1: \documentclass[aps,amssymb, showpacs]{revtex4}
2: \usepackage[english]{babel}
3: \usepackage{bm}
4: \usepackage{epsfig}
5: \topmargin = -1.5cm
6: %\oddsidemargin = 10pt
7: \hoffset= 0.0cm
8: \textwidth=480pt
9: \textheight=680pt
10:
11: \newcommand{\beq}{\begin{equation}}
12: \newcommand{\eeq}{\end{equation}}
13: \newcommand{\ben}{\begin{eqnarray}}
14: \newcommand{\een}{\end{eqnarray}}
15: \newcommand{\non}{\nonumber \\}
16: \newcommand{\noi}{\noindent}
17: \newcommand{\p}{\partial}
18: \newcommand{\h}{\hbar}
19:
20:
21:
22: \begin{document}
23:
24: \title{Quantum Mechanics of a Charged Particle in an Axial Magnetic Field}
25: \author{Ashok Das}
26: \affiliation{Department of Physics and Astronomy, University of
27: Rochester, Rochester, NY, USA}
28: \author{J. Frenkel and S. H. Pereira}
29: \affiliation{Instituto de F\'{i}sica, Universidade de S\~{a}o Paulo, S\~{a}o Paulo, Brazil}
30: \author{J. C. Taylor}
31: \affiliation{DAMTP, Centre for Mathematical Sciences, Cambridge University, Cambridge, UK}
32:
33: \bigskip
34: \begin{abstract}
35: We study some aspects of the quantum theory of a charged particle
36: moving in a time-independent, uni-directional magnetic field.
37: When the field is uniform, we make a few clarifying remarks on the use
38: of angular momentum eigenstates and momentum eigenstates with the
39: diamagnetism of a free electron gas as an example. When the field is
40: non-uniform but weakly varying, we discuss both perturbative and
41: non-perturbative methods for studying a quantum mechanical system. As an application, we derive the
42: quantized energy levels of a charged particle in a Helmholtz coil,
43: which go over to the usual Landau levels in the limit of a uniform field.
44: \end{abstract}
45:
46: \pacs{03.65.Ge}
47:
48: \maketitle
49:
50: \section{Introduction}
51:
52: Although the quantum theory of a charged particle moving in a
53: uni-directional constant magnetic field has been studied over many decades (for some
54: text-book treatments, see for example \cite{magnetic}),
55: we would like to make some clarifying remarks which we believe are
56: useful. In studying this problem, one has to make a choice of the
57: gauge. Different gauge choices, of course, lead to a trivial phase
58: redefinition of the wave function. Moreover, corresponding to specific
59: gauges, one chooses either a wave function which is an eigenstate of
60: the momentum operator (which has continuous eigenvalues) or that of
61: the angular momentum operator (which has discrete eigenvalues). In
62: section II we point out that such a choice is unessential and derive
63: an explicit relation between the two bases. This is then used in the
64: treatement of diamagnetism of a free electron gas with angular
65: momentum eigenstates. In section III, we study the case when the magnetic field has a weak
66: $z$-dependence (where the $z$-axis is in the direction of the field).
67: Then, there must be also a weak radial field and we work under
68: conditions where it is consistent to treat this component of the field
69: perturbatively.
70: We find the energy eigenvalues for certain models of the $z$-dependence.
71: These models could approximate the field between two parallel
72: circular currents. In section IV, we discuss the form of the
73: wave-function for large quantum numbers associated with the
74: $z$-motion, using a non-perturbative
75: approach in a slowly varying field. As an application, we study in section V the
76: structure of the quantized energy levels of a charged particle in a
77: Helmholtz coil, which are specified by two quantum numbers. The first
78: is associated with the motion in the $(x,y)$ plane, while the second
79: characterizes the motion in the $z$-direction. This energy spectrum
80: may be regarded as a generalization of the Landau levels
81: \cite{magnetic} to the case of a slowly varying magnetic field. A
82: brief conclusion is presented in section VI.
83:
84:
85:
86: \def\L{ \Lambda}
87: \def\p{\partial}
88: \def\f{\phi}
89: \def\m{\mu}
90: \def\n{\nu}
91: \def\l{\langle}
92: \def\r{\rho}
93: \def\D{{\cal D}}
94: \def\q{\psi}
95: \def\g{\gamma}
96: \def\B{{\cal B}}
97: \def\E{{\cal E}}
98:
99:
100: \section{Uniform magnetic field}
101:
102: We take the field to be in the $z$-direction, and to have $z$-component $B$.
103: We take the particle's charge and mass to be $e$ and $m$.
104: Two gauges for the vector potential which are commonly used are
105: \begin{equation}
106: {\bf A}= (-By/2,Bx/2,0) \label{a1}
107: \end{equation}
108: and
109: \begin{equation}
110: {\bf A}= (-By, 0, 0).
111: \end{equation}
112: But the difference between these two gauges can only be trivial,
113: since the wave-functions just differ by the phase factor
114: \begin{equation}
115: \omega=\exp[i eBxy/(2c\hbar )].
116: \end{equation}
117: Therefore we might just as well choose (\ref{a1}).
118:
119: For shortness, we define
120: \begin{equation}
121: \B = eB/c
122: \end{equation}
123: The Hamiltonian in gauge (\ref{a1}) is
124: $$
125: H={1\over 2m}[(p_x-\B y/2)^2+(p_y+\B x/2)^2+p_z^2]
126: $$
127: \begin{equation}
128: ={1\over 2m}[p_x^2+p_y^2+p_z^2+\B^2(x^2+y^2)/4 +\B L]=H'+p_z^2/2m,\label{a5}
129: \end{equation}
130: where
131: \begin{equation}
132: p_x=-i\hbar {\p \over \p x},~~p_y=-i\hbar {\p \over \p y},~~
133: p_z=-i\hbar {\p \over \p z},
134: \end{equation}
135:
136: \begin{equation}
137: L=xp_y-yp_x\,.\end{equation}
138:
139: We define also
140: \begin{equation}
141: P_x=p_x+ \B y/2,~~ P_y=p_y- \B x/2. \label{a8}
142: \end{equation}
143: Then the operators $P_x,P_y$ and $L$ each commute with $H$ and $p_z$. But they don't
144: commute with each other:
145: \begin{equation}
146: [P_x,P_y]= i\hbar \B,~ ~[L,P_x]= i\hbar P_y,~ ~[L,P_y]=- i\hbar P_x
147: \end{equation}
148:
149: We may choose the wave-function $\psi (x,y,z)$ to be an eigenfunction
150: of $p_z$ with eigenvalue $\hbar k_z$:
151: \begin{equation}
152: \psi(x,y,z)=e^{ik_z z}\phi(x,y).
153: \end{equation}
154: \begin{equation}
155: H\psi=e^{ik_z z} [H'+\hbar^2k_z^2/(2m)]\phi,
156: \end{equation}
157: and from now on we will be concerned with $H'$ and $\phi(x,y)$.
158:
159: We may choose $\phi$ to be an eigenfunction of any one of the three operators
160: $P_x,P_y,L$. We discuss the two cases (i) eigenfunction of $P_x$ (by
161: rotational invariance we can equally well choose it to be an
162: eigenstates of $P_y$) and (ii)
163: eigenfunction of $L$.
164: These wave-functions may be written in a more compact form if we define a magnetic length $l$ by
165: \begin{equation}
166: l= \sqrt{\hbar/\B}.
167: \end{equation}
168:
169: In case (i), we have
170: \begin{equation}
171: \phi_{k_x,n}(x,y)= {1\over \sqrt{2\pi }}e^{ik_x x}e^{-i
172: xy/2l^2}\xi_n(y), \label{a13}
173: \end{equation}
174: where $\hbar k_x$ is the eigenvalue of $P_x$, and $\xi_n$ satisfies
175: \begin{equation}
176: {\hbar^2\over 2m}\left[ - {\p^2\over \p y^2}+ \{(y/l^2)-k_x\}^2 \right]\xi_n(y)=E'\xi_n(y),\label{a14}
177: \end{equation}
178: $E'$ being the eigenvalue of $H'$. This equation describes a simple
179: harmonic oscillator whose equilibrium point is shifted, and
180: the eigenfunctions have the form
181: \begin{equation}
182: \xi_n(y)=l^{-1/2}u_n[(y-\eta)/l],
183: \end{equation}
184: where $n$ is a positive integer,
185: \begin{equation}
186: \eta=
187: l^2k_x\, ,~~E'={\hbar \B \over m}(n+{1\over 2} )\label{a16}
188: \end{equation}
189: and $u_n(w)$ is the real, normalized solution of
190: \begin{equation}
191: \left[ -{\p ^2 \over \p w^2}+w^2 \right]u_n(w)=(2n+1)u_n(w).
192: \end{equation}
193:
194:
195: Next we take the case (ii), where we use energy eigenfunctions which are
196: also eigenfunctions of angular momentum
197: $L$ with eigenvalue $\hbar M$. We denote these normalized
198: wave-functions by
199: \begin{equation}
200: \zeta_{n,M}(x,y).\label{a18}
201: \end{equation}
202:
203:
204: Eigenfunctions of type (i), in equation (\ref{a13}), are non-normalizable
205: and have a continuous degeneracy, whereas those of type (ii) in
206: (\ref{a18}) are normalized and have a discrete degeneracy.
207: Nevertheless, it should be possible to express each type in terms of the other.
208: To this end, we consider a superposition of the form
209: \begin{equation}
210: \chi_{n,f}=\int dk_x f(k_x)\phi_{k_x,n}(x,y)\label{a19}
211: \end{equation}
212: where $f$ is some function to be determined and
213: \begin{equation}
214: \int dx dy \chi^*_{n',f}\chi_{n,f}=\delta_{nn'}\int dk_x |f(k_x)|^2 .
215: \end{equation}
216:
217:
218: In order to make (\ref{a19}) an eigenstate of $L$, we require
219: $$
220: \hbar M\chi_{n,f}=L\chi_{n,f}
221: $$
222: $$
223: =-i\hbar(2\pi l)^{-1/2}\int dk_x f(k_x) \left(x{\p \over \p y}-y{\p \over \p x}
224: \right)
225: $$
226: \begin{equation}
227: \times
228: \exp[ik_x x -i xy/2l^2]u_n[(y/l)-lk_x].\label{a25}
229: \end{equation}
230: We convert $x$ and $x^2$ into derivatives of $e^{ik_x x}$
231: and integrate by parts to put these derivatives onto $f$ and $u$.
232: In this way, we find that equation (\ref{a25}) is satisfied provided $f$
233: obeys
234: \begin{equation}
235: -{ 1\over l^2}{d^2 f\over d k_x^2}+l^2k_x^2 f = [2(n-M)+1]f.
236: \end{equation}
237: This equation is consistent, and its normalised solution is just
238: \begin{equation}
239: f(k_x)=l^{1/2} u_{n-M}[lk_x ].
240: \end{equation}
241: Thus, finally, the required eigenfunction of $L$ is determined to be
242: \begin{equation}
243: \zeta_{n,M}(x,y)
244: =l^{1/2}
245: \int dk_x u_{n-M}(lk_x)\phi_{k_x,n}(x,y).\label{a28}
246: \end{equation}
247:
248: Using the ortho-normality of the $u_n$ coefficients in (\ref{a28}), we can invert this
249: equation to find the momentum eigenfunctions in terms of the angular
250: momentum eigenfunctions:
251: \begin{equation}
252: \phi_{k_x,n}(x,y)=l^{1/2}\sum_{M=-\infty}^{n}u_{n-M}(lk_x)\zeta_{n,M}(x,y).
253: \end{equation}
254:
255:
256: The text-book treatment (see for example \cite{diamagnetism}) of the diamagnetism of a free electron gas is usually
257: formulated in terms of the eigenfunctions of momentum (\ref{a13}). The sample
258: is considered to have a finite size with dimensions $S_x, S_y$
259: in the $x$- and $y$-directions, in which case states contribute only for
260: \begin{equation}
261: |\eta | <S_y.
262: \end{equation}
263: So the number of states with a given energy and value of $k_z$
264: is
265: \begin{equation}
266: {\B S_x S_y \over 2 \pi \hbar}={S_xS_y \over 2\pi l^2}\label{a42}
267: \end{equation}
268: It ought to be possible to carry out the analysis using the angular momentum
269: eigen-functions (\ref{a18}). In that case, it is natural to consider a
270: cylindrical sample, axis along the $z$ axis, with radius $R$.
271: The asymptotic form of the wave function (\ref{a18}), for large $\rho$, is proportional to
272: \begin{equation}
273: e^{iM\phi} \rho^{2n-M} \exp(-\rho^2/4l^2),
274: \end{equation}
275: and this has its maximum at
276: \begin{equation}
277: \rho_{max}=\sqrt{2(2n-M)}\,\,l.
278: \end{equation}
279: Thus we expect states to contribute for
280: \begin{equation}
281: 2n-M < {R^2 \over 2l^2}.
282: \end{equation}
283:
284: Then, for given $n$, the range of $M$ is
285: \begin{equation}
286: 2n-{R^2\over 2l^2}\leq M \leq n,
287: \end{equation}
288: so that the number of possible values of $M$ is
289: \begin{equation}
290: {R^2 \over 2l^2}-n. \label{a47}
291: \end{equation}
292: The relevant values of $n$ are of order $(mkTl^2/\hbar^2)$ which is small
293: compared to $(R^2 /l^2)$ for typical values of $B$, $R$ and $T$ (temperature). Thus (\ref{a47}) gives a factor
294: \begin{equation}
295: {\pi R^2\over 2 \pi l^2 }
296: \end{equation}
297: which is proportional to the area just as in (\ref{a42}).
298:
299: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
300:
301:
302: \section{Weakly varying magnetic field}
303:
304: We now allow the magnetic field to vary slowly with $z$, but we restrict
305: ourselves to motion near the $z$-axis. Along the axis, we use an expansion
306: \begin{equation}
307: B_z(z)=B_0[1+b_1 (z/a)^2+b_2 (z/a)^4+...]\label{a30}
308: \end{equation}
309: where $a$ is a characteristic length.
310:
311: A realistic case would be a pair of two similar current loops, each perpendicular to
312: the common $z$-axis and of radius $a$, separated by a distance
313: $2d$. This system, known as the Helmholtz coil, is shown in Fig. 1.
314:
315: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
316: \begin{figure}[htb]
317: \begin{center}
318: %\begin{tabular}{|c|}
319: %\hline
320: %\begin{minipage}[b]{10cm}
321: %\makebox[10cm][c]{\includegraphics[scale=0.5]{bobina.eps}}
322: \epsfig{file=bobina.eps, scale=0.5}\\
323: %\end{minipage}
324: %\hline
325: %\end{tabular}
326: \end{center}
327: \caption{Schematic illustration of a Helmholtz coil.}
328: \end{figure}
329: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
330:
331:
332:
333: Near the origin the field has an expansion
334: of the above form, where $B_0$ is determined by the current
335: carried by the loops and (in terms of $r=(d/a)^2$)
336: \begin{equation}
337: b_1= {3(4r-1) \over 2(r+1)^2},~~~b_2={15(1-12r+8r^2) \over 8(r+1)^4}.\label{a31}
338: \end{equation}
339:
340: Off the axis, the magnetic field must have some radial component.
341: Up to second order (that is the $b_1$ term),
342: we take the vector potential to be
343: \begin{equation}
344: {\bf A}={1 \over 2}B_0[1+b_1(z^2-\rho^2/4)/a^2](-y,x,0). \label{a32}
345: \end{equation}
346: This form follows by rotational symmetry (if the field is generated by
347: currents in circles about the axis), together with the gauge
348: choice $\mbox{\boldmath$\nabla$}\cdot{\bf A}$$=0$ and Maxwell's equation
349: $\nabla^2 {\bf A}=0$.
350:
351: As the leading approximation, we keep just the $b_1 z^2$ term in (\ref{a30})
352: (we shall see that the $\rho^2$ term is effectively smaller).
353: We seek an energy eigenfunction of the Hamiltonian (\ref{a5}), with $\B$
354: now $z$-dependent, of the form
355: \begin{equation}
356: W(z)\Phi(x,y).
357: \end{equation}
358: Then the equation for $W$ is (we now define $l$ to be $\sqrt{\hbar/\B_0}$)
359: \begin{equation}
360: - {d^2W\over dz^2}+(2n+1){b_1z^2\over a^2l^2}W={2m\Delta E\over \hbar^2}W,
361: \end{equation}
362: where the second term on the left comes from $E'$ in (\ref{a14}) when $B$ is
363: $z$-dependent.
364: This is just the equation for a
365: simple harmonic oscillator and, for $k_z=0$, the energy associated with the $z$ motion is
366: \begin{equation}
367: \Delta E={\hbar^2 \over mal}\sqrt{ b_1 (2n+1)}(N+{1\over 2}).
368: \end{equation}
369:
370: We can now treat the $z^4$ term in (\ref{a30}) and the $\rho^2$ term in (\ref{a32})
371: as perturbations. These terms are of the same order. This is because
372: the transverse size of the wave-function is of order $l$ whereas the
373: longitudinal size is of order $\sqrt{la}$. Note that, as an example,
374: for a magnetic flux density of 1 tesla, $l=2.5 \times 10^{-8} m$.
375:
376: Treating the $z^4$ term by perturbation theory, we find an additional
377: energy
378: \begin{equation}
379: \Delta E' = {3\hbar^2 \over 4ma^2}{b_2\over b_1(2n+1)}(1+2N+2N^2).\label{a36}
380: \end{equation}
381: Note that this is independent
382: of $B$ and usually very small if $a$ is macroscopic. However, for large values of $N$, the contributions from the $b_2$ term and higher
383: terms in (\ref{a30}) may be significant.
384:
385: We also treat by perturbation theory the $\rho^2$ term in (\ref{a32}).
386: To this end
387: we require, inserting the $\rho^2$ term in (\ref{a32}) into the Hamiltonian (\ref{a5}),
388: \begin{equation}
389: \Delta E''=-{\hbar b_1\over 16ma^2 l^2}\int dx dy\phi^*_{n,M} (2L\rho^2+\hbar\rho^4/l^2)\phi_{n,M}.
390: \end{equation}
391: These expectation values can be worked out using:
392: \begin{equation}
393: \int dx dy \psi^*\rho^2 \psi= (2n-M+1)l^2,\label{a38}
394: \end{equation}
395: \begin{equation}
396: \int dx dy \psi^* \rho^4 \psi =[(2n-M+1)^2+2n(n-M)+2n-M+1]l^4.\label{a39}
397: \end{equation}
398: As a result, we have
399: \begin{equation}
400: \Delta E''=-{\hbar^2b_1\over 16 ma^2}[6n(n+1)-M(M+1)+2(1-nM)].\label{a40}
401: \end{equation}
402: This is independent of $B$ and, in general, is much smaller than
403: (\ref{a36}) for large values of $N$.
404:
405:
406: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
407:
408:
409: \section{Non-perturbative approach to a slowly varying field}
410:
411: We have seen that when the quantum number associated with the
412: $z$-motion is large, $N >> n$, the dependence of $B_z$ upon $\rho$
413: may be neglected in first approximation. In this case, the vector
414: potential may be written as:
415: \begin{equation}
416: {\bf A}= {1 \over 2} B_z(z)(-y,x,0) \,.\label{a49}
417: \eeq
418: \noi
419: This leads to the Schr\"odinger equation:
420: \beq
421: {\p^2 \psi \over \p \rho^2} + {1 \over \rho} {\p \psi \over \p \rho} -
422: \bigg({M^2 \over \rho^2}+ {e B_z M \over \hbar c} + {e^2 B_z^2
423: \over 4\hbar^2 c^2}\rho^2 \bigg)\psi + {\p^2 \psi \over \p z^2} = -
424: {2m \over \hbar^2} E \psi \,, \label{a50}
425: \eeq
426: \noi
427: where we have taken the energy eigenfunction to be also an
428: eigenfunction of the angular momentum $L_z$ with eigenvalue $\hbar
429: M$. Let us consider a solution of the form:
430: \beq
431: \psi(\phi, \rho, z) = F(\phi, \rho, z)f(z) \label{a51}
432: \eeq
433: \noi
434: and assume, since the field is slowly varying, that:
435: \beq
436: \bigg| {\p F \over \p z} \bigg| << \bigg| {\p F \over \p \rho} \bigg|
437: \,. \label{a52}
438: \eeq
439: \noi
440: Then, the Schr\"odinger equation (\ref{a50}) takes the form:
441: \beq
442: {1 \over F} \bigg[ {\p^2 F \over \p \rho^2} + {1 \over \rho}{\p F
443: \over \p \rho}\bigg] - {M^2 \over \rho^2} - {eB_zM \over \hbar c}-
444: {e^2 B_z^2 \over 4 \h^2 c^2}\rho^2 = - \bigg[ {1 \over f}{d^2f \over d z^2}
445: + {2m \over \h^2} E \bigg]\,. \label{a53}
446: \eeq
447: \noi
448: The right hand side of (\ref{a53}) depends on $z$ only, so that we
449: have:
450: \beq
451: {d^2f \over d z^2} + {2m \over \h^2} E f = {2m \over \h^2} E'(z) f
452: \label{a54}
453: \eeq
454: \noi
455: and
456: \beq
457: {\p^2 F \over \p \rho^2} + {1 \over \rho}{\p F \over \p \rho} - \bigg(
458: {M^2 \over \rho^2} + {eB_z M \over \h c} + {e^2 B_z^2 \rho^2 \over 4
459: \h^2 c^2}\bigg)F = -{2m \over \h^2} E'(z) F \,. \label{a55}
460: \eeq
461: \noi
462: But we know how to solve (\ref{a55}), in which $z$ is just a
463: parameter. Comparing with (\ref{a16}), we see that:
464: \beq
465: E'(z) = {e \h \over mc}(n+ {1 \over 2}) B_z(z) \,. \label{a56}
466: \eeq
467: \noi
468: Thus, (\ref{a54}) may be written as:
469: \beq
470: {d^2 f \over d z^2}+{2m\over \h^2} \bigg[ E - {e \h (2n+1)\over 2mc}
471: B_z(z)\bigg] f = 0 \,, \label{a57}
472: \eeq
473: \noi
474: which is just an ordinary differential equation.
475:
476: We can now check the
477: assumption (\ref{a52}). The solutions of (\ref{a55}) have the form
478: (\ref{a18}), namely:
479: \beq
480: F(\phi, \rho, z) = \zeta_{n,M}(\phi, \sqrt{{eB_z\over 2\h c}}\,\rho)
481: \,. \label{a58}
482: \eeq
483: \noi
484: Hence, the condition (\ref{a52}) requires that:
485: \beq
486: {\rho\over B_z}{dB_z\over dz} << 1 \,. \label{a59}
487: \eeq
488: \noi
489: Although this cannot be satisfied for all $\rho$, we note that the
490: solution (\ref{a58}) contains the factor:
491: \beq
492: \exp[-{eB_z\over 4\h c}\rho^2]\,,\label{a60}
493: \eeq
494: \noi
495: so that typical values of $\rho$ are of order $\sqrt{\h c/eB_z}$. Thus, we expect (\ref{a59}) to be valid for:
496: \beq
497: \sqrt{{\h c \over e}}{1\over B_z^{3/2}}{dB_z\over dz}<< 1\,. \label{a61}
498: \eeq
499: \noi
500: This can be satisfied provided we take $B_z$ to be varying slowly
501: enough. For instance, if $B_z(z)$ has the form (\ref{a30}) with $|z|$ of order $a$, then the left hand side of (\ref{a61}) is of order $l/a$, which is very small for typical values of $B_0$ and $a$.
502:
503: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
504:
505:
506: \section{Quantized energy levels in a Helmholtz coil}
507:
508: A simple example of a magnetic mirror is provided by the pair of
509: current loops
510: shown in Fig. 1. Then, as can be seen from equations (\ref{a30}) and
511: (\ref{a31}) for $2d > a$, the magnetic field can increase enough in
512: the region near the current loops, so that a charged particle may
513: eventually be reflected out of this region towards the center of the
514: coil. The classical restoring force which may confine the particle along the
515: $z$-axis is provided by the $\rho$-component of the magnetic field:
516: \beq
517: {\bf B}(z) = -{1\over 2}\rho {\p B_z\over \p z}\hat{\mbox{\boldmath$\rho$}} + B_z
518: \hat{{\bf z}} \,. \label{a62}
519: \eeq
520: which can be derived from (\ref{a49}) using ${\bf B} =
521: \mbox{\boldmath$\nabla$}\times{\bf A}$.
522:
523: In this section, we will discuss the quantized energy levels of a
524: charged particle in a Helmholtz coil. To this end, we will analyse the basic equation
525: (\ref{a57}), where
526: \beq
527: B_z(z) = {1 \over 2} B_0 (d^2 + a^2)^{3/2}\bigg\{ {1 \over[(z+d)^2 +
528: a^2]^{3/2}} + {1 \over[(z-d)^2 +
529: a^2]^{3/2}} \bigg\} \,. \label{a63}
530: \eeq
531: The general behavior of this field, for $2d>a$, is shown in Fig. 2.
532: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
533: \begin{figure}[htb]
534: \begin{center}
535: %\begin{tabular}{|c|}
536: %\hline
537: \epsfig{file=kelly01.eps, scale=0.3}\\
538: %\hline
539: %\end{tabular}
540: \end{center}
541: \caption{Pattern of the field $B_z(z)$ in a Helmholtz coil. Circles
542: indicate the classical turning points.}
543: \end{figure}
544: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
545:
546: Since (\ref{a57}) was derived under the assumption that $N>>1$, our
547: analysis will be more appropriate in the semiclassical regime. This may
548: be studied conveniently in the WKB approximation, which is
549: particularly useful since we are dealing with a slowly varying
550: potential:
551: \beq
552: V(z) = { e\h \over 2mc}(2n+1)B_z(z)\,.\label{a64}
553: \eeq
554: One of the most interesting results arising from the WKB method is a
555: semiclassical estimate for the quantized energy levels in a
556: potential. Matching the WKB wave function at each of the classical
557: turning points, which are determined by the
558: relation $E=V(z)$, leads to the Bohr-Sommerfeld quantization
559: condition:
560: \beq
561: \int_{-Z_+}^{Z_+}\sqrt{2m[E - V(z)]}dz = (N+{1\over 2})\pi \h\,,
562: \label{a65}
563: \eeq
564: where the classical turning points for bound motion, $\pm Z_+$, are
565: situated inside the well as shown in Fig. 2. (We neglect, in first approximation, the very small
566: probability of tunneling through the potential barrier.)
567:
568: In general, equation (\ref{a65}) is rather complicated and can
569: only be solved numerically. However, when $d/a$ is somewhat larger
570: than $1/2$, it may be solved in closed form, since in this case $z/a$
571: will be effectively of order $1/2$ or smaller. Then, we can expand the
572: potential $V(z)$ [see (\ref{a63}) and (\ref{a64})] up to terms which
573: are of quartic order in $z/a$, as shown in equations (\ref{a30}) and
574: (\ref{a31}). In this approximation, the quantization condition
575: (\ref{a65}) may be written in the form:
576: \beq
577: \int_{-Z_+}^{Z_+} \sqrt{\E -1 - b_1 \big({z\over a}\big)^2 -
578: b_2\big({z\over a}\big)^4}dz = \pi \sqrt{{c\h \over (2n+1)e B_0}}
579: (N+{1\over 2})\,, \label{a66}
580: \eeq
581: where the dimensionless parameter $\E$ is defined by:
582: \beq
583: \E = {2mc\over (2n+1)e\h B_0}E\,.\label{a67}
584: \eeq
585: We can now determine explicity the positions of the
586: classical turning points, which are given by:
587: \beq
588: Z_\pm = a\bigg[ {-b_1 \pm \sqrt{b_1^2 + 4b_2 (\E -1)}\over
589: 2b_2}\bigg]^{1/2}\,, \label{a68}
590: \eeq
591: where $\pm Z_-$ are the turning points for unbound motion, which are
592: situated outside the well as shown in Fig. 2. Then, the integral appearing in (\ref{a66}) can be
593: evaluated in closed form in terms of the complete elliptic integrals
594: of the first (${\bf F}$) and second (${\bf E}$) kind \cite{tables}, with the result:
595: \beq
596: {2\sqrt{-b_2}\over 3a^2} Z_- \bigg[ \big(Z_+^2 - Z_-^2\big) {\bf
597: F}\bigg({Z_+\over Z_-}\bigg) +\big(Z_+^2 + Z_-^2\big) {\bf
598: E}\bigg({Z_+\over Z_-}\bigg)\bigg]=\pi \sqrt{{c\h\over
599: (2n+1)eB_0}}(N+{1\over 2})\,. \label{a69}
600: \eeq
601: This is a transcendental equation which determines implicitly the
602: energy in terms of the quantum number $N$. The energies of the bound states have an upper bound $E^{max}$ which corresponds
603: to the maximum value of the potential. At this energy, $Z_+ = Z_-$, so
604: that using (\ref{a67}) and (\ref{a68}) we obtain:
605: \beq
606: E^{max}_n= {e\h B_0\over 2mc}(2n+1)\bigg(1-{b_1^2\over
607: 4b_2}\bigg)\,. \label{a70}
608: \eeq
609:
610: When $Z_+ = Z_-$, the equation
611: (\ref{a69}) simplifies considerably and fixes the
612: maximum value of the quantum number $N$, which is given by the relation:
613: \beq
614: N_{max}+ {1\over 2}= {\sqrt{2(2n+1)}\over 3\pi}{b_1^{3/2}\over |b_2|}{a\over
615: l}\,, \label{a71}
616: \eeq
617: where $l=\sqrt{c\h / e B_0}$ is the magnetic length. Hence, $N_{max}$ is
618: in general a very large number, which is in acordance with our
619: previous assumption.
620:
621: A more explicit relation giving $E$ as a function of $N$ can be obtained
622: by solving numerically the equation (\ref{a69}). As an example, the numerical
623: solution for $n=0$ and $d/a=3/5$ is shown in Fig. 3. Here, the
624: numerical values of $\E^{max}=1.028$ and $N_{max}=1.12\times 10^6$
625: are in good agreement with the corresponding results obtained from the
626: closed form expressions (\ref{a70}) and (\ref{a71}). One can see that the numerical solution
627: can be fitted reasonably well, for large quantum numbers, by
628: a simple phenomenological form like $N^{5/6}$.
629:
630:
631: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
632: \begin{figure}[htb]
633: \begin{center}
634: %\begin{tabular}{|c|}
635: %\hline
636: \epsfig{file=figure3.eps, scale=0.3}\\
637: %\hline
638: %\end{tabular}
639: \end{center}
640: \caption{Numerical solution (crossed) of Eq. (\ref{a69}) for the
641: energy spectrum and plot (solid) of the form $N^{5/6}$ for large $N$.}
642: \end{figure}
643: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
644: Finally, we note from (\ref{a30}) and (\ref{a31}) that for $d/a=1/2$, $b_1=0$ and $b_2=-144/125$. Then, the
645: system of the two current loops provides a practically uniform field in the
646: central region of the Helmholtz coil. In this case, (\ref{a70})
647: reduces to the well known expression for the Landau levels which occur
648: in a constant field. Therefore, one may regard the
649: quantized energies described by (\ref{a69}) and (\ref{a70}),
650: as an extension of the Landau levels to the case of a slowly varying magnetic field.
651:
652: \section{Conclusion}
653:
654: In this paper, we have discussed several aspects concerning the
655: quantum behavior of a charged particle in a static magnetic field. We
656: have treated the issue of the relation between the choice of gauge
657: and the choice of the diagonal operators which commute with the
658: Hamiltonian of the system. We have also developed some approaches
659: which may be useful for physical applications in a slowly varying
660: magnetic field. These methods have been applied to study the quantized
661: energy levels of a charged particle in a weakly varying field. Such a
662: field may be present, for example, in a magnetic mirror like the
663: Helmholtz coil. We have shown that this energy spectrum represents an
664: interesting extension of the well known Landau levels which occur in an
665: uniform magnetic field.
666:
667:
668: \vspace{0.5cm}
669:
670: This work was supported in part by US DOE Grant number DE-FG
671: 02-91ER40685, by CAPES, CNPq and FAPESP, Brazil.
672:
673: \section*{Bibliography}
674: \begin{enumerate}
675: \bibitem{magnetic}
676: L. D. Landau and E. M. Lifshitz, {\it Quantum Mechanics: The Non-Relativistic Theory}, Butterworth-Heinemann, 1981;
677: L. E. Ballentine, {\it Quantum Mechanics}, Prentice Hall, 1990;
678: S. Gasiorowicz, {\it Quantum Physics} (2nd edition), Wiley, 1996;
679: E. Merzbacher, {\it Quantum Mechanics} (3rd edition), Wiley, 1998;
680: J. Schwinger, {\it Quantum Mechanics}, Springer, 2001.
681:
682: \bibitem{diamagnetism}
683: R. E. Peierls, {\it Quantum Theory of Solids}, Oxford University Press, 2001.
684:
685: \bibitem{tables}
686: I. S. Gradshteyn and M. Ryzhik, {\it Tables of Integral Series and
687: Products}, Academic Press, New York, 1980.
688:
689: \end{enumerate}
690: \end{document}
691:
692: