1: % LateX file prepared by Li-Jen Chen
2:
3: %%%%%%%%%%%%%%%%%%%% USER DEFINED VARIABLES %%%%%%%%%%%%%%%%%%%%%%%%%%
4:
5: %%%%%%%%%% Journals %%%%%%%%%%
6: \def\GRL {{\it Geophys.\ Res.\ Lett.\/}}
7: \def\GMono {{\it Geophys.\ Monog.\/}}
8: \def\JGR {{\it J. Geophys.\ Res.\/}}
9: \def\PF {{\it Phys. Fluids.}}
10: \def\PRL {{\it Phys. Rev. Lett.}}
11: \def\PL {{\it Phys. Lett.}}
12: \def\PRep {{\it Phys. Rep.}}
13: \def\PR {{\it Phys. Rev.}}
14: \def\PSr {{\it Phys. Scr.}}
15: \def\NPG {{\it Nonlinear Processes in Geophys.}}
16: \def\bra {{\langle}}
17: \def\ket {{\rangle}}
18: \def\erf {\mathop{\rm erf}\nolimits}
19: \def\sech {\mathop{\rm sech}\nolimits}
20: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
21:
22: \documentstyle[aps,prl,epsf,twocolumn]{revtex}
23:
24: \draft
25:
26: \begin{document}
27:
28: \wideabs{
29:
30: \title{Trapped and Passing Electrons in BGK Solitary Waves}
31: \author{Li-Jen Chen and George K. Parks*}
32: \address{Physics Department, University of Washington, Seattle, WA 98195}
33: \address{*Also at Space Science Laboratory, University of California, Berkeley}
34:
35: \maketitle
36:
37: \begin{abstract}
38: This paper reexamines the physical roles of trapped and passing electrons
39: in Bernstein-Greene-Kruskal (BGK) electron
40: solitary waves, also called the BGK electron holes (EH).
41: It is shown that the shielding of the positive core is
42: achieved by electrons trapped and oscillating inside
43: the potential energy trough, instead of the thermal screening
44: from the ambient plasma as previously thought.
45: The size of a BGK EH is therefore not
46: restricted to be greater than the Debye length $\lambda_D$.
47: \end{abstract}
48:
49: \pacs{PACS numbers: 52.35.Sb, 52.35.Mw, 52.35.Fp}
50: }
51:
52: %MAIN TEXT
53:
54: In 1957, Bernstein, Greene and Kruskal \cite{BGK57} solved the one-dimensional,
55: time-independent Vlasov-Poisson equations and obtained the general solutions
56: for electrostatic nonlinear traveling waves, including solitary potential pulses.
57: Their derivation emphasized the special role played by the particles trapped
58: in the potential energy troughs. They demonstrated that one
59: could construct waves of arbitrary shapes by assigning the distribution of
60: trapped particles suitable for the desired wave form.
61:
62: In 1967, Roberts and Berk \cite{Roberts67} provided a quasi-particle picture
63: for the electron phase space holes (EH) based on the results of a numerical
64: experiment on two-stream instability.
65: They followed the evolution of the electron phase space
66: boundary between $f=0$ and $f=1$
67: using the time-dependent Vlasov-Poisson equations with the
68: ``water-bag'' model in which the distribution function only takes
69: discrete values, 0 and 1. They interpreted the elliptical empty
70: ($f=0$) region associated
71: with a positive charge observed in the late stage of the nonlinear development
72: as a BGK EH. In order to explain the coalescence of neighboring EHs,
73: a negative effective mass was assigned to each EH to compensate for the
74: Coulomb repulsion of two positive EHs.
75: Thus, they suggested a quasi-particle picture that BGK EHs have positive charge
76: and negative mass. This picture is in use even today to interpret results in
77: computer simulations of EH disruptions
78: \cite{Saeki98}, and to model the mutual interaction of
79: electrostatic solitary waves in space plasma \cite{Krasovsky99}.
80:
81: It was not until 1979 that BGK EHs were experimentally realized by the Ris\mbox{\o}
82: laboratory experiments \cite{Saeki79,Lynov79}. By applying large amplitude
83: potential pulses in a plasma-loaded wave guide, solitary potential pulses were excited,
84: including EHs and Korteweg-de Vries (KdV) solitons \cite{Washimi66}.
85: Investigations of the mutual interaction of EHs showed that two EHs
86: close to each other would coalesce if they have almost equal velocity and
87: they would pass through each other if their relative velocity was large \cite{Saeki79}.
88: The coalescence was interpreted in terms of the positive EH picture derived earlier
89: \cite{Saeki79,Lynov79}.
90:
91: The analytical work that followed the Ris\mbox{\o} experiments mainly focused on constructing
92: solutions and obtaining the corresponding width-amplitude relations to facilitate the
93: comparison between BGK EHs and KdV solitons \cite{Turikov84,Schamel86,Lynov85}.
94: In addressing the quasi-particle picture of EHs,
95: Schamel \cite{Schamel86} concluded that EHs were positively charged, screened by the
96: ambient electrons over many Debye lengths ($\lambda_D$), and
97: had negative mass \cite{Schamel86}.
98: This conclusion supports the positive EH picture previously obtained from the numerical
99: experiments \cite{Roberts67,Berk70} and that
100: the minimum size of EHs of several $\lambda_D$ is a consequence of thermal screening by
101: the ambient electrons.
102:
103: Turikov \cite{Turikov84} followed the BGK approach and constructed the trapped
104: electron distribution for a Maxwellian ambient electron distribution and
105: various solitary potential profiles. He restricted his study to
106: BGK EHs with phase space density zero in a hole center and the
107: results showed that the potential width increases with increasing amplitude.
108: This behavior is different from that of the KdV solitons whose width decreases
109: with increasing amplitude.
110: He also numerically simulated the temporal evolution of the EHs for different Mach
111: numbers to study the EH stability and found that EHs are quasi-stable
112: for Mach numbers less than 2.
113: One of the main conclusions that Turikov made was that EHs are purely
114: kinetic nonlinear objects in which trapped electrons play an important
115: role, but exactly what physical role trapped electrons played was not
116: addressed.
117:
118: Space-borne experiments now show that electrostatic solitary waves (ESW) are
119: ubiquitous in Earth's magnetospheric
120: boundaries, shock, geomagnetic tail and auroral ionosphere
121: \cite{Bostrom88,Matsumoto94,Franz98,Ergun98,Ergun99,Bale98}.
122: While detailed properties of these solitary waves are continuously being
123: studied, it has been shown that in a number of cases the ESWs have
124: features that are consistent with BGK electron \cite{Muschietti99}
125: and ion \cite{Malkki89} mode solitary waves.
126: A statistical study by FAST satellite observations \cite{Ergun98,Ergun99} in the auroral
127: ionosphere has revealed that solitary pulses with a positive potential typically
128: have a Gaussian
129: half width ranging from less than one $\lambda_D$ to several $\lambda_D$ with
130: a mean of $1.80 \lambda_D$ and a standard deviation of $1.13 \lambda_D$.
131: However, the statistical analysis strongly favored large amplitude pulses
132: \cite{Ergun98,Ergun99} and smaller EHs if they existed were not sampled.
133: This question of how small EHs can be
134: is an important issue associated with how a collisionless plasma supports
135: nonlinear waves and needs to be further investigated.
136:
137: To examine the physical roles played by trapped and passing
138: electrons, we start first with simple physical arguments and then
139: perform analytical
140: calculations using the same formulation used by Turikov \cite{Turikov84},
141: except that we relax the restriction of empty-centered EHs and obtain a more general
142: width-amplitude relation. The number density profiles of trapped and
143: passing electrons are calculated to quantify the separate contributions from
144: trapped and passing electrons to the charge density.
145: Implications of our results to the positive EH picture and minimum
146: size of BGK EHs are made.
147:
148: We first discuss heuristically the behavior of
149: electrons in the vicinity of a potential pulse. Figures \ref{f1}(a)-\ref{f1}(c)
150: show the general form of a positive solitary potential pulse ($\phi(x)$),
151: the corresponding bipolar electric field ($E=-\partial \phi/\partial x$)
152: and the total charge density ($\rho=-\partial^2 \phi/\partial x^2$).
153: The charge
154: density is positive at the core, negative at the boundary, and zero outside
155: the solitary potential.
156: Figure \ref{f1}(d) shows the potential energy trough with an electron passing by
157: (open circle) and a trapped electron (solid circle) at its turning point.
158: Consider the phase space trajectories of electrons passing by the potential
159: and those that are trapped in the potential shown in Figure \ref{f1}(e).
160: The dashed line marked by electrons with zero total energy
161: is the boundary of the trapping region inside which electrons are trapped
162: and outside which electrons are untrapped.
163: The total energy, $w=\frac{m}{2}v^2-e\phi$, is a constant of motion.
164: A passing electron ($w>0$) moves with a constant velocity outside
165: the potential and the speed increases when it encounters the
166: potential pulse and then decreases back to its original value as it moves away.
167: A trapped electron ($w<0$) bounces back and forth between its two turning points
168: in the potential.
169: Since there is no source or sink for the particles, the density is inversely
170: proportional to the velocity.
171: We can thus deduce that the density of passing electrons is constant outside
172: and becomes
173: smaller as $\phi$ increases. No excess negative charge results from the
174: passing electrons. On the other hand, trapped electrons have density maxima
175: at their turning points, and so must be responsible for the excess negative
176: charge.
177: The charge density variation (Figure \ref{f1}(c)) needed to be self-consistent with
178: the solitary potential pulse is thus a net balance of the negative charge from
179: trapped electrons
180: and the positive charge due to the density decrease of passing electrons
181: since the ion density is assumed uniform.
182: From this simple picture, one can see that in BGK solitary waves it is
183: the trapped electrons traveling with the solitary potential that screen
184: out the positive core.
185: Our picture is different from the picture of a positive object in
186: a plasma whose screening is achieved by the thermal motion of the plasma
187: (Debye shielding).
188:
189: The entire trapping region that consists of the total electron density
190: enhancement at the flanks and depletion at the core is a physical entity
191: produced by the self-consistent interaction between the plasma particles and
192: the solitary potential pulse. This defines the physical identity of one
193: BGK EH.
194: The total charge of the entire trapping region is zero, and therefore it
195: follows that two separated BGK EHs do not interact and the concept of
196: negative mass is not needed.
197:
198: We now use the approach formulated by BGK \cite{BGK57} to quantify
199: the above arguments
200: and further demonstrate that the results are independent of the strength of the
201: nonlinearity.
202: The time-independent, coupled Vlasov and Poisson equations with the assumption of a
203: uniform neutralizing ion background take the following form:
204: \begin{equation}
205: v\frac{\partial f(v, x)}{\partial x}+\frac{1}{2}\frac{\partial \phi}
206: {\partial x}\frac{\partial f(v, x)}{\partial v}=0, \label{Veq}
207: \end{equation}
208: \begin{equation}
209: \frac{\partial^2 \phi}{\partial x^2}=\int_{-\infty}^{\infty}f(v, x)dv-1,
210: \label{Peq}
211: \end{equation}
212: where f is the electron distribution function and the quantities have been
213: normalized with the units of the Debye length $\lambda_D$,
214: the ambient electron thermal energy $T_e$, and the electron thermal velocity
215: $v_t=\sqrt{2T_e/m}$.
216: The total energy $w=v^2-\phi$ under this convention.
217: Any $f=f(w)$ is a solution to Eq. \ref{Veq} as can be readily verified.
218: Recognizing this, Eq. \ref{Peq} can be
219: re-written in the following form,
220: \begin{equation}
221: \frac{\partial^2 \phi}{\partial x^2}=\int_{-\phi}^0 dw\frac{f_{tr}(w)}
222: {2\sqrt{w+\phi}}+\int_0^{\infty}dw \frac{f_p(w)}{2\sqrt{w+\phi}}-1, \label{phi2}
223: \end{equation}
224: where $f_{tr}(w)$ and $f_p(w)$ are the trapped and passing electron
225: phase space densities at energy $w$, respectively.
226: The first integral on the RHS of Eq. \ref{phi2} is the trapped
227: electron density, and the second integral the passing electron density.
228: Prescribe the solitary potential as a Gaussian,
229: \begin{equation}
230: \phi(\psi, \delta, x)=\psi \exp{(-x^2/2\delta^2)},
231: \end{equation}
232: and the passing electron distribution a Maxwellian where
233: the density has been normalized to 1 outside the solitary potential,
234: \begin{equation}
235: f_p(w)=\frac{2}{\sqrt{\pi}}\exp{(-w)}.
236: \end{equation}
237: Following the BGK approach, we obtain the trapped electron distribution,
238: \begin{eqnarray}
239: f_{tr}(\psi, \delta, w) & = & \frac{4\sqrt{-w}}{\pi \delta^2}
240: \left[1-2\ln{(\frac{-4w}{\psi})}\right] \nonumber \\
241: & & +\frac{2\exp{(-w)}}{\sqrt{\pi}} \left[1-\erf(\sqrt{-w}) \right].
242: \label{ftr2}
243: \end{eqnarray}
244: The first term in $f_{tr}$ stems from $\partial^2\phi/\partial x^2$ term
245: in Eq. \ref{phi2} and has a single peak at $w=\frac{-\psi}{4e^{3/2}}$.
246: This term is 0 at $w=0^-$, goes negative at $w=-\psi$, and
247: will always be single peaked even for other bell-shaped
248: solitary potentials (for example, $\sech^2(x/\delta)$ and $\sech^4(x/\delta)$,
249: see Figure 4 of \cite{Turikov84} for the special case of empty-centered
250: EHs). Although the peak location may vary, it will not be at the end
251: points, 0 and $-\psi$.
252: The second term arising from the integral of the passing
253: electron distribution decreases monotonically from $w=0^-$ to $w=-\psi$.
254: The end point behavior of the two terms implies that
255: $f_{tr}(w=0^-) > f_{tr}(w=-\psi)$.
256: Combining the behavior of the two terms in $f_{tr}$, it can be concluded that
257: $f_{tr}(0>w\geq -\psi)\geq f_{tr}(w=-\psi)$.
258: This feature of $f_{tr}$ is essential in making
259: a solitary pulse, and it manifests itself at the peak of the potential as
260: two counterstreaming beams.
261: For the solution to be physical, $f_{tr}$ has to be nonnegative.
262: $f_{tr}(\psi, \delta, w=-\psi)\geq 0$ suffices this requirement,
263: yielding the width-amplitude ($\delta$-$\psi$) relation,
264: \begin{equation}
265: \delta \geq \left[\frac{2 \sqrt{\psi}(2\ln4-1)}{\sqrt{\pi}e^{\psi}
266: [1-\erf(\sqrt{\psi})]}\right]^{1/2}. \label{WA}
267: \end{equation}
268: The equal sign in Ineq. \ref{WA} corresponds to the case of empty-centered EHs
269: studied by Turikov \cite{Turikov84}.
270: We plot Ineq. \ref{WA} in Figure \ref{f2}.
271: A point in the shaded region represents an allowed EH with a given $\psi$ and
272: $\delta$. The shaded region includes all of the
273: allowed $\psi$ and $\delta$ for the range of values shown.
274: For a fixed $\delta$, all $\psi\leq\psi_0$ are
275: allowed, where $\psi_0$ is such that
276: $f_{tr}(\psi_0,\delta,w=-\psi_0)=0$; while for a fixed $\psi$, all
277: $\delta\geq\delta_0$ are allowed,
278: where $\delta_0$ is such that $f_{tr}(\psi,\delta_0,w=-\psi)=0$.
279: The corresponding physical meaning is that empty-centered EHs give the largest
280: amplitude ($\psi_0$) for a fixed width, and the smallest width ($\delta_0$)
281: for a fixed amplitude as comparing to EHs that are not empty-centered.
282: Note that by distributing a finite number of
283: electrons at rest at the bottom of the potential energy trough,
284: the plasma supports smaller amplitude or larger width
285: pulses than the amplitude and width of empty-centered EHs.
286: This inequality relation is dramatically different from that of KdV
287: solitons whose width-amplitude relation is a one-to-one mapping.
288:
289: With $f_p$ and $f_{tr}$, we can now calculate the passing
290: and trapped electron densities separately and obtain
291: \[
292: n_p(x)=\exp{(\phi)}\left[1-\erf(\sqrt{\phi})\right],
293: \]
294: \[
295: n_{tr}(x)={\frac{-\phi \left[1+2\ln(\phi/\psi)\right]}{\delta^2}}+1-
296: \exp{(\phi)}\left[1-\erf(\sqrt{\phi})\right].
297: \]
298: Another way to obtain $n_{tr}$ comes directly from
299: solving Eq. \ref{phi2} as the first term on the RHS is exactly
300: $n_{tr}$.
301: Solving Eq. \ref{phi2} for $n_{tr}$ is simpler, but
302: without the knowledge of $f_{tr}$, one is not guaranteed whether the
303: particular set of ($\psi$, $\delta$) is physically allowed ($f_{tr}\geq0$).
304: Note that even for $\phi\ll 1$, $n_p\sim 1-2\sqrt{\phi/\pi}$ is different from
305: the leading terms of a Boltzmann distribution, $e^{\phi}$. The physical
306: meaning is that under self-consistent interaction of electrons and the
307: solitary potential, electrons in the vicinity of the potential are
308: not in local thermal equilibrium, in contrast to the starting point
309: of local thermal equilibrium in obtaining the
310: thermal screening length \cite{Debye23}.
311:
312: To study the contributions from trapped and passing electrons to the
313: charge density ($-\partial^2 \phi/\partial x^2$) and how such contributions
314: are affected by various parameters, we show in Figure \ref{f3} plots
315: of $n_{tr}$ and $n_p$ and the charge density
316: $\rho$ as a function of $x$ for several values of $\psi$ and $\delta$.
317: Figure \ref{f3} (a) and (b) plot $100\times n_{tr}(x)$, $100\times
318: [n_p(x)-1]$, and $100\times \rho(x)$ for
319: ($\psi$, $\delta$)=($2\times 10^{-5}$,$0.1$). For an ambient plasma with
320: $T_e=700 eV$ and $\lambda_D=100 m$ as found at ionospheric heights by FAST satellite
321: in the environment of BGK EHs \cite{Ergun99}, this case corresponds to $\psi=
322: 1.4\times 10^{-2} V$ and $\delta=10 m$. As shown, in this weakly nonlinear case,
323: the maximum perturbation in $n_p$ is only $0.5\%$ and in $n_{tr}$ $0.4\%$.
324: The perturbation in the charge density $\rho$ is $\leq 0.2\%$, and occurs all
325: within one $\lambda_D$.
326:
327: $n_p(x)$, $n_{tr}(x)$, and
328: $\rho(x)$ for ($\psi$, $\delta$)=(5, 4.4) in Figure \ref{f3} (c) and (d).
329: This choice corresponds to a point close to the curve
330: $f_{tr}(w=-\psi)=0$ in Figure 2 and is an extremely nonlinear case.
331: One can see that the total charge density perturbation goes
332: $\sim 10\%$ negative and $\sim 25\%$ positive, corresponding respectively
333: to electron density enhancement and depletion.
334: With similar format, Figure \ref{f3} (e) and (f) plots a case with same $\delta$ and
335: $\psi=1$ to illustrate the change in $n_p$, $n_{tr}$, and $\rho$
336: of an EH with equal width but smaller amplitude. By locating this case
337: in Figure \ref{f2}, one notices that farther away from the $f_{tr}(w=-\psi)=0$
338: curve, the dip in $n_{tr}$ is filled up and the charge
339: density perturbation only increases to $5\%$ positive and $2\%$ negative.
340:
341: These examples demonstrate how trapped electrons
342: produce negative charge density perturbations and passing electrons
343: positive charge density perturbations owing to the decrease in
344: their number density.
345: It is always true that $n_{tr}\geq 0$, since the number density cannot be
346: negative,
347: and therefore trapped electrons always contribute to negative charge density
348: regardless of the strength of the nonlinearity.
349: This result disagrees with
350: the picture that the positive core is due to
351: a deficit of deeply trapped electrons, and that this positive core is screened
352: out by the ambient electrons \cite{Schamel86}. It is also different from the
353: conclusion
354: that the trapped electrons are screened out by the resonant or nonresonant
355: passing electrons depending on the EH velocity in \cite{Krasovsky97}.
356:
357: We now turn to the issue of minimum size of EHs.
358: From the illustrations of Figure \ref{f3}, one sees that
359: the charge density variation is the net balance
360: of the negative charge produced by trapped electrons and the positive charge
361: density produced by a depletion of passing electrons inside the solitary
362: potential.
363: The trapped electrons must distribute and oscillate in
364: such a way to yield the desired negative charge at the flanks
365: to shield out the positive core.
366: The entire solitary object is a self-consistent and self-sustained object with
367: zero total charge and does not require screening from the ambient plasma.
368: Thus, the size of EHs are not restricted to be greater than $\lambda_D$.
369:
370: In summary, we demonstrated that the positive core of the EH is
371: shielded exclusively by trapped electrons oscillating between
372: their turning points and resulting in the excess negative charge.
373: A BGK EH consists of an electron density enhanced region and
374: depletion region, and this means that the total charge for a BGK EH is zero.
375: It thus follows that two separated EHs do not interact and the concept of
376: negative mass is not needed. There does not exist
377: a minimum size for BGK EHs since they do not rely on the thermal
378: screening from the ambient plasma. These features are independent of the particular
379: choice of potential profile and passing electron distribution, and also
380: independent of the strength of nonlinearity, because the only principle
381: upon which the arguments are built is conservation of charge
382: (the continuity equation).
383: The restriction to empty-centered EHs which was used previously is relaxed
384: to obtain a more general width-amplitude relation of
385: an inequality form.
386:
387: Finally, note that the arguments and results we obtained for BGK
388: electron solitary waves can be
389: analogously applied to BGK ion solitary waves.
390:
391: One of the authors (Chen) is grateful to Bill Peria for the discussions
392: on the electric field experiment onboard FAST satellite.
393: The research at the University of Washington
394: is supported in part by NASA grants NAG5-3170 and NAG5-26580.
395:
396: \begin{thebibliography}{10}
397: \bibitem{BGK57}
398: I. B. Bernstein, J. M. Greene, and M. D. Kruskal, \PR, {\bf 108}, 546 (1957)
399: \bibitem{Roberts67}
400: K. V. Roberts and H. L. Berk, \PRL, {\bf 19}, 297 (1967)
401: \bibitem{Saeki98}
402: K. Saeki and H. Genma, \PRL, {\bf 80}, 1224 (1998)
403: \bibitem{Krasovsky99}
404: V. L. Krasovsky, H. Matsumoto, and Y. Omura, \NPG, {\bf 6}, 205 (1999)
405: \bibitem{Saeki79}
406: K. Saeki, et al., \PRL, {\bf 42}, 501 (1979)
407: \bibitem{Lynov79}
408: J. P. Lynov, et al. \PSr, {\bf 20}, 328 (1979)
409: \bibitem{Washimi66}
410: H. Washimi and T. Taniuti, \PRL, {\bf 17}, 996 (1966)
411: \bibitem{Turikov84}
412: V. A. Turikov, \PSr, {\bf 30}, 73 (1984)
413: \bibitem{Schamel86}
414: H. Schamel, \PRep, {\bf 140}, pp172-173 (1986)
415: \bibitem{Lynov85}
416: J. P. Lynov, et al., \PSr, {\bf 31}, 596 (1985)
417: \bibitem{Berk70}
418: H. L. Berk, C. E. Nielsen, and K. V. Roberts, \PF, {\bf 13}, 980 (1970)
419: \bibitem{Bostrom88}
420: R. Bostr\"om, et al., \PRL, {\bf 61}, 82 (1988)
421: \bibitem{Matsumoto94}
422: H. Matsumoto, et al., \GRL, {\bf 21}, 2915 (1994)
423: \bibitem{Franz98}
424: J. R. Franz and P. M. Kintner, J. S. Pickett, \GRL, {\bf 25},
425: 1277, (1998)
426: \bibitem{Ergun98}
427: R. E. Ergun, et al., \PRL, {\bf 81}, 826 (1998)
428: \bibitem{Ergun99}
429: R. E. Ergun, et al., \NPG, {\bf 6}, 187 (1999)
430: \bibitem{Bale98}
431: S. D. Bale, et al., \GRL, {\bf 25}, 2929 (1998)
432: \bibitem{Muschietti99}
433: L. Muschietti, et al., \GRL, {\bf 26}, 1093 (1999)
434: \bibitem{Malkki89}
435: A. M\"alkki, et al., \PSr, {\bf 39}, 787 (1989)
436: \bibitem{Debye23}
437: P. Debye and E. H\"uckel, Physik. Zeits. {\bf 24}, 185, (1923)
438: \bibitem{Krasovsky97}
439: V. L. Krasovsky, H. Matsumoto, and Y. Omura, \JGR, {\bf 102}, 22131 (1997)
440: \end{thebibliography}
441:
442: \begin{figure}
443: \epsfxsize=8.0cm
444: \epsffile{Fig1.eps}
445: \caption{Please see the text for explanations.}
446: \label{f1}
447: \end{figure}
448:
449: \begin{figure}
450: \epsfxsize=8.0cm
451: \epsffile{Fig2.eps}
452: \caption{the width-amplitude relation of BGK EHs that are not restricted to be empty-centered
453: for a Gaussian potential and Maxwellian ambient electron distribution}
454: \label{f2}
455: \end{figure}
456:
457: \begin{figure}
458: \epsfxsize=8.5cm
459: \epsffile{Fig3.eps}
460: \caption{Trapped electron density ($n_{tr}$), passing electron density ($n_p$),
461: and charge density ($\rho$) for ($\psi$,$\delta)=(2\times 10^{-5}, 0.1$) in
462: (a) and (b), ($\psi$,$\delta)=(5, 4.4)$ in (c) and (d), and ($\psi$,$\delta)=
463: (1, 4.4)$ in (e) and (f). }
464: \label{f3}
465: \end{figure}
466:
467: \end{document}
468: