: plane. Explicitly, we have
641: \begin{equation}
642: {\bf E}_{\textrm{\footnotesize eff}} =
643: \frac{\partial_{\theta} \alpha (r,\theta)}{2r} \ {\bf e}_{r} +
644: E_{\theta} (r,\theta) \, {\bf e}_{\theta} ,
645: \end{equation}
646: where we have left out the explicit form of the $\theta$ component of
647: ${\bf E}_{\textrm{\footnotesize eff}}$, as it does not contribute to
648: the MAB phase effect. The effect of this field is equivalent to that
649: of an AC system that consists of four charged lines in the $y$
650: direction sitting at the four conical intersections at $r=0$ and
651: $r=2k/g$ for $\theta = \pi/3 , \pi , 5\pi/3$, all of which with the
652: charge per unit length being proportional to $\frac{1}{2}$. Thus, the
653: MAB effect for the $E\otimes \varepsilon$ Jahn-Teller system resembles
654: exactly that of the AC effect for an electrically neutral
655: spin$-\frac{1}{2}$ particle encircling a certain configuration of
656: charged lines perpendicular to the plane of motion.
657:
658: Note that the phase shift $\gamma_{{\textrm{\footnotesize AC}}}$
659: only depends upon the enclosed charge, but is independent of the shape
660: of the dipole's path. On the other hand, the dipole feels the gauge
661: invariant electric field ${\bf E}$, which defines a preferred gauge in
662: terms of the gauge invariant effective vector potential $(\mu /c^2)
663: \hat{{\bf n}} \times {\bf E}$ and which causes a nontrivial, essentially
664: local and nontopological autocorrelation among the spin variables, as
665: demonstrated in \cite{peshkin95}. From this perspective, we believe
666: that the present analogy between the AC and MAB effects further
667: strengthens the local and nontopological interpretation of MAB
668: for this Jahn-Teller system.
669:
670: \section{Conclusions}
671: Arguably one of the most intriguing discoveries of the second
672: half of the 20$th$ century was that of a nonlocal and topological
673: interference effect for a charged particle moving around a magnetic
674: flux line, made by Aharonov and Bohm (AB) \cite{aharonov59}.
675: Analogues of this remarkable effect have since then been found,
676: such as the Aharonov-Casher (AC) effect \cite{aharonov84} and
677: its Maxwell dual, the so-called He-McKellar-Wilkens (HMW) effect
678: \cite{he93,wilkens94}. Common to these analogue effects are that
679: they only occur under certain restrictions on some additional degree
680: of freedom: the spin direction for AC and the direction of an electric
681: dipole for HMW. These additional restrictions turn out to make these
682: effects essentially different from the standard AB, in that they do
683: not obey all the properties for being nonlocal and topological.
684:
685: Does the molecular Aharonov-Bohm (MAB) effect share the fate of these
686: other analogue effects? This question was examined quite recently by
687: the present author \cite{sjoqvist02}, in the particular case of the
688: $E\otimes \varepsilon$ Jahn-Teller system. In identifying the
689: restrictions for this system, which are the conditions of attaining
690: the Born-Oppenheimer regime and the electronic state being in one of
691: two instantaneous energy eigenstates, this issue can be addressed very
692: much along the line of the other analogue effects. The outcome was:
693: the MAB effect for this system is neither nonlocal nor topological in
694: the sense of the standard AB effect. In this paper, this result has
695: been generalized to any molecular system with real electronic
696: Hamiltonian and where a nontrivial MAB effect shows up. In addition, a
697: notion of preferred gauge for such systems, defined by an effective
698: vector potential whose open path integral is the open path Berry phase
699: for the electronic motion, has been suggested.
700:
701: One of the early motivations preceding the analysis in
702: \cite{sjoqvist02} was to examine whether there is a relation
703: between the MAB effect in the $E\otimes \varepsilon$ Jahn-Teller
704: system and the AC effect, based upon the simple observation that the
705: original vibronic Hamiltonian for this Jahn-Teller system resembles
706: exactly that of an electrically neutral spin$-\frac{1}{2}$ in a
707: certain field configuration. Indeed, by transforming to a molecular
708: frame that co-moves with the nuclear pseudorotation, this analogue was
709: made explicit in \cite{sjoqvist02} in the case of either linear or
710: quadratic coupling. In the present paper, this result has been
711: extended to the linear $+$ quadratic case, leading to an anisotropic
712: effective electric field originating from four charged lines sitting
713: at the four conical intersections in nuclear configuration space.
714:
715: \section*{Acknowledgments}
716: I wish to express my deep gratitude to Prof. Osvaldo Goscinski,
717: especially for introducing me to the subject of geometric phases and
718: how they appear in molecular systems, but also for numerous
719: discussions and collaboration during the past 10 years. Osvaldo's
720: great impact on my scientific thinking makes it a very special honour
721: to dedicate this paper on his 65$th$ birthday. I also wish to thank
722: Mauritz Andersson, Henrik Carlsen, Marie Ericsson, Gonzalo Garc\'{\i}a
723: de Polavieja, Magnus Hedstr\"om, and Niklas Johansson for discussions
724: and collaboration over the years on issues related to the present
725: paper. This work was supported by the Swedish Research Council.
726: \begin{thebibliography}{99}
727: \bibitem{longuet58} H.C. Longuet-Higgins, U. \"{O}pik,
728: M.H.L. Pryce and R.A. Sack,
729: Proc. Roy. Soc. London Ser. A {\bf 244}, 1 (1958).
730: \bibitem{longuet61} H.C. Longuet-Higgins,
731: Adv. Spectr. {\bf 2}, 429 (1961).
732: \bibitem{herzberg63} G. Herzberg and H.C. Longuet-Higgins,
733: Disc. Frad. Soc. {\bf 35}, 77 (1963).
734: \bibitem{longuet75} H.C. Longuet-Higgins,
735: Proc. Roy. Soc. London Ser. A {\bf 344}, 147 (1975).
736: \bibitem{berry84} M.V. Berry,
737: Proc. Roy. Soc. London Ser. A {\bf 392}, 45 (1984).
738: \bibitem{kendrick97} B. Kendrick,
739: Phys. Rev. Lett. {\bf 79}, 2431 (1997).
740: \bibitem{vonbusch98} H. von Busch, V. Dev, H.-A. Eckel, S. Kasahara,
741: J. Wang, W. Demtr\"{o}der, P. Sebald, and W. Meyer,
742: Phys. Rev. Lett. {\bf 81}, 4584 (1998).
743: \bibitem{kupperman93} A. Kupperman and Y.M. Wu,
744: Chem. Phys. Lett. {\bf 205}, 577 (1993).
745: \bibitem{kendrick96} B. Kendrick and R.T. Pack,
746: J. Chem. Phys. {\bf 104}, 7475 (1996).
747: \bibitem{adhikari00} S. Adhikari and G.D. Billing,
748: Chem. Phys. {\bf 259}, 149 (2000).
749: \bibitem{sjoqvist94} E. Sj\"{o}qvist and O. Goscinski,
750: Chem. Phys. {\bf 186}, 17 (1994).
751: \bibitem{ham87} F.S. Ham,
752: Phys. Rev. Lett. {\bf 58}, 725 (1987).
753: \bibitem{ham90} F.S. Ham,
754: J. Phys.: Condens. Matter {\bf 2}, 1163 (1990).
755: \bibitem{rios96} P. De Los Rios, N. Manini, and E. Tosatti,
756: Phys. Rev. B {\bf 54}, 7157 (1996).
757: \bibitem{moate96} C.P. Moate, M.C.M. O'Brien, J.L. Dunn, C.A. Bates,
758: Y.M. Liu, and V.Z. Polinger,
759: Phys. Rev. Lett. {\bf 77}, 4362 (1996).
760: \bibitem{stone76} A.J. Stone,
761: Proc. Roy. Soc. London Ser. A {\bf 351}, 141 (1976).
762: \bibitem{varandas79} A.J.C. Varandas, J. Tennyson, and J.N. Murrell,
763: Chem. Phys. Lett. {\bf 61}, 431 (1979).
764: \bibitem{xantheas90} S. Xantheas, S.T. Elbert, and K. Ruedenberg,
765: J. Chem. Phys. {\bf 93}, 7519 (1990).
766: \bibitem{ceotto00} M. Ceotto and F.A. Gianturco,
767: J. Chem. Phys. {\bf 112}, 5820 (2000).
768: \bibitem{johansson03} N. Johansson and E. Sj\"{o}qvist,
769: Phys. Rev. Lett. {\bf 92}, 060406 (2004).
770: \bibitem{mead79} C.A. Mead and D.G. Truhlar,
771: J. Chem. Phys. {\bf 70}, 2284 (1979);
772: \bibitem{mead80} C.A. Mead,
773: Chem. Phys. {\bf 49}, 23 (1980); {\it Ibid.} {\bf 49}, 33 (1980).
774: \bibitem{aharonov59} Y. Aharonov and D. Bohm,
775: Phys. Rev. {\bf 115}, 485 (1959).
776: \bibitem{sjoqvist02} E. Sj\"{o}qvist,
777: Phys. Rev. Lett. {\bf 89}, 210401 (2002).
778: \bibitem{pati95a} A.K. Pati,
779: J. Phys. A {\bf 28}, 2087 (1995).
780: \bibitem{pati95b} A.K. Pati,
781: Phys. Rev. A {\bf 52}, 2576 (1995).
782: \bibitem{samuel88} J. Samuel and R. Bhandari,
783: Phys. Rev. Lett. {\bf 60}, 2339 (1988).
784: \bibitem{mukunda93} N. Mukunda and R. Simon,
785: Ann. Phys. (N.Y.) {\bf 288}, 205 (1993).
786: \bibitem{peshkin95} M. Peshkin and H.J. Lipkin,
787: Phys. Rev. Lett. {\bf 74}, 2847 (1995).
788: \bibitem{sjoqvist97} E. Sj\"{o}qvist and M. Hedstr\"{o}m,
789: Phys. Rev. A {\bf 56}, 3417 (1997).
790: \bibitem{garcia98} G. Garc\'{\i}a de Polavieja and E. Sj\"{o}qvist,
791: Am. J. Phys. {\bf 66}, 431 (1998).
792: \bibitem{englman99} R. Englman and A. Yahalom,
793: Phys. Rev. A {\bf 60}, 1802 (1999);
794: \bibitem{englman00} R. Englman, A. Yahalom, and M. Baer,
795: Eur. Phys. J. D {\bf 8}, 1 (2000).
796: \bibitem{aharonov84} Y. Aharonov and A. Casher,
797: Phys. Rev. Lett. {\bf 53}, 319 (1984).
798: \bibitem{larsson03} P. Larsson and E. Sj\"{o}qvist,
799: Phys. Rev. A {\bf 68}, 042109 (2003).
800: \bibitem{wagh98} A.G. Wagh, V.C. Rakhecha, P. Fischer, and A. Ioffe,
801: Phys. Rev. Lett. {\bf 81}, 1992 (1998).
802: \bibitem{sjoqvist01} E. Sj\"{o}qvist,
803: Phys. Lett. A {\bf 286}, 4 (2001).
804: \bibitem{liyin90} L. Yin and O. Goscinski,
805: Int. J. Quantum Chem. {\bf 37}, 249 (1990).
806: \bibitem{zwanziger87} J.W. Zwanziger and E.R. Grant,
807: J. Chem. Phys. {\bf 87}, 2954 (1987).
808: \bibitem{he93} X.G. He and B.H.J. McKellar,
809: Phys. Rev. A {\bf 47}, 3424 (1993).
810: \bibitem{wilkens94} M. Wilkens,
811: Phys. Rev. Lett. {\bf 72}, 5 (1994).
812: \end{thebibliography}
813: \end{document}
814:
815:
816:
817:
818:
819:
820:
821:
822:
823:
824:
825:
826:
827:
828:
829:
830:
831:
832: