1: %% This document created by Scientific Word (R) Version 2.5
2:
3:
4: \documentclass[12pt]{article}
5: \usepackage{amssymb}
6:
7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8: \usepackage{sw20aip}
9:
10: %TCIDATA{TCIstyle=article/art2.lat,aip,article}
11:
12: %TCIDATA{Created=Mon Nov 28 15:21:59 2005}
13: %TCIDATA{LastRevised=Wed Sep 13 13:30:44 2006}
14: %TCIDATA{Language=American English}
15:
16: \input{tcilatex}
17: \begin{document}
18:
19: \date{2-24-2006}
20: \title{Coulomb oscillation in the hydrogen atom and molecule ion }
21: \author{Manfred Bucher \\
22: %EndAName
23: Physics Department, California State University, Fresno\\
24: Fresno, Califonia 93740-8031}
25: \maketitle
26:
27: \begin{abstract}
28: Semiclassical oscillation of the electron through the nucleus of the $H$
29: atom yields both the exact energy and the correct orbital angular momentum
30: for $l=0$ quantum states. Similarly, electron oscillation through the nuclei
31: of $H_{2}^{+}$ accounts for a stable molecule ion with energy close to the
32: quantum mechanical solution. The small discrepancy arises from the neglect
33: of the electron's wave nature.
34:
35: PACS numbers: 03.65.Sq, 31.10.+z, 31.20.Pv
36: \end{abstract}
37:
38: \section{INTRODUCTION}
39:
40: Two of the reasons why the old quantum theory of Bohr and Sommerfeld was
41: abandoned in the mid 1920s were the theory's failure to give the correct
42: multiplet structure of the hydrogen atom and the stability of the hydrogen
43: molecule ion, $H_{2}^{+}$.\cite{1} The old vector model of angular momentum%
44: \cite{2} gave, for a given principal quantum number $n$, sublevels with
45: angular quantum numbers $l=1,2,...,n$.\cite{3} Spectroscopic evidence,
46: however, showed multiplicities of spectral line splitting in a magnetic
47: field according to $l=0,1,...,n-1$. Max Jammer, in his review,\cite{2} notes
48: that ``the old quantum theory could never resolve this inconsistency.''
49:
50: A treatment of the hydrogen molecule ion with Sommerfeld's quantization
51: conditions had been Wolfgang Pauli's doctoral thesis of 1922.\cite{4} Pauli
52: found its molecular binding energy to be positive (non-binding)---contrary
53: to (later) experimental findings. Martin Gutzwiller\cite{5} thinks that
54: ``the solution of this problem can be rated, with only slight exaggeration,
55: as the most important in quantum mechanics, because if an energy level with
56: a [more] negative value [than of a free hydrogen atom] can be found, then
57: the chemical bond between two protons by a single electron has been
58: explained.''
59:
60: Both dilemmas of the old quantum theory can be resolved, though, with a
61: single extension: an oscillation of the electron through the nucleus
62: (nuclei) of the atom (molecule). In essence this solution is already
63: formally included in Sommerfeld's theory of the hydrogen atom\cite{6} but
64: was explicitly omitted by Sommerfeld and his school as being \textit{%
65: unphysical}. The case in point, obtained with Sommerfeld's quantization
66: conditions for radial and angular motion, is a quantum state with zero
67: angular action, characterized by an angular quantum number $l$ = 0. What is
68: its orbit?
69:
70: The geometry of an $nl$ Sommerfeld ellipse is given by its semimajor axis, $%
71: a_{nl}=(r_{B}/Z)n^{2}$, and semiminor axis, $b_{nl}=(r_{B}/Z)n\sqrt{l(l+1)}$.%
72: \cite{7} Here $r_{B}=h^{2}/4\pi ^{2}me^{2}$ is the Bohr radius in terms of
73: fundamental constants, serving as an atomic distance unit, and $Ze$ is the
74: nuclear charge. An $(n,0)$ orbit is thus a line ellipse with its nuclear
75: focus at one end and its empty focus at the other. This case was regarded as
76: unphysical because of the electron's collision with the nucleus---an
77: uncritical adaptation from celestrial mechanics. A closer inspection
78: confirms that a line ellipse with \textit{terminal} nuclear focus is indeed
79: unphysical---but for quite a different reason!\bigskip
80:
81: Figure 1 Here
82:
83: {\small Fig. 1. Partial trajectory of an extranuclear orbit }$XX^{\prime }$%
84: {\small \ and of a penetrating orbit }$PP^{\prime }${\small \ through
85: nucleus }$N${\small . The dotted line }$S${\small \ shows the major symmetry
86: axis of }$XX^{\prime }${\small .}\bigskip
87:
88: Leaving quantization conditions momentarily aside, what would happen if we
89: \textit{continuously} decrease an angular quantum number $\lambda $ while
90: keeping the principal quantum number $n$ constant? We then would get more
91: and more slender ellipses with the same length of major axis, $2a_{n}$. By
92: basic electric theory, the nuclear Coulomb potential outside a \textit{\
93: finite-size} nucleus $N$ is given by the point potential as if all nuclear
94: charge, $+Ze$, was concentrated at the center of the nucleus. This holds as
95: long as the electron orbit stays outside the nucleus. Two borderline cases
96: are illustrated in Fig. 1. For a very small $\lambda $ value, say $0<\lambda
97: _{X}\ll 1$, we obtain a very slender elliptical orbit with partial
98: trajectory $XX^{\prime }$ about the nucleus. Further decrease of $\lambda $
99: to $\lambda _{P}$ \TEXTsymbol{<} $\lambda _{X}$ causes an intrusion of the
100: electron into the finite nucleus of radius $r_{N}$ (see trajectory $PN$ in
101: Fig. 1). Once inside, at a distance $r<r_{N}$ from the center, then, by
102: Gauss's law, only a fraction of the nuclear charge, $Z^{\prime }(r)e<Ze$,
103: acts on the electron via centripetal force.\cite{8} Accordingly, the
104: electron's exit trajectory $NP^{\prime }$ is no longer symmetric to its
105: approach trajectory $PN$ with respect to the major axis $S$ of the (partial)
106: ellipse $XX^{\prime }$. In the extreme case of a head-on penetration of the
107: nucleus, $\lambda =0$, there is no centripetal force at all! The electron
108: will then, with almost constant speed, traverse the nucleus, continue, with
109: decreasing speed, to the opposite turning point of its line orbit and revert
110: its motion periodically. We want to call the electron's straight-line
111: oscillation in the Coulomb potential of a finite-size nucleus a \textit{%
112: ``Coulomb oscillator.''}
113:
114: \section{HYDROGEN ATOM}
115:
116: For the formal treatment of the non-relativistic Coulomb oscillator we
117: designate the $z$ axis along the line orbit, with turning points at $z=\pm
118: \Gamma $ and nuclear position at $z=0$ (see Fig. 2). The electron's total
119: energy $E$ at position $z$ must equal the potential energy at a turning
120: point,
121:
122: \begin{equation}
123: E=\frac{1}{2}mv^{2}-\frac{Ze^{2}}{|z|}=-\frac{Ze^{2}}{\Gamma }. \tag{1}
124: \end{equation}
125:
126: \noindent This gives the electron's speed along the $z$ axis,
127:
128: \begin{equation}
129: v=\pm e\sqrt{\frac{2Z}{m}}\sqrt{\frac{1}{|z|}-\frac{1}{\Gamma }}. \tag{2}
130: \end{equation}
131:
132: \noindent Figure 3 displays its dependence on the axial position as a
133: two-wing curve cusped at the nucleus. Atomic units $(a.u.)$ are used, that
134: is, the Bohr radius $r_{B}$ and the ``Bohr speed'' $v_{B}=2\pi
135: e^{2}/h=\alpha c$---the electron speed in the ground-state Bohr orbit of the
136: $H$ atom---with fine-structure constant $\alpha $ $\approx $ 1/137 and speed
137: of light $c$. The curve's wing along the positive $z$ axis, $|z|=r$, gives
138: the \textit{radial} speed, $v_{r}(r)=|v(z)|$, necessary for Sommerfeld's
139: \textit{radial} quantization condition,
140:
141: \begin{equation}
142: \oint p_{r}dr=m\oint v_{r}(r)dr=n_{r}h. \tag{3}
143: \end{equation}
144: \noindent
145:
146: \noindent Here $p_{r}$ is the radial momentum, $n_{r}=1,2,...$ is the radial
147: quantum number, and $h$ is Planck's quantum of action. Integration is over
148: one period of the radial motion, $z=+\Gamma \rightarrow 0\rightarrow +\Gamma
149: $. Graphically, the radial quantization is illustrated in Fig. 3 by the area
150: under \textit{one} wing of the speed curve. For a line orbit, $l=0$ , the
151: radial quantum number equals the principal quantum number, $n\equiv
152: n_{r}+l=n_{r}$.
153:
154: In order to express the radial quantization in terms of \textit{axial}
155: motion we employ a ``fold-out factor,'' $\phi =2$, to compensate for the
156: doubling of integration range in the extension from the radial one-wing
157: speed curve to the axial two-wing curve. The quantization is thus restated,
158:
159: \begin{equation}
160: \frac{1}{\phi }\oint p_{z}dz=\frac{m}{\phi }\oint v(z)dz=n_{z}h, \tag{3'}
161: \end{equation}
162:
163: \noindent with axial quantum number $n_{z}=n_{r}=n$ and integration over the
164: axial double-wing range, $z=+\Gamma \rightarrow \ -\Gamma \rightarrow \
165: +\Gamma $. By symmetry we can restrict the axial action integral to one
166: quarter of the oscillation, say $z=+\Gamma \rightarrow \ 0$,
167:
168: \begin{equation}
169: \frac{1}{\phi }\oint p_{z}dz=-\frac{4m}{\phi }\int_{\Gamma }^{0}v(z)dz=-%
170: \frac{4e}{\phi }\sqrt{2Zm}\int_{\Gamma }^{0}\sqrt{\frac{1}{z}-\frac{1}{
171: \Gamma }}dz=nh. \tag{4}
172: \end{equation}
173:
174: \noindent Here the electron's motion in the negative $z$ direction is
175: accounted for by the negative sign. Although the electron's speed through a
176: point nucleus diverges, $v(0)=\infty $, the action integral, Eq. (4), stays
177: finite and determines the quantized amplitude $\Gamma _{n}$ of the Coulomb
178: oscillator. Graphically the amplitude $\Gamma _{n}$ must be such that it
179: stretches the speed curve horizontally to the extent that the area under one
180: wing, $A_{n}=nh$, represents the quantized action. The analytic solution,
181: derived in Appendix A, is
182:
183: \begin{equation}
184: \Gamma _{n}=2\frac{r_{B}}{Z}n^{2}. \tag{5}
185: \end{equation}
186:
187: \noindent Inserting Eq. (5) into Eq. (1) yields the quantized energy,
188:
189: \begin{equation}
190: E_{n}=-\frac{Z^{2}e^{2}}{\Gamma _{n}}=-\frac{Z^{2}}{n^{2}}R_{y}, \tag{6}
191: \end{equation}
192:
193: \noindent in terms of the Rydberg energy unit, $R_{y}=2\pi
194: ^{2}me^{4}/h^{2}=13.6$ $eV$, and in agreement with the energy of the $n$th
195: Bohr orbit.
196:
197: Note that Eq. (5) gives the amplitude of the $n$th Coulomb oscillator as
198: \textit{twice} the radius of the $n$th Bohr orbit or of the semimajor axis
199: of an $nl$ Sommerfeld ellipse, $r_{n}=a_{nl}=(r_{B}/Z)n^{2}$. For
200: comparison, the time-average radial distance of a Kepler orbit\cite{9} of
201: major and minor semiaxes $a$ and $b$, respectively, is $\langle r\rangle
202: _{t}=(3a^{2}-b^{2})/(2a)$. A line ellipse ($b=0$) has then $\langle r\rangle
203: _{t}=\frac{3}{2}a$. Applied to an $nl$ Sommerfeld orbit,\cite{9} its average
204: size, $\langle r_{nl}\rangle _{t}=(r_{B}/Z)[3n^{2}-l(l+1)]/2$, is in
205: agreement with the corresponding quantity from quantum mechanics,\cite{10} $
206: \langle r_{nl}\rangle =$ $\smallint $ $\psi ^{*}r\psi d^{3}r$. Thus the
207: time-average radial distance of a Coulomb oscillator is $\langle
208: r_{n0}\rangle _{t}=\frac{3}{2}(r_{B}/Z)n^{2}$. For the ground state of the
209: hydrogen atom, $n=1$, this gives $\langle r_{10}\rangle _{t}=\frac{3}{2}
210: r_{B} $, as is well-known from quantum mechanics.\cite{10}
211:
212: The concept of the electron's semiclassical Coulomb oscillation is
213: consistent with the Fermi-contact term of hyperfine interaction for $l=0$
214: states, which arises from the presence of the electron \textit{inside} the
215: nucleus. This is familiar from quantum mechanics\cite{11} and can be
216: interpreted semiclassically as a local-field effect.\cite{12}
217:
218: To be sure, the extension of Sommerfeld's theory by the Coulomb oscillator
219: resolves the discrepancy of the old quantum theory with spectroscopy,
220: mentioned above, only at the \textit{low} end of angular quantum numbers, $%
221: l=0$. The resolution at the high end---repeal of the circular Bohr orbit, $%
222: l\neq n$---involves an analysis in terms of space quantization which is
223: beyond the scope of the present study.
224:
225: \section{HYDROGEN MOLECULE ION}
226:
227: \subsection{Formalism}
228:
229: A hydrogen molecule ion, $H_{2}^{+}$, consists of two proton nuclei and one
230: electron. We assume, in adiabatic approximation, the protons located at
231: fixed positions $z=\pm c$ on the molecular axis (see Fig. 4). The term
232: ``Coulomb oscillator'' denominates again the motion of a point electron, now
233: along either the line through the protons ($z$ axis) and with turning points
234: $\pm C$, or along the perpendicular line through the midpoint ($y$ axis)
235: with turning points $\pm B$. The molecule ion's total energy $E$ at any
236: position on the axis, $-C\leqslant z\leqslant +C$, must equal the potential
237: energy at the turning point $C$,
238:
239: \begin{equation}
240: E=\frac{1}{2}mv^{2}-\frac{e^{2}}{|z+c|}-\frac{e^{2}}{|z-c|}+\frac{e^{2}}{2c}
241: =-\frac{e^{2}}{C+c}-\frac{e^{2}}{C-c}+\frac{e^{2}}{2c}. \tag{7}
242: \end{equation}
243:
244: \noindent For electron positions beyond the protons, $z>c$, this gives an
245: electron speed
246:
247: \begin{equation}
248: v_{out}=\pm \frac{2e}{\sqrt{m}}\sqrt{\frac{z}{z^{2}-c^{2}}-\frac{C}{
249: C^{2}-c^{2}}}. \tag{8a}
250: \end{equation}
251:
252: \noindent For positions between the protons, $0<z<c$, the corresponding
253: speed is
254:
255: \begin{equation}
256: v_{in}=\pm \frac{2e}{\sqrt{m}}\sqrt{\frac{c}{c^{2}-z^{2}}-\frac{C}{
257: C^{2}-c^{2}}}. \tag{8b}
258: \end{equation}
259:
260: \noindent The speed expressions will be used in the action integral,
261:
262: \begin{equation}
263: \frac{1}{\phi }\oint p_{z}dz=\frac{2m}{\phi }%
264: \int_{C}^{-C}v(z)dz=A_{z}=A_{out}+A_{in}, \tag{9}
265: \end{equation}
266:
267: \noindent with outer contribution
268:
269: \begin{equation}
270: A_{out}=\frac{2m}{\phi }\left[
271: \int_{C}^{c}v_{out}(z)dz+\int_{-c}^{-C}v_{out}(z)dz\right] \tag{10a}
272: \end{equation}
273:
274: \noindent and inner contribution
275:
276: \begin{equation}
277: A_{in}=\frac{2m}{\phi }\int_{c}^{-c}v_{in}(z)dz. \tag{10b}
278: \end{equation}
279:
280: \noindent Here $\phi $ is a fold-out factor to be specified below.
281:
282: If the conditions are such that the electron swings along the $z$ axis
283: through the midpoint $0$, that is, $v_{in}(0)>0$, then there exists also an
284: oscillation along the $y$ axis with lateral speed $u$, having the same total
285: energy,
286:
287: \begin{equation}
288: E=\frac{1}{2}mu^{2}-\frac{2e^{2}}{\sqrt{y^{2}+c^{2}}}+\frac{e^{2}}{2c}=-%
289: \frac{2e^{2}}{\sqrt{B^{2}+c^{2}}}+\frac{e^{2}}{2c}. \tag{11}
290: \end{equation}
291:
292: \noindent Equal energy at the axial and lateral turning points, $E(C)=E(B)$,
293: Eqs. (7) and (11), determines the latters' geometric dependence,
294:
295: \begin{equation}
296: B=\sqrt{C^{2}+\frac{c^{4}}{C^{2}}-3c^{2}}. \tag{12}
297: \end{equation}
298:
299: \noindent Solving Eq. (11) for the lateral speed,
300:
301: \begin{equation}
302: u=\pm \frac{2e}{\sqrt{m}}\sqrt{\frac{1}{\sqrt{y^{2}+c^{2}}}-\frac{1}{\sqrt{%
303: B^{2}+c^{2}}},} \tag{13}
304: \end{equation}
305:
306: \noindent provides the integrand of the action integral over a lateral
307: oscillation,
308:
309: \begin{equation}
310: A_{y}=\frac{1}{\phi }\oint p_{y}dy=\frac{2m}{\phi }\int_{B}^{-B}u(y)dy.
311: \tag{14}
312: \end{equation}
313:
314: Subtraction of the protons' mutual repulsion from the total energy $E$ of
315: axial or lateral motion, Eqs. (7) and (11), gives the \textit{electronic}
316: energy,
317:
318: \begin{equation}
319: E_{el}=E-\frac{e^{2}}{R}, \tag{15}
320: \end{equation}
321:
322: \noindent in its dependence on the proton separation, $R=2c$. This brackets
323: the molecular problem with known atomic results, Eq. (6), in the limits of $%
324: R=\infty $ (free $H$ atom, $Z=1$) and $R=0$ (free $He^{+}$ ion, $Z=2$). For
325: those cases, as well as any proton-proton distance $R$ between, we keep the
326: action constant,
327:
328: \begin{equation}
329: A=A_{z}+A_{y}=nh. \tag{16}
330: \end{equation}
331:
332: \noindent Equation (16) is the \textit{Einstein }quantization condition\cite
333: {13}---a generalization of Sommerfeld's quantization over separable
334: variables---where the \textit{quantum sum} equals the sum of action
335: integrals over topologically independent paths in phase space.\cite{14}
336:
337: Here we treat the molecule ion only in its \textit{ground state}, $n=1$.
338: Analytic solutions of the action integrals, Eqs. (10ab) and (14), are
339: complicated due to elliptic functions. We therefore integrate numerically
340: and visualize the integrals by the area under the corresponding speed
341: curves. The bold curve in Fig. 5 shows the axial electron speed $v(z)$ for a
342: \textit{far} proton separation, $R=6$ $r_{B}$. The electron, in its
343: semiclassical motion, then oscillates only about (and through) the right
344: proton. The area under the speed curve, Eq. (8ab), proportionally represents
345: the ground state's unity of action, $m\oint v(z)dz=\phi h$, with a fold-out
346: factor $\phi =2$ in analogy to the free-atom case, Eq. (4). The thin curve
347: shows, for comparison, the axial electron speed in a free $H$
348: atom---familiar from Fig. 3---centered at the same proton position, $+c$.
349: The pull from the left proton (at $-c$) on the oscillating electron can be
350: seen by the distortion of the speed function $v(z)$ and the redistribution
351: of the area under the curve.
352:
353: When, with closer proton separation $R$, as in Fig. 6, the electron swings
354: past the midpoint $0$, then the single-cusp speed curve $v(z)$ from Fig.
355: 5---akin in shape to letter $\Lambda $---becomes double-cusped (akin to
356: letter M) and symmetric with respect to the bisector ($y$ axis). Now there
357: is also a lateral oscillation with speed $u(y)$ having the same total energy
358: $E$. The equality of $E$ in both cases can be seen in Fig. 6 by the equality
359: of axial and lateral speed at the midpoint, $v(0)=u(0)$---a position where
360: the electron experiences the same potential in either case. For convenience
361: the lateral speed $u(y)$, though perpendicular to the proton axis, is
362: displayed in Fig. 6 together with the axial speed $v(z)$. The lateral speed
363: curve $u(y),$ drawn dotted, is readily recognized by its dome shape ($\cap $%
364: ). At the bifurcation value of the proton separation, $\check{R}$, where the
365: electron starts swinging though the midpoint $0$, the axial speed curve $%
366: v(z) $ changes from its one-centered $\Lambda $ shape to a two-centered M
367: shape. The \textit{area} under the axial speed curve then abruptly doubles, M%
368: $(\check{R}-\delta )$ $\approx 2\Lambda (\check{R}+\delta )$, upon a very
369: small change in proton separation, $\delta \ll $ $\check{R}$. In order to
370: keep the action integral continuous at $\check{R}$, the sudden area doubling
371: is compensated by a corresponding doubling of the fold-out factor from $\phi
372: $ $=2$ for $R$ $>$ $\check{R}$ to $\phi =4$ for $R$ $<\check{R}.$ The same
373: fold-out factor, $\phi $ $=4$, must be used for the lateral action integral $%
374: A_{y}$, Eq. (14), as will become clear shortly.
375:
376: The unity of action, $A=1h$, is visualized again in Fig. 6. To this end we
377: compare the right half of the bold M-shape curve of axial electron speed in
378: the molecule with the thin curve $\Lambda (H)$ of the axial speed in a free $%
379: H\,$atom positioned at the right nucleus, $+c$. Due to attraction from the
380: left nucleus, the right wing of the M curve is smaller than that of $\Lambda
381: (H)$ by the area of lobe $L$. On the other hand, the left flank of the $%
382: \Lambda (H)$ curve that extends over the negative $z$ axis is smaller than
383: the left wing of the lateral speed curve $\cap $ by the area of slice $S$.
384: The area under the free-atom curve is then
385: \begin{equation}
386: \Lambda (H)\approx \frac{1}{2}\text{M}+L+\frac{1}{2}\cap -S. \tag{17a}
387: \end{equation}
388:
389: \noindent The areas of lobe and slice are comparable,
390: \begin{equation}
391: L\approx S. \tag{17b}
392: \end{equation}
393:
394: \noindent When the tiny notch to the right of the saddle point of M is taken
395: into account, then the approximations (17ab) become equations and combine to
396: \begin{equation}
397: \text{M}+\cap =2\Lambda (H). \tag{18}
398: \end{equation}
399:
400: \noindent The area under both the axial and lateral speed curves is thus
401: four times the area under \textit{one} wing of the free-atom curve, M$+\cap
402: =4\times \frac{1}{2}\Lambda (H)$. Since the latter represents one quantum of
403: action, $h$, the combined area M$+\cap $ visualizes its \textit{double}
404: fold-out, $\phi =4$.
405:
406: With very close proximity of the nuclei, $R\rightarrow 0$, the crests of the
407: M curve start merging while its saddle point, $v(0)$, keeps rising. In the $%
408: R=0$ limit of fusing nuclei the axial electron speed becomes that of a free $%
409: He^{+}$ ion, M$\rightarrow $\ $\Lambda (He^{+})$, familiar from Fig. 3.
410: Concurrently, the lateral speed curve $\cap $ rises at its peak, $u(0)$ $=$ $%
411: v(0)$, and narrows at its base until it, too, turns into the speed curve of
412: the free $He^{+}$ ion, $\cap \rightarrow $\ $\Lambda (He^{+})$. In the $R=0$
413: limit the three curves merge, M$(0)=\cap (0)$ $=$ $\Lambda (He^{+})$.
414:
415: The results of the Coulomb-oscillator approach will be compared with another
416: semiclassical calculation of $H_{2}^{+}$, by Strand and Reinhardt.\cite{15}
417: These authors, like Pauli,\cite{4} separate the equation of motion in
418: spheroidal coordinates, $\xi $ $=(r_{+}+r_{-})/2c$ , $\eta =(r_{+}-r_{-})/2c$
419: and $\varphi $ by virtue of the constants of the motion: total energy $E$,
420: angular momentum $\mathbf{M}$ about the $z$ axis, and a component of the
421: bifocal Runge-Lenz vector, $\Omega _{c}$.\cite{16} Here $r_{+}$ $(r_{-})$ is
422: the distance of the electron from the nucleus at $+c$ $(-c).$ Strand and
423: Reinhardt (SR) solve the ensuing one-dimensional differential equations with
424: classical Poisson-bracket techniques. They find the electron's trajectories
425: conditionally periodic\cite{17} and regionally confined due to restrictions
426: from $E$, $\mathbf{M}$ and $\Omega _{c}$. However, unlike Pauli, who used
427: Sommerfeld quantization, SR employ the Einstein-Brillouin-Keller (EBK)
428: quantization conditions,
429:
430: \begin{equation}
431: A_{j}=\oint p_{j}dj=(n_{j}+\frac{1}{2})h,\text{ }j=\xi ,\text{ }\eta
432: \tag{19ab}
433: \end{equation}
434:
435: \noindent and
436:
437: \begin{equation}
438: A_{\varphi }=\oint p_{\varphi }d\varphi =n_{\varphi }h. \tag{19c}
439: \end{equation}
440:
441: The background of EBK quantization touches on the foundations of classical
442: and quantum mechanics.\cite{18} For the present purpose its essential
443: rationale may be summarized as follows: A semiclassical treatment envolves
444: turning points of radial, or other librating motion. Any tunneling through
445: ``forbidden'' regions of negative kinetic energy is ruled out. Viewed in
446: terms of the quantum mechanical WKB (Wentzel-Kramers-Brillouin)
447: approximation, the hard reflection of a wavefunction at a turning point
448: corresponds to a so-called ``loss'' of phase (phase shift by $\pi $)
449: compared to the soft reflection caused by tunneling (phase shift by $\pi /2$
450: ). This shortcoming can be remedied with EBK quantization conditions by
451: addition of a value of 1/4, for each librational turning point, to the
452: corresponding quantum number. Such is the case for the electron's elliptical
453: and hyperbolic librations in the above quantization, Eq. (19ab), but not for
454: a rotation about the $z$ axis, Eq. (19c). Strand and Reinhardt call these
455: quantization conditions ``primitive'' to distinguish them from more
456: sophisticated ones, specified below.
457:
458: A semiclassical treatment of a free $H$ atom with EBK quantization has
459: recently been presented in these pages.\cite{19} In this case the isotropic
460: symmetry permits a separation of variables in spherical coordinates, $r$, $%
461: \theta $ and $\varphi $, and the EBK quantization conditions are like Eqs.
462: (19abc) except for $j=r,$ $\theta $. The atomic ground state is
463: characterized by the quantum numbers ($n_{r},$ $n_{\theta },$ $n_{\varphi
464: })=(0,0,0)$. Accordingly, the action in the atom's ground state, $A_{1}=$ $%
465: 1h $, is attributed only to tunneling at the radial and latitudinal turning
466: points (the former being the nucleus). Applying EBK quantization to the
467: ground state of $H_{2}^{+}$, denoted $1s\Sigma _{g}$ in molecular
468: spectroscopy, SR likewise assign the quantum numbers $(n_{\xi },n_{\eta
469: },n_{\varphi })=(0,0,0)$.
470:
471: \subsection{Results}
472:
473: Energies of the Coulomb oscillator (CO), in adiabatic dependence on the
474: proton separation, are listed in Appendix B and shown in Fig. 7 in
475: comparison with exact quantum mechanical (QM) results and the semiclassical
476: calculation by SR.\cite{15} The lower part of Fig. 7 shows the electronic
477: energy $E_{el}(R)$ of the ground state, $1s\Sigma _{g}$. At \textit{large}
478: proton separations, $R>6$ $r_{B}$, both the SR calculation (circles) and the
479: CO approach (crosses) agree excellently with the exact QM values (curve).
480: This is the situation where the electron stays near one nucleus (see Fig.
481: 5). As Fig. 7 further shows, such agreement ceases once the classical
482: electron motion leads beyond the molecular bisector, which happens for
483: proton separations below the bifurcation value, $R<\check{R}\approx 5.57$ $%
484: r_{B}$ (see Fig. 6). Interestingly, the deviations of CO and SR from QM are
485: opposite over the entire range. The SR results are \textit{discontinuous} at
486: a certain proton separation, $R^{*}\approx 1.38$ $r_{B}$. Remarkably, at (or
487: near) that value, $R^{*}$, the CO energy crosses the curve of the QM
488: solution.
489:
490: Adding to the electronic energy $E_{el}$ the proton-proton repulsion gives
491: the \textit{total} energy $E$, Eq. (15). The middle part of Fig. 7 shows by
492: the dashed curve the exact total energy $E(R)$ of the $1s\Sigma _{g}$ ground
493: state, obtained from QM and, by symbols, the corresponding CO and SR values.
494: The solid dot at the minimum of the curve shows the QM equilibrium energy,
495: in agreement with experiment, $E_{0}=-1.20$ $R_{y}$, and the equilibrium
496: internuclear distance, $R_{0}=2.00$ $r_{B}$. The CO energy ($\times $) comes
497: out too low, due to the inaccuracy of its $E_{el}$, with an equilibrium
498: value $E_{0}(CO)$ $\approx -1.38$ $R_{y}$ at $R_{0}(CO)$ $\approx $ $2.5$ $%
499: r_{B}$. Conversely, the SR energy ($\square $) comes out too high with $%
500: E_{0}(SR)$ $\approx -1.05$ $R_{y}$ at $R_{0}(SR)$ $\approx $ $5$ $r_{B}$.
501: Since the CO and SR results deviate about equal and oppositely from QM,
502: their \textit{average} ($\Diamond $) is close to the exact values with a
503: minimum of $E_{0}[\frac{1}{2}(CO+SR)]$ $\approx -1.19$ $R_{y}$ at $R_{0}[%
504: \frac{1}{2}(CO+SR)]$ $\approx $ $2.5$ $r_{B}$.
505:
506: The molecular binding energy is the difference of $E$ in the molecule and in
507: the constituting atoms, here, $E_{0}(H_{2}^{+})-$ $E(H)$. The energy of a
508: free hydrogen atom, $E(H)=-1$ $R_{y}$, is indicated in Fig. 7 by the fine
509: horizontal line. Both semiclassical treatments, CO and (barely) SR, yield
510: molecular binding energies with \textit{negative} values and thus a \textit{%
511: \ stable} molecule ion. Why are they more successful than the early
512: attempts, in the 1920s, by Pauli,\cite{4} and independently Niessen,\cite{20}
513: with the Sommerfeld quantization conditions of the old quantum theory?
514:
515: For the same reason that Sommerfeld had excluded the angular quantum number $%
516: l=0$ for the $H$ atom---avoidance of electron collision with the
517: nucleus---both Pauli and Niessen excluded electron motion in the nuclear
518: plane of the $H_{2}^{+}$ molecule ion. They then found the lowest admissible
519: quantum state to be $(n_{\xi },$ $n_{\eta },$ $n_{\varphi })=(0,1,1)$,
520: denoted $2p\Pi _{u}$ in molecular spectroscopy, with $E_{0}(P,N)=-0.52$ $%
521: R_{y}$ at $R_{0}(P,N)=5.53$ $\pm 0.01$ $r_{B}$, depicted by the triangle in
522: the top part of Fig. 7. The QM energy\cite{21} of that quantum state is $
523: E_{0}(2p\Pi _{u})=-0.27$ $R_{y}$ at $R_{0}(2p\Pi _{u})$ $\approx $ $8$ $%
524: r_{B},$ marked by the solid dot at the minimum of the dotted curve. Since
525: both these energy values are higher than the ground state of a free hydrogen
526: atom, they give rise to \textit{positive} molecular binding energies and
527: thus to spontaneous dissociation, $H_{2}^{+}(2p\Pi _{u})$ $\rightarrow $ $H$
528: $+$ $H^{+}$. Qualitatively, Pauli's and Niessen's finding of energetic
529: \textit{instability} is borne out by quantum mechanics for this exited state
530: of $H_{2}^{+}$ (Pauli's argument\cite{4} about ``dynamical stability''
531: notwithstanding). The deviation of their historical value $(\triangle )$
532: from the (dotted) QM curve is remarkably small---comparable to those of the
533: CO and SR results for the ground state. Pauli's and Niessen's misfortune,
534: though, was that they \textit{\ misinterpreted} their result as the molecule
535: ion's ground state---an assessment with fateful consequences in the
536: development of quantum theory.
537:
538: \subsection{Discussion}
539:
540: Why do the semiclassical results of the Coulomb oscillator and of SR's
541: quantization deviate from the QM solution of the $H_{2}^{+}$ molecule ion?
542: Strand and Reinhardt explain the deviation of their primitive quantization
543: from QM with effective potential barriers arising from constrictions due to
544: conservation of both the energy $E$ and the bifocal Runge-Lenz component $%
545: \Omega _{c}$ (the angular momentum vanishes for the ground state, $\mathbf{M}%
546: =0$). The most drastic consequence of those barriers is the discontinuity of
547: $E_{el}$ at $R^{*}$ (see Fig. 7) and the large deviations at closer proton
548: separation, $R<R^{*}$. While the simulation of quantum mechanical tunneling
549: beyond the semiclassical turning points of librations is adequately achieved
550: by EKB quantization under the isotropic symmetry of a free $H$ \textit{atom},%
551: \cite{19} Eq. (19) is less successful under the lower symmetry of $H_{2}^{+}$
552: . When SR remedy the situation with ``unified'' semiclassical quantization
553: conditions, then $E_{el}$ agrees, for all practical purposes, with QM. Those
554: unified quantization condition, going well beyond Eq. (19), are
555: sophisticated in their dependence on $E_{el}$, $\Omega _{c},$ and the
556: hyperbolic turning points $\eta _{\pm }$. They will not be discussed here.
557:
558: The reason for the deviation of the CO results from the QM values is the
559: \textit{neclect} of the electron's \textit{wave} nature in the underlying
560: quantization condition, Eq. (16). In proposing his wave hypothesis de Broglie%
561: \cite{22} already showed that the quantization condition of the Bohr model, $%
562: A_{n}=nh$, is equivalent to $n$ standing waves along the $nth$ Bohr orbit.
563: If $s$ denotes the position along the Bohr orbit, then the de Broglie wave
564: can be expressed as $w(s)=\sin [2\pi a_{n}(s)/h]$, with the variable $%
565: a_{n}(s)=(A_{n}/S_{n})\int_{0}^{s}ds^{\prime }$ along the orbit's
566: circumference $S_{n}=2\pi r_{n}$. A generalization gives for the Coulomb
567: oscillator of a free $H$ \textit{atom} in the ground state ($n=1$), its de
568: Broglie wave as
569: \begin{equation}
570: w(z)=\sin [2\pi a(z)/h] \tag{20a}
571: \end{equation}
572:
573: \noindent with
574: \begin{equation}
575: a(z)=m\int_{0}^{z}v(z^{\prime })dz^{\prime } \tag{20b}
576: \end{equation}
577: and the speed $v(z^{\prime })$ from Eq. (2). This de Broglie wave, shown by
578: the line curve in Fig. 8 for a free $H$ atom positioned at $+c$, has a node
579: at the nucleus and at each turning point. By the above characterization,
580: those turning points are ``soft,'' and it is their softness that ensures the
581: exact energy of the free atom. However, when Eq. (20ab) is applied to $%
582: H_{2}^{+}$ for proton separations beyond the bifurcation value, $R>\check{R}$%
583: , with axial speed from Eq. (8ab) and integration away from the occupied
584: nucleus, $\int_{c}^{z}...$, then the de Broglie wave is found to be
585: ``truncated'' (no nodes) at both the outer and inner turning point (see Fig.
586: 8, small circles). Those turning points are ``hard'' and give rise to
587: incorrect energies. Qualitatively, an augmentation of the truncated de
588: Broglie wave with (exponential) ``tunneling tails,'' determined by the
589: negative kinetic energy in the classically ``forbidden'' region, would
590: ``soften'' the turning points. This would give rise to an \textit{effective}
591: far turning point farther out, $C_{eff}$ $>C$, and accordingly raise the CO
592: energy, Eq. (7), toward the QM result.
593:
594: When the proton separation is below bifurcation, $R<\check{R}$, then Eq.
595: (20b) should be integrated from the midpoint $0$ in both the axial and
596: lateral direction rather than favoring one nucleus with the node of the de
597: Broglie wave. Again, the de Broglie wave is found to be truncated at the
598: axial and lateral turning points, $C$ and $B$, respectively (see Fig. 9). An
599: exception exits for the proton separation $R^{*}$. The de Broglie wave,
600: shown in Fig. 10, then has a minimum at both turning points, $w(C)=w(B)=-1$,
601: according to action values of $A_{z}=\frac{3}{4}h$ and $A_{y}=\frac{1}{4}h$.
602: Such turning points seem to be ``benign''---reminiscent of the soft turning
603: points in the free-atom case---and cause the CO energy $E_{el}(R^{*})$ in
604: Fig. 7 to be \textit{exact}. At still smaller proton separation, $R<R^{*}$,
605: the de Broglie wave is truncated again (not shown). The limit $R=0$
606: corresponds to a Coulomb oscillator in the free $He^{+}$ ion which, like in
607: the free $H$ atom for $R=\infty $, has de Broglie nodes at the turning
608: points and an exact energy value.
609:
610: If the explanation that the CO energy $E_{el}(R)$ deviates from the QM curve
611: because of the \textit{neglect} of \textit{wave} effects is valid, then this
612: sheds new light on the SR results. The opposite sign of the CO and SR
613: deviations then suggests that SR's primitive EBK quantization, while
614: appropriate for a free atom, simulates \textit{too much} \textit{wave effects%
615: } under the lower symmetry of the $H_{2}^{+}$ molecule ion. However, the
616: \textit{average} of both those semiclassical quantizations seems to be an
617: excellent compromise, as evidenced by the close agreement of the
618: corresponding total energy $(\Diamond )$ with the (dashed) QM curve in Fig.
619: 7.
620:
621: In conclusion, semiclassical quantization can rise to Gutzwiller's challenge
622: and ``explain'' the chemical bond in the paradigm molecule,\cite{23} $%
623: H_{2}^{+}$, by a combination of classical mechanics, quantization, and
624: moderate wave effects.
625:
626: \section{\noindent \textbf{ACKNOWLEDGMENTS}}
627:
628: I thank Duane Siemens and Ernst Mohler for valuable discussions. Many thanks
629: to Preston Jones for help with computer integration and graphics. I also
630: thank Professor Gutzwiller for advice and his kind encouragement.
631:
632: \section{APPENDIX A: QUANTIZATION}
633:
634: By Eq. (4) the action integral of the atomic Coulomb oscillator is
635:
636: \begin{equation}
637: A=\frac{1}{\phi }\oint p_{z}dz=-\frac{4e}{\phi }\sqrt{2Zm}\int_{\Gamma }^{0}%
638: \sqrt{\frac{1}{z}-\frac{1}{\Gamma }}dz. \tag{4'}
639: \end{equation}
640:
641: \noindent For a comparison with integral tables we change notation to $x=z$
642: and use the abbreviation $a=-1/\Gamma $. Then
643:
644: \begin{equation}
645: \int \sqrt{\frac{1}{z}-\frac{1}{\Gamma }}dz=\int \sqrt{\frac{1}{x}+a}dx=\int
646: \frac{\sqrt{X}}{x}dx \tag{21}
647: \end{equation}
648:
649: \noindent with $X=ax^{2}+x$. Integration tables give
650:
651: \begin{equation}
652: \int \frac{\sqrt{X}}{x}dx=\sqrt{X}+\frac{1}{2}\int \frac{dx}{\sqrt{X}}.
653: \tag{22}
654: \end{equation}
655:
656: \noindent The first term on the rhs, evaluated at the limits of the $%
657: \int_{\Gamma }^{0}$ integration, vanishes. The last integral in Eq. (22),
658: tabulated as
659:
660: \begin{equation}
661: \int \frac{dx}{\sqrt{X}}=(-\sqrt{\Gamma })\arcsin (1-2x/\Gamma ), \tag{23}
662: \end{equation}
663:
664: \noindent and evaluated at the limits, $x=0$ and $x=\Gamma $, contributes
665:
666: \begin{equation}
667: -[\arcsin (1)-\arcsin (-1)]\sqrt{\Gamma }=-\pi \sqrt{\Gamma }. \tag{24}
668: \end{equation}
669:
670: \noindent Combining Eqs. (4'), (22) and (24), together with a fold-out
671: factor $\phi $ = 2, gives the action integral, to be equated with the
672: Sommerfeld quantization condition,
673:
674: \begin{equation}
675: A_{n}=-\frac{4e}{\phi }\sqrt{2Zm}\frac{1}{2}(-\pi \sqrt{\Gamma })=nh.
676: \tag{25}
677: \end{equation}
678:
679: \noindent We square Eq. (25) and solve for the amplitude of the quantized
680: Coulomb oscillator,
681:
682: \begin{equation}
683: \Gamma _{n}=2\frac{r_{B}}{Z}n^{2}, \tag{26}
684: \end{equation}
685:
686: \noindent in terms of the Bohr radius $r_{B}$.
687:
688: \section{APPENDIX B: DATA}
689:
690: \noindent TABLE I. Electronic ground-state energy $E_{el}$ for various
691: nuclear separations $R$ of the $H_{2}^{+}$ molecule from quantum-mechanical
692: calculations (QM, Ref 21), the ``primitive'' semiclassical quantization of
693: Strand and Reinhardt (SR, Ref. 15), and the present Coulomb-oscillator
694: approach (CO).
695:
696: \begin{tabular}{llll}
697: R $(a.u.)$ & QM $(R_{y})$ & SR $(R_{y})$ & CO $(R_{y})$ \\
698: 0.00 & -4.00 & -4.00 & -4.00 \\
699: 0.25 & -3.80 & -4.75 & -3.35 \\
700: 0.50 & -3.47 & -4.27 & -3.12 \\
701: 0.75 & -3.11 & -3.76 & -2.93 \\
702: 1.00 & -2.90 & -3.34 & -2.81 \\
703: 1.25 & -2.68 & -3.07 & -2.67 \\
704: 1.50 & -2.50 & -2.19 & -2.55 \\
705: 1.75 & -2.34 & -2.09 & -2.45 \\
706: 2.00 & -2.21 & -1.99 & -2.36 \\
707: 2.50 & -1.99 & -1.79 & -2.18 \\
708: 3.00 & -1.82 & -1.66 & -2.02 \\
709: 3.50 & -1.69 & -1.56 & -1.87 \\
710: 4.00 & -1.59 & -1.49 & -1.75 \\
711: 5.00 & -1.45 & -1.40 & -1.54 \\
712: 6.00 & -1.36 & -1.33 & -1.38 \\
713: 8.00 & -1.26 & -1.25 & -1.27 \\
714: 10.00 & -1.20 & -1.20 & -1.20 \\
715: 12.00 & -1.17 & -1.17 & -1.17
716: \end{tabular}
717:
718: \begin{thebibliography}{99}
719: \bibitem{1} Other failures were its inability to give the brightness of
720: spectral lines and the unsuccessful extension to the $He$ atom.
721:
722: \bibitem{2} M. Jammer, \textit{The Conceptual Development of Quantum
723: Mechanics} (McGraw-Hill, New York, 1966), p. 129.
724:
725: \bibitem{3} Instead of the traditional notation $k$ for the angular quantum
726: number in the old quantum theory, we will use the letter $l$ $(=k)$ to
727: facilitate the connection with quantum mechanics.
728:
729: \bibitem{4} W. Pauli, \textit{``\"{U}ber das Modell des
730: Wasserstoffmolek\"{u}lions,''} Annalen der Physik \textbf{68}, 177-240
731: (1922). A history of Pauli's thesis is given by C. P. Enz, \textit{``No Time
732: to be Brief: A Scientific Biography of Wolfgang Pauli''} (Oxford UP, 2002),
733: pp. 63-74.
734:
735: \bibitem{5} M. C. Gutzwiller, \textit{Chaos in Classical and Quantum
736: Mechanics} (Springer, New York, 1990), p. 36.
737:
738: \bibitem{6} A. Sommerfeld,\textit{\ ``Zur Quantentheorie der
739: Spektrallinien,''} Annalen der Physik, \textbf{51}, 1-94 (1916).
740:
741: \bibitem{7} This length of semiminor axis has been modified from
742: Sommerfeld's original expression, $b_{nl}=(r_{B}/Z)nk$, to achieve agreement
743: with expressions from quantum mechanics.
744:
745: \bibitem{8} If the nuclear charge is uniformly distributed, then $Z^{\prime
746: }(r)=(r/r_{N})^{3}Z$, and the electron's motion inside the nucleus is
747: harmonic.
748:
749: \bibitem{9} M. Bucher, D. Elm and D. P. Siemens, \textit{``Average position
750: in Kepler motion,''} Am. J. Phys. \textbf{66}, 929-930 (1998).
751:
752: \bibitem{10} L. Pauling and E. B. Wilson, \textit{Introduction to Quantum
753: Mechanics} (Dover, New York, 1935) p. 144.
754:
755: \bibitem{11} Although almost all quantum texts use the weaker statement
756: ``at the nucleus.''
757:
758: \bibitem{12} M. Bucher,``The electron inside the nucleus: An almost
759: classical derivation of isotropic hyperfine interaction,'' Eur. J. Phys.
760: \textbf{21}, 19-22 (2000).
761:
762: \bibitem{13} A. Einstein, \textit{``Zum Quantenansatz von Sommerfeld und
763: Epstein,''} Verh. Dtsch. Phys. Ges. \textbf{19}, 82-92 (1917). A summary
764: with modern comments is given by A. D. Stone, ``\textit{Einstein's unknown
765: insight and the problem of quantizing chaos,''} Phys. Today, Aug. 2005, pp.
766: 37-43.
767:
768: \bibitem{14} Eq. (3), introduced as Sommerfeld quantization for
769: familiarity's sake, is also an Einstein quantization condition.
770:
771: \bibitem{15} M. P. Strand and W. P. Reinhardt, \textit{``Semiclassical
772: quantization of the low lying electronic states of }$\mathit{H}_{\mathit{2}%
773: }^{\mathit{+}}$\textit{,''} J. Chem. Phys. \textbf{70}, 3812-27 (1979).
774:
775: \bibitem{16} The (one-center) Runge-Lenz vector, $\mathbf{\Omega }=\mathbf{v%
776: }\times \mathbf{L}-Ze^{2}\mathbf{r}/r$, of a Sommerfeld orbit of angular
777: momentum $\mathbf{L}$ has a magnitude proportional to the ellipse's
778: eccentricity $\varepsilon $ and points from the nucleus toward the
779: perihelion; see J. Morehead, ``\textit{Visualizing the extra symmetry of the
780: Kepler problem,''} Am. J. Phys. \textbf{73}, 234-239 (2005). For a Coulomb
781: oscillator ($L=0$, $\varepsilon =1$), $\mathbf{\Omega }$ oscillates and thus
782: is \textit{not conserved.} Neither is $\Omega _{c}$ of the two-center
783: \textit{Coulomb oscillator} conserved; see H. A. Erikson and E. L. Hill, ``%
784: \textit{A Note on the one-electron state of diatomic molecules,''} Phys.
785: Rev. \textbf{75}, 29-31 (1949); and also C. A. Coulson and A. Joseph, ``%
786: \textit{A constant of the motion for the two-center Kepler problem,''} Int.
787: J. Quantum Chem. \textbf{1} , 337-347 (1967).
788:
789: \bibitem{17} The trajectory will eventually revisit an arbitrarily close
790: neighborhood of any previous point.
791:
792: \bibitem{18} Ref. 5, pp. 207-215.
793:
794: \bibitem{19} L. J. Curtis and D. G. Ellis, ``\textit{Use of the
795: Einstein-Brillouin-Keller action quantization,''} Am. J. Phys. \textbf{72},
796: 1521-1523 (2004).
797:
798: \bibitem{20} K. F. Niessen, \textit{``Zur Quantentheorie des
799: Wasserstoffmolek\"{u}lions,''} Annalen der Physik \textbf{70}, 129-134
800: (1923).
801:
802: \bibitem{21} M. M. Madsen and J. M. Peek, At. Data \textbf{2}, 171 (1971);
803: E. Teller and H. L. Sehlin, in \textit{Physical Chemistry, An Advanced
804: Treatise} (Academic, New York, 1970), Vol. \textbf{5}, p. 35.
805:
806: \bibitem{22} L. de Broglie, \textit{``Ondes et quanta,''} Comptes Rendus
807: \textbf{177}, 507-510 (1923).
808:
809: \bibitem{23} K. Ruedenberg, ``\textit{The physical nature of the chemical
810: bond,''} Rev. Mod. Phys. \textbf{34}, 326-276 (1962).
811: \end{thebibliography}
812:
813: \section{FIGURE CAPTIONS}
814:
815: Fig. 1. Partial trajectory of an extranuclear orbit $XX^{\prime }$ and of a
816: penetrating orbit $PP^{\prime }$ through nucleus $N$. The dotted line $S$
817: shows the major symmetry axis of $XX^{\prime }$.
818:
819: Fig. 2. Line orbit of a Coulomb oscillator with nucleus at origin $0$ and
820: turning points at $\pm \Gamma $.
821:
822: Fig. 3. Axial electron speed $v$ vs. position $z$ of an electron in Coulomb
823: oscillation for the ground state, $n=1$, of a hydrogen atom $H$ ($Z=1$,
824: solid curve) and a helium ion $He^{+}$ ($Z=2$, dashed curve). The area under
825: one wing of each curve represents the radial action, $A=1h$.
826:
827: Fig. 4. Axial Coulomb oscillation in an $H_{2}^{+}$ molecule ion between
828: axial turning points $\pm C$ and through nuclei at $\pm c$; perpendicular
829: oscillation between lateral turning points $\pm B$ and through midpoint $0$.
830:
831: Fig. 5. Axial speed $v$ vs. position $z$ in the ground state of an $%
832: H_{2}^{+} $ molecule ion (bold curve) and, for comparison, of a free $H$
833: atom (thin curve). Circles indicate the axial the positions of the nuclei,
834: here with a \textit{large} separation, $R=6$ $r_{B}$.
835:
836: Fig. 6. Axial speed $v$ vs. position $z$ in the ground state of an $%
837: H_{2}^{+} $ molecule ion (M-shaped bold curve) and, for comparison, of a
838: free H atom ($\Lambda $-shaped thin curve, centered at the right nucleus).
839: The $\cap $ -shaped dotted curve, centered at $0$, shows the lateral speed $%
840: u $ vs. the perpendicular position $y$ in the molecule. Circles indicate the
841: axial positions of the nuclei, here with a \textit{small} separation, $R=2$ $%
842: r_{B}$.
843:
844: Fig. 7. Dependence of energies of $H_{2}^{+}$ on internuclear distance $R$.
845: \textit{Bottom:} Electronic energy $E_{el}$ from quantum mechanics (QM,
846: solid curve), primitive semiclassical quantization by Strand and Reinhardt
847: (SR, $\bigcirc $) and the present Coulomb-oscillator approach (CO, +).
848: \textit{Middle}: Total energy $E$ of the $1s\Sigma _{g}$ ground state from
849: QM (dashed curve) with minimum ($\bullet $), values by SR ($\square $) and
850: CO ($\times $) and their average ($\diamond $), energy of a free $H$ atom
851: (dashed horizontal line). \textit{Top}: Total energy $E$ of the $2p\Pi _{u}$
852: state from QM (dotted curve) with minimum ($\bullet $), and historical value
853: by Pauli and Niessen ($\bigtriangleup $).
854:
855: Fig. 8. de Broglie wave of the free $H$ atom Coulomb oscillator (curve) and
856: of the $H_{2}^{+}$ Coulomb oscillator (small circles) for the same proton
857: separation as in Fig. 5, $R=6$ $r_{B}$. Large circles indicate the axial
858: positions of the nuclei.
859:
860: Fig. 9. de Broglie wave of the $H_{2}^{+}$ Coulomb oscillator along axial
861: distance $OC$ from Fig. 4 (right side of graph) and lateral distance $OB$
862: (left side). Large circles indicate the axial positions of the nuclei, here
863: with the same separation as in Fig. 6, $R=2$ $r_{B}$.
864:
865: Fig. 10. Same as Fig. 9 but for the proton separation $R^{*}=1.38$ $r_{B}$
866: where the semiclasssical energy is exact.
867:
868: \end{document}
869: