1: \documentstyle[aps,preprint]{revtex}
2: %\usepackage{graphics,epsf}
3: \newcommand{\bea}{\begin{eqnarray}}
4: \newcommand{\beq}{\begin{equation}}
5: \newcommand{\eea}{\end{eqnarray}}
6: \newcommand{\eeq}{\end{equation}}
7: \begin{document}
8: \title{Semiclassical analysis of a two-electron quantum
9: dot in a magnetic field: dimensional phenomena}
10: \author{R.G. Nazmitdinov $^{1,2}$, N. S. Simonovi\'c $^1$,
11: and Jan M. Rost $^1$}
12: \address{
13: $^1$ Max-Planck-Institut f\"ur Physik komplexer
14: Systeme, D-01187 Dresden, Germany\\
15: $^2$ Bogoliubov Laboratory of Theoretical Physics,
16: Joint Institute for Nuclear Research, 141980 Dubna, Russia}
17:
18: \maketitle
19:
20: \begin{abstract}
21: While the dynamics for three-dimensional axially symmetric
22: two-electron quantum dots with parabolic confinement potentials
23: is in general non-separable we have found an exact separability
24: with three quantum numbers for specific values of the magnetic field.
25: Furthermore, it is shown that the magnetic properties such as the
26: magnetic moment and the susceptibility are sensitive
27: to the presence and strength of a vertical confinement.
28: Using a semiclassical approach the calculation of
29: the eigenvalues reduces to simple quadratures providing a transparent and
30: almost analytical quantization of the quantum dot energy levels which
31: differ from the exact energies only by a few percent.
32: \end{abstract}
33:
34: \vskip 0.3cm
35:
36: PACS numbers: 73.21.La, 03.65.Sq, 75.75.+a, 05.45.Mt
37:
38: \vskip 1cm
39:
40:
41: Current nanofabrication technology allows one to control the size and
42: shape of quantum dots \cite{Tur,Bim,Tap}. Due to the confinement
43: of the electrons in all three
44: spatial directions the energy spectrum is quantized creating
45: excellent experimental and theoretical opportunities
46: to study {\it controlled} single-particle and collective
47: dynamics at the atomic scale.
48: For example, depending on the experimental setup, the spectrum of
49: a quantum dot displays shell structure \cite{Kouw,Mac} or follows
50: predictions of random matrix theory (for a review see \cite{Al}).
51: Furthermore, it becomes possible to trace the transition
52: from a quantum mechanical to an almost classical regime.
53:
54: Few-electron quantum dots have attracted special attention \cite{Kouw,Mak},
55: since they may provide a natural realization of a quantum bit \cite{Tur}.
56: The simplest quantum dot with the essential features of more complex
57: systems contains two electrons.
58: Experimental data, including
59: transport measurements \cite{Su} and spin oscillations in
60: the ground state under a perpendicular magnetic field \cite{Ash},
61: have been explained quantum mechanically as a result of the interplay
62: between the two-dimensional lateral confinement potential,
63: electron correlations and
64: the magnetic field \cite{Mer,Din}.
65: While in these experiments effects of the
66: third spatial dimension are somewhat hidden, they naturally come into play
67: with a tilted magnetic field \cite{MHP}. However, even with a perpendicular
68: magnetic field the vertical confinement has at least
69: two important consequences which will be worked
70: out in the following:
71: first, it changes the magnetic moment and
72: susceptibility with respect to the 2D results, second, the generically
73: non-separable 3D dynamics
74: becomes separable for certain values of
75: the magnetic field. We will use a
76: semiclassical description
77: which offers a simple and accurate approach to
78: explore the effects of dimensionality in quantum dots.
79: In contrast to a circular (2D) two-electron quantum dot
80: whose classical dynamics is always separable and therefore regular, the corresponding
81: 3D-system with axial symmetry is in general a
82: non-integrable problem with typical features of mixed dynamics
83: (regular/chaotic).
84:
85: The Hamiltonian for the 3D two-electron quantum dot reads
86:
87: \beq
88: \label{ham}
89: H = \sum_{j=1}^2{\Bigg\{} \frac{1}{2m^*}
90: ({\bf p}_j \!-\!\!\frac{e}{c} {\bf A}_j)^2+\frac{m^*}{2}\left[
91: \omega _0^2(x_{j}^2\!+\!y_{j}^2)+\omega _z^2z_{j}^2 \right]\! {\Bigg\}}
92: + V_C + H_{\rm spin},
93: \eeq
94: where $V_C = \alpha/{|{\bf r}_1 - {\bf r}_2|}$ is the Coulomb energy
95: ($\alpha = e^2/(4\pi \varepsilon \varepsilon_0)$)
96: and
97: $H_{\rm spin} = g^*({\bf s}_1 + {\bf s}_2) \!\cdot\! {\bf B}$ describes
98: the Zeeman energy.
99: Here $m^*$ and $ g^*$ are the effective electron mass and
100: $g$-factor, respectively, and $\varepsilon$ is the dielectric constant.
101: The confining potential is approximated with a 3D axially-symmetric
102: harmonic oscillator
103: and $\hbar \omega_z \neq \hbar \omega_0$ are the energy scales of
104: confinement
105: in the $z$-direction and in the $xy$-plane, respectively.
106: For the typical voltage $\sim 1$ V applied to the gate, the
107: confining potential is some eV deep which is large compared
108: to the few meV of the confining frequency \cite{Bim,Tap}.
109: Hence, the electron
110: wave function is localized close to the minimum of the well which
111: always can be approximated by a parabolic potential.
112: In real samples the electron-electron interaction is usually
113: screened. However, the pure Coulomb interaction should suffice
114: to understand the main features of the system.
115: For the perpendicular magnetic field
116: $({\bf B}\parallel z)$ we choose a gauge described by the
117: vector ${\bf A} =[{\bf B}\times {\bf r}]/2 = \frac{1}{2}B(-y, x,0)$.
118: Introducing the relative and center-of-mass coordinates
119: ${\bf r} = {\bf r}_1 - {\bf r}_2$, $ {\bf R} =
120: \frac{1}{2}({\bf r}_1+{\bf r}_2)$,
121: the Hamiltonian, Eq.(\ref{ham}), can be separated into the
122: center-of-mass (CM) $H_{\rm CM}$ and relative-motion (RM)
123: $H_{\rm rel}$ terms: $H=H_{\rm CM}+H_{\rm rel}+H_{\rm spin}$.
124: The solution to the CM-Hamiltonian is well known \cite{Fock} and
125: the effect of the Zeeman energy has been discussed in \cite{Mer,Din}.
126: In the following we will concentrate on the dynamics of $H_{\rm rel}$.
127:
128: For our analysis it is convenient to use cylindrical {\it scaled}
129: coordinates, $\tilde\rho = \rho/l_0$, ${\tilde p}_{\rho} =
130: p_{\rho}l_0/\hbar$, $\tilde z = z/l_0$, $ {\tilde p}_{z} = p_z
131: l_0/\hbar$, where $l_{0}=(\hbar/\mu\omega_0)^{1/2}$ is the
132: characteristic length of the confinement potential with the reduced
133: mass $\mu = m^*/2$. The strength parameter $\alpha$ of the Coulomb
134: repulsion goes over to $\lambda = 2\alpha/(\hbar \omega_0 l_0)$.
135: Using the effective mass $m^*=0.067 m_e$, the dielectric constant
136: $\varepsilon=12$, which are typical for GaAs, and the confining
137: frequency $\hbar \omega_0$=3 meV, we obtain $\lambda \approx 3 $.
138: Hereafter, for the sake of simplicity, we drop
139: the tilde, i.e. for the scaled variables we use the same symbols as
140: before scaling.
141:
142: In these variables the Hamiltonian for the relative motion
143: takes a particular simple form (in units of $\hbar \omega_0$)
144: \beq
145: \epsilon \equiv \frac{H_{\rm rel}}{\hbar\omega_0} = \frac{1}{2}
146: \left[ p_\rho^2 + \frac{m^2}{\rho^2} + p_z^2 +
147: \left(\frac{\omega_\rho}{\omega_0}\right)^{\!\!2}\!\rho^2 +
148: \left(\frac{\omega_z}{\omega_0}\right)^{\!\!2}\!z^2
149: + \frac{\lambda}{\sqrt{\rho^2+z^2}} \right] - \frac{\omega_L}{\omega_0}\,m,
150: \label{relham}
151: \eeq
152: where $m = l_z/\hbar$, $\omega_L=eB/2m^*\!c$
153: is the Larmor frequency and
154: \begin{equation}
155: \label{frequ}
156: \omega_{\rho}=(\omega_{L}^{2}+\omega_{0}^{2})^{1/2}
157: \end{equation}
158: is the effective confinement frequency in the $\rho$-coordinate which
159: depends through $\omega_{L}$ on the magnetic field.
160:
161: Due to the cylindrical symmetry, the $z$-component $l_z \equiv p_\phi$
162: of the angular momentum is conserved and the motion in $\phi$ is
163: separated from the motion in the $(\rho, z)$-plane. Since the Coulomb
164: term couples the two coordinates, the problem is in
165: general non-integrable which is reflected in the Poincar\'e sections
166: shown in Fig.1 for increasing magnetic field. The chosen
167: ratio $\omega_z/\omega_0 = 3$ is of the same
168: order of magnitude as in the experiment \cite{MHP}. For $\omega_L
169: = 0$ and small values of $m$ the motion is mainly chaotic (see Fig.1a).
170: With the magnetic field the frequency of oscillations
171: along the $\rho$-coordinate can be controlled which leads to
172: qualitatively different dynamical situations (Fig.1b-d).
173: For equal effective confinement frequencies
174: $\omega_\rho^2=\omega_z^2$, the Hamiltonian Eq.(\ref{relham})
175: becomes separable in spherical coordinates and the dynamics is
176: integrable (Fig.1c).
177: For two other limiting cases, the dynamics is nearly integrable,
178: namely in the limit $m\to \infty$ and for $\omega_{z}\to \infty$. The
179: latter case represents a two-dimensional quantum dot,
180: classically, we have $p_{z}, z\to 0$ in this limit.
181:
182: The semiclassical quantization of the {\it circular 2D quantum
183: dot} is particularly simple since it reduces to a 1D WKB quantization
184: of the $\rho$-motion due to the separability of the problem. For given $m$ and $p_{z}=z=0$
185: the momentum $p_{\rho}$ determined from Eq.(\ref{relham}) enters the action
186: integral
187: \beq
188: I_{\rho} = \frac{\hbar}{2\pi}\oint p_{\rho}\,d\rho =
189: \frac{\hbar}{\pi}\int_{\rho_{\rm min}}^{\rho_{\rm max}} |p_\rho|\,d\rho,
190: \label{action}
191: \eeq
192: with the turning points $\rho_{\rm min}$, $\rho_{\rm max}$ as
193: the positive roots of
194: equation $p_\rho(\rho) = 0$. The WKB quantization conditions
195: \beq
196: I_{\rho}(\epsilon) = \hbar\,(n_{\rho} +
197: \hbox{$\frac{1}{2}$}), \quad n_\rho = 0,1,..., \quad m = 0,\pm 1,...
198: \eeq
199: determine the energy levels. For non-interacting electrons ($\lambda = 0$) the
200: analytical calculation of the action integral leads to the
201: (quantum mechanically exact) eigen-energies
202: \beq
203: \epsilon = \sqrt{1+\left(\frac{\omega_L}{\omega_0}\right)^{\!\!2}}\,
204: (2 n_\rho\! + \vert\,m\,\vert + 1) - \frac{\omega_L}{\omega_0}\,m,
205: \eeq
206: which are the well known Fock-Darwin energies \cite{Fock}.
207: For $\lambda \neq 0$, we calculate the action
208: integral Eq.(\ref{action}) numerically with a
209: few iterations to determine the quantum eigenvalues.
210:
211: The energy spectra for non-interacting and interacting
212: electrons are shown in Fig.2. In the interacting
213: case the semiclassical result, although not exact (the error is less
214: than $1\%$),
215: reproduces very well the quantum mechanical results
216: \cite{Din,Mer}.
217:
218: Turning now to
219: the {3D quantum dot} we have seen that the dynamics is separable
220: for
221: $\omega_z^2 = \omega_{\rho}^2 \equiv{\omega_L^*}^2+\omega_{0}^2 $
222: and the Hamiltonian Eq.~(\ref{relham}) in scaled spherical coordinates
223: takes the form
224: \beq
225: \epsilon = \frac{1}{2} \left\{
226: p_r^2 + \left(\frac{\omega_z}{\omega_0}\right)^{\!\!2}\!r^2 +
227: \frac{\lambda}{r} + \frac{({\bf l}/\hbar)^2}{r^2} \right\} -
228: \frac{\omega_L^*}{\omega_0}\,m\,.
229: \label{relsp}
230: \eeq
231: In this case the square of the total angular momentum ${\bf l}^2$ is
232: an additional integral of motion. Therefore, the classical dynamics
233: reduces again to a one-dimensional, radial problem. Using
234: Eq.(\ref{relsp}) and calculating the action integral for the
235: radial motion analogous to that in Eq.(\ref{action}) (i.e. with $r$
236: instead of $\rho$), we obtain the energy levels from the standard WKB
237: quantization conditions
238: \bea
239: &&I_r(\epsilon) = \hbar\,(n_r + \hbox{$\frac{1}{2}$}), \quad
240: \vert\,{\bf l}\,\vert = \hbar\,(l + \hbox{$\frac{1}{2}$}), \nonumber\\
241: &&n_r, l = 0, 1, ..., \quad m = 0, \pm 1, ..., \pm l.
242: \eea
243: Note that it is only the magnetic field which generates the spherical
244: symmetry of the problem and therefore its separability leading to
245: three good quantum numbers $n_r$, $l$ and $m$.
246:
247: In the general case of an {\it axially symmetric 3D quantum dot} we
248: have non-integrable motion and a semiclassical quantization is neither
249: straight forward nor does it give results which allow for a simple
250: understanding of the dynamics. Therefore, we make use of the fact
251: that in real samples the confining potential in the $z$-direction is
252: much stronger than in the $xy$-plane which allows us to analyze the 3D
253: non-integrable system with the
254: 'removal of resonances' method (RRM) \cite{LL}. To
255: lowest order the RRM consists of averaging the Hamilton function over
256: the fastest angle of the unperturbed motion $(\lambda=0)$ after
257: rewriting coordinates and momenta in terms of action-angle
258: variables $(J_{\rho}, J_z, \theta_{\rho}, \theta_z)$:
259: \bea
260: &&\!\!\!\!\rho^2 = \frac{\omega_0}{\omega_{\rho}}\left(2j_\rho+|m|-
261: 2\sqrt{j_{\rho}(j_{\rho}+|m|)}\cos 2\theta_{\rho}\right)\quad \quad, \\
262: &&\!\!\!\!z^2 = \frac{2j_z\omega_0}{\omega_z}\,\sin^2\!\theta_z \quad,
263: \eea
264: and $p_{\rho}={\dot{\rho}}$, $p_z={\dot{z}}$.
265: Here, $j_z=J_z/\hbar$ and $j_{\rho}=J_{\rho}/\hbar$.
266: If $\omega_z > \omega_\rho$ one averages over the angle
267: $\theta_z = \omega_zt$. As a result, the motion effectively
268: decouples into an unperturbed motion in the $z$-coordinate governed
269: by the potential
270: $(\omega_z / \omega_0)^2z^2/2$ and into the relative motion in the
271: $\rho$-coordinate governed by the effective potential
272: \beq
273: V_{\rm eff}(\rho, j_z) =
274: \frac{1}{2}\!\left(\frac{\omega_\rho}{\omega_{0}}\right)^{\!\!2}\!\rho^2
275: + \frac{m^2}{2\rho^2}
276: + \frac{\lambda}{\pi\rho}\,K\!\left(\!-2\frac{\omega_{0}}{\omega_z}
277: \frac{j_z}{\rho^2}\right),
278: \label{effham}
279: \eeq
280: where $K$ is the first elliptic integral.
281: Hence, the effective Hamiltonian reads
282: \beq
283: \epsilon = \frac{p_\rho^2}{2} + V_{\rm eff} -
284: \frac{\omega_L}{\omega_0}\,m
285: + \frac{\omega_z}{\omega_0}\,j_z.
286: \label{scaleffen}
287: \eeq
288: Applying a similar procedure as in the 2D case, we calculate the
289: action integral numerically. The momentum $p_\rho$ is determined from
290: Eq.(\ref{scaleffen}) and the turning points $\rho_{\rm min}$,
291: $\rho_{\rm max}$
292: are as usual the (positive) roots of the equation $p_\rho(\rho) = 0$.
293: Finally, the WKB-quantization conditions
294: \bea
295: && I_{\rho}(\epsilon) = \hbar\,(n_{\rho} + \hbox{$\frac{1}{2}$}), \quad
296: j_z = n_z + \hbox{$\frac{1}{2}$},\nonumber\\
297: && n_\rho, n_z = 0,1,2,..., \quad m = 0,\pm 1, \pm 2,...,
298: \eea
299: determine the energy levels.
300:
301: Comparing the exact results for eigen-energies for the spherical case
302: $\omega_z/\omega_\rho = 1$
303: we found good agreement even for large
304: values of the magnetic field (Fig.3a) although RRM is expected
305: to work best for $\omega_\rho/\omega_z <1$. Without magnetic field we
306: have $\omega_\rho/\omega_z =1/3$ which means that the motion in $z$ and $\rho$
307: approximately decouples justifying the widely used 2D approximation. This
308: is also reflected in the small difference between 2D and 3D results
309: (compare Fig.2b with Fig.3a at $\omega_{L}=0$).
310: Turning on the magnetic field increases the coupling of the dynamics in
311: $\rho$ and $z$ which allows the two electrons eventually to access
312: the full 3D space. As a consequence, the electrons can avoid each other
313: more effectively and the Coulomb interaction has a smaller effect on the
314: 3D spectrum than on the 2D spectrum which is most clearly visible for the
315: $m=0$ energies, see Fig.3a. We can understand this effect
316: quantitatively by averaging the elliptic integral in
317: Eq.(\ref{effham}) over the unperturbed ($\lambda =0$) motion in $\rho$.
318: It gives rise to an effective charge in the Coulomb interaction
319: $V_C\approx{\lambda_{\mathrm{ eff}}}/2{\rho}$, where
320: \beq
321: \lambda_{\mathrm{ eff}}=\frac{2\lambda}{\pi^2}\int_0^\pi\!\!
322: K\!\left(\!-\frac{\omega_{\rho}/\omega_z}
323: {1+|m|-{\sqrt{1+2|m|}}\cos{2\theta_\rho}}\right)\!d\theta_{\rho}
324: \label{efk}
325: \eeq
326: for $n_{\rho}=n_z=0\quad (j_ \rho=j_z=1/2)$.
327: The 3D energy quantized with this effective charge for the repulsion
328: is close to the full interaction (dotted line in Fig.3a).
329:
330: The effective charge $\lambda_{\mathrm{ eff}}/\lambda$ as a function
331: of $\omega_\rho/\omega_z$ for different $m$
332: is shown in Fig.3b.
333: The maximum repulsion at $\omega_\rho/\omega_z= 0$ corresponds
334: with $\omega_z\to\infty$ to the 2D case. The 3D case without magnetic
335: field starts for our parameters $\omega_\rho/\omega_z= 1/3$ at some
336: value $\lambda_{\mathrm{ eff}}/\lambda <1$ which decreases
337: further for increasing
338: $\omega_\rho/\omega_z$, i.e., increasing magnetic field.
339: This explains quantitatively through the effective charge the difference
340: of the effect of a magnetic field on a quantum spectrum in 2D and 3D cases.
341: However, this difference becomes weaker for larger $m$ as it is seen in
342: Fig.3b.
343:
344:
345: Although
346: the ground state as a function of the magnetic field is formed piecewise
347: by levels of {\it increasing} $m$ and alternating singlet-triplet character
348: (see Fig.3a) the magnetic properties of the ground state nevertheless
349: reveal the dimensional difference between 2D and 3D.
350: At temperature $T=0$ the dot is
351: in the ground state and the magnetic moment and the magnetic
352: susceptibility are defined by
353: $\mu_{\rm mag} = -\partial E_{\rm gr}/\partial B$
354: and $\chi = \partial \mu_{\rm mag} /\partial B$, respectively.
355: Both quantities exhibit discontinuities as a function of the magnetic
356: field due to the symmetry
357: changes (with respect to $m$ and spin).
358: We find that these discontinuities shift
359: when going from the 2D quantum dot to the 3D case as shown
360: in Fig.4.
361:
362: By relaxing the restriction of two dimensions for a quantum dot and
363: working in the physical three-dimensional space we have investigated
364: physical examples of non-integrable systems close to integrability.
365: For this situation the RRM method is naturally justified, since the
366: confining frequencies in quantum dots obey the condition $\hbar
367: \omega_z \gg \hbar\omega_0$. The WKB-approach provides a simple and
368: transparent way to calculate the spectrum of the 3D
369: two-electron quantum dot even for {\it small} values of the quantum
370: numbers. We have found
371: that at {\it specific} values of the magnetic field
372: $\omega_L^*=\sqrt{\omega_{z}^2 - \omega_{0}^2}$
373: an axially-symmetric quantum dot exhibits spherical symmetry and its
374: dynamics becomes completely separable with three integrals of motion
375: and three corresponding quantum numbers. We have shown that the
376: confinement in the $z$-direction, neglected in the 2D description of
377: quantum dots, does have an influence on the spectrum and consequently
378: also on magnetic properties of the dot. In fact, by changing the
379: confining frequency in the $z$-direction only slightly one can
380: increase or decrease the magnetic moment and the susceptibility, i.e.
381: one can control the magnetic properties of the two-electron quantum
382: dot.
383:
384: \begin{thebibliography}{99}
385: \bibitem{Tur}
386: R.~Turton,
387: {\it The Quantum Dot. A Journey into Future Microelectronics}
388: (Oxford University Press, New York, 1995)
389:
390: \bibitem{Bim}
391: D.~Bimberg, M.~Grundmann, and N.N.~Ledentsov,
392: {\it Quantum Dot Heterostructures}
393: (John Willey \& Sons, New York, 1998)
394: \bibitem{Tap}
395: T.~Chakraborty,
396: {\it Quantum Dots}
397: (Elsevier, Amsterdam, 1999)
398:
399: \bibitem{Kouw}
400: L.P.~Kouwenhoven, D.G.~Austing and T.~Tarucha,
401: Rep.Prog.Phys. {\bf 64}, 701 (2001)
402:
403: \bibitem{Mac}
404: M.~Macucci, K.~Hess, and G.J.~Iafrate,
405: Phys.\ Rev.\ B {\bf 48}, 17354 (1993);
406: J. Appl.\ Phys. {\bf 77}, 3267 (1995);
407: A.~Wojs and P.~Hawrylak,
408: Phys.Rev. B {\bf 53}, 10841 (1996);
409: W.D.~Heiss and R.G.~Nazmitdinov,
410: Phys.\ Lett.\ A {\bf 222}, 309 (1996);
411: Phys.Rev. {\bf B55}, 16310 (1997);
412: JETP Letters, {\bf 68}, 915 (1998)
413:
414: \bibitem{Al}
415: Y.~Alhassid, Rev.Mod.Phys. {\bf 72}, 895 (2000)
416:
417:
418:
419: \bibitem{Mak}
420: P.A.~Maksym, H.~Imamura, G.P.~Mallon and H.~Aoki,
421: J.Phys.: Condens. Matter {\bf 12}, R299 (2000)
422: \bibitem{Su}
423: Bo Su, V.J.~Goldman, J.E.~Cunningham,
424: Phys.Rev.B {\bf 46}, 7644 (1992)
425: \bibitem{Ash}
426: R.C.~Ashoori, H.L.~Stormer, J.S.~Weiner,
427: L.N.~Pfeiffer, K.W.~Baldwin and K.W.~West,
428: Phys. Rev. Lett., {\bf 71}, 613 (1993);
429: R.C.~Ashoori, Nature (London) {\bf 379}, 413 (1996)
430:
431: \bibitem{Mer}
432: U.~Merkt, J.~Huser and M.~Wagner,
433: Phys.Rev.B {\bf 43}, 7320 (1991);
434: M.~Wagner, U.~Merkt and A.V.~Chaplik,
435: Phys. Rev.B {\bf 45}, 1951 (1992)
436: \bibitem{Din}
437: M.~Dineykhan and R.G.~Nazmitdinov,
438: Phys. Rev.B {\bf 55}, 13707 (1997);
439: J. Phys.: Condens.Matter {\bf 11}, L83 (1999)
440: \bibitem{MHP}
441: B.~Meurer, D.~Heitmann, and K.~Ploog,
442: Phys.Rev. B {\bf 48}, 11488 (1993)
443: \bibitem{Fock}
444: V.~Fock, Z. Phys. {\bf 47}, 446 (1928);
445: C.G.~Darwin, Proc.Cambridge Philos.Soc. {\bf 27}, 86 (1930)
446:
447: \bibitem{LL}
448: A.J.~Lichtenberg and M.A.~Lieberman,
449: {\it Regular and Chaotic Dynamics}, Second Edition
450: (Springer-Verlag, New York, 1992)
451: \end{thebibliography}
452:
453: \vskip 1cm
454:
455: {\bf Figure captions:}
456: \vskip 0.2cm
457:
458:
459: {\bf Fig.1.} Poincar\'e surfaces of sections $z = 0$, $p_z > 0$
460: of the relative motion for
461: the axially-symmetric 3D two-electron quantum dot ($\omega_z/\omega_0 = 3$,
462: $\lambda = 3$, $m = 0$, $\epsilon = 5$) in the
463: magnetic field for: (a) $\omega_L =
464: 0$, (b) $\omega_L/\omega_0 = 2.5$, (c) $\omega_L/\omega_0 = \sqrt{8}$ and (d)
465: $\omega_L/\omega_0 = 3.3$. The section (c) indicates that for the corresponding
466: value of the magnetic field the system is integrable.
467:
468:
469: {\bf Fig.2.} The energy spectrum of a circular 2D quantum dot (in units
470: $\hbar\omega_0$) as a function of the ratio $\omega_L/\omega_0$ for
471: $n_\rho=0$ and $m = 0,...,9$ in the cases: (a) $\lambda=0$ and
472: (b) $\lambda=3$.
473:
474:
475:
476: {\bf Fig.3.} (a) The comparison between energy levels (in units
477: $\hbar\omega_0$) of the axially-symmetric 3D quantum dot with
478: $\omega_z/\omega_0 = 3$ and $\lambda=3$ for $n_\rho=n_z = 0$
479: and $m = 0,...,9$ obtained using the RRM (full lines) and
480: exact results for the spherical case (circles).
481: The inset shows a good agreement between the RRM and the exact
482: results. The dashed and dotted lines display the energy level with $m=0$
483: for the 2D and 3D cases with
484: $\lambda_{\mathrm{ eff}}$
485: at $\omega_0/\omega_z=0$ and $1/3$, respectively.
486: (b) The dependence of the effective strength
487: of the Coulomb interaction $\lambda_{\mathrm{ eff}}/\lambda$
488: on the ratio $\omega_{\rho}/\omega_z$.
489:
490:
491: {\bf Fig.4.} Magnetic moments $\mu_{\rm mag}$
492: (a) in the units of effective Bohr
493: magneton $\mu_B^* = (m_e/m^*)\,\mu_B$
494: and the magnetic susceptibility $\chi$ (b)
495: for the 2D (dashed lines) and 3D (full lines) cases
496: as a function of the magnetic field strength (in $\omega_L/\omega_0$-units).
497: We use the same parameters as in Figs.2,3.
498:
499: \end{document}
500:
501: