1:
2:
3: \documentstyle[12pt,aps,epsf]{revtex}
4: \renewcommand{\baselinestretch}{1.2}
5: \input{epsf}
6: \begin{document}
7:
8: \title{Saturable Absorber, Coherent Population Oscillations,
9: \vskip 1mm and "Slow Light"}
10: \author{V.S.Zapasskii and G.G.Kozlov}
11:
12: \maketitle
13: \hskip20pt
14: \vskip20pt
15: \hskip100pt {\it e}-mail: gkozlov@photonics.phys.spbu.ru
16: \vskip20pt
17: All-Russia Research Center "Vavilov State Optical Institute",
18: St. Petersburg, 199034 Russia
19: \begin{abstract}
20: The paper presents a critical analysis of publications
21: devoted to one of the methods for production of the so-called
22: "slow light" (the light with an anomalously low group velocity)
23: due to extremely high local steepness of the dispersion curve of the medium. The
24: method in point employs for this purpose the effect of coherent
25: population oscillations accompanied by burning of a narrow
26: spectral hole in the homogeneously broadened absorption spectrum. Physical model of the effect proposed in the studies
27: under consideration is based on analysis of response of a
28: nonlinear medium to a low-frequency intensity modulation of
29: the incident light beam. We show that all the observations
30: described in these papers can be easily interpreted in the
31: framework of the simplest model of saturable absorber and have
32: nothing to do with the effects of hole burning and group velocity
33: reduction.
34: \end{abstract}
35: \newpage
36: \section{INTRODUCTION}
37:
38:
39: After publication of the first impressive results on group
40: velocity reduction by several orders of magnitude \cite{1,2}, the
41: effects of slowing down, stoppage, storage, and release of light
42: attracted attention of a great number of researchers. The
43: experiment \cite{2}, which demonstrated the light velocity reduction
44: by more than 7 orders of magnitude, was performed with
45: sodium atoms at the temperatures close to that of the Bose-Einstein condensation (hundreds of nK) and was based on the
46: effect of electromagnetically induced transparency. The paper \cite{2} initiated a great
47: number of publications, with some of them, in particular,
48: claiming observation of the light velocity reduction of similar
49: scale under much less exotic experimental conditions (e.g., at
50: room temperature). To create a steep dispersion of the refractive index,
51: along with the effect of electromagnetically induced
52: transparency, there has been used the effect of coherent
53: population oscillations (CPO). Observation of the CPO-based
54: hole burning in the absorption spectra of ruby and
55: alexandrite crystals was reported in \cite{3,4}. Then, in \cite{5,6}, there has been
56: announced observation, in these crystals, of "slow" and "fast" light, i.e., the light with "ultralow" and, in this particular
57: case, "negative" group velocity. More recently, other
58: publications have appeared that supported the proposed
59: interpretation \cite{7,8}, and, as a result, the CPO effect was included into the
60: arsenal of the "slow light" methods (see, e.g.,
61: \cite{9,10}).
62:
63: The goal of this publication is to show that the light-pulse delay
64: detected in the above experiments is a well known effect of nonlinear
65: optics. It has nothing in common with the CPO effect
66: and is not connected with the group velocity reduction
67: in the medium. In our opinion, the experimental data of \cite{3,4,5,6} provide no evidence for the "slow light" observation, and their interpretation is erroneous.
68:
69: \section{THE EFFECT OF COHERENT POPULATION
70: OSCILLATIONS}
71:
72: The role of population oscillations in the coupling of the light
73: waves with close frequencies propagating in a resonant
74: nonlinear medium was first noted, as far as we know, by W.
75: Lamb in 1964 \cite{11}. Analysis of the properties of a saturable absorber
76: performed by Schwartz and Tan in 1967 \cite{12} has shown that, in
77: the presence of a strong (saturating) pump wave, a weak probe
78: wave scanned in the vicinity of the strong wave frequency
79: displays an apparent dip in the absorption spectrum with the center at
80: the pump frequency and with the width
81: corresponding to the inverse absorption relaxation time. The
82: effect had a simple and visual physical interpretation. The
83: combined action of two monochromatic waves, due to their
84: interference, gives rise to a modulation of the total field intensity at
85: the beat frequency. When this frequency appears to be
86: sufficiently low (comparable with the inverse population
87: relaxation time or lower), the light intensity modulation is
88: transformed into modulation of populations (and absorption) of
89: the optically active centers. Under these conditions, interaction
90: of the monochromatic pump wave with the oscillating
91: susceptibility of the medium destroys its monochromaticity and
92: gives birth to sidebands, with one of them being exactly
93: coincident in frequency with the probe wave and thus contributing to
94: its intensity. This is revealed as an apparent bleaching of the
95: medium (or as a dip in the absorption spectrum) in the vicinity
96: of the pump frequency. The width of such a dip is, as a rule,
97: smaller than homogeneous width of the optical transition. For
98: this reason, this effect was often considered as "hole burning" in
99: a homogeneously broadened spectrum.
100:
101: The interest to the CPO effect in 70s of the last century was
102: stimulated, on the one hand, by its role in the laser emission
103: dynamics and, on the other, by the possibility to use it for
104: getting information about lifetimes of excited states in
105: nonluminescent or rapidly relaxing systems. In those years, the CPO effect
106: has been thoroughly studied both theoretically (see, e.g.,
107: \cite{13,14}) and experimentally \cite{15,16}. The experimental studies
108: mainly demonstrated applicability of this spectroscopic phenomenon to
109: kinetic measurements. In spite of the fact that this effect is linear with respect to the probe wave and is phenomenologically close to the well-known effect of hole-burning in
110: the inhomogeneously broadened spectra, it results from the multiwave interaction and is, evidently, of essentially different nature.
111:
112: The conclusion about observation of a narrow spectral dip ($\Delta\nu \approx 37$
113: Hz) in the homogeneously broadened absorption
114: spectrum of ruby crystal was made in \cite{3} based on analysis
115: of the crystal's response to the low-frequency intensity
116: modulation of the light beam (Ar$^+$-laser). The dip was
117: treated as a usual spectral feature in the complex refractive
118: index of the medium. In \cite{5}, to this narrow spectral dip, in
119: conformity with the Kramers-Kronig relations, was assigned an
120: extremely steep slope of the dispersion curve, and the
121: observation of "slow light" with a group velocity of 57.5 $\pm$
122: 0.5 m/s was reported. Similar experiments performed with
123: alexandrite crystals were reported in \cite{4,6}. This crystal,
124: behaved, at certain wavelengths, as an {\it inverse} saturable
125: absorber (with the absorption growing with light intensity), and the
126: CPO effect, in this crystal, resulted in "anti-hole burning" in
127: the absorption spectrum. The experimental observations of \cite{6} have
128: been interpreted in terms of "negative" group velocity of light
129: ("fast" light). These papers provided the basis for a new
130: approach to the problem of implementation of the
131: "slow" and "fast" light. To evaluate the novelty and validity of
132: this interpretation, let us turn to basic properties of a saturable
133: absorber.
134:
135: \section{SATURABLE ABSORBER}
136:
137: Characteristic properties of a saturable absorber attracted attention of
138: researchers in the early laser epoch, in 60s, when the
139: phthalocyanine dyes were found to be efficient for laser Q-switching. In
140: those years, their temporal and intensity-related characteristics
141: have been studied in sufficient detail \cite{17,18,19,20}. The simplest
142: saturable absorber is known to be well modeled by a two-level
143: system with a high dephasing rate $T_2^{-1} (T_2^{-1} >> T_1^{-1}, \Omega$,
144: where $T_1$ is the population relaxation time and
145: $\Omega$ is the Rabi frequency). As a good example of the
146: saturable absorber may also serve ruby crystal (Fig. 1a). The
147: energy structure of chromium ion in ruby is not two-level, but a
148: high relaxation rate of the excited state $^4F_2$ (($|b\rangle$ in Fig. 1) to
149: the metastable levels $2\bar{A}$ and $\bar{E}$ ($|c\rangle$) makes it possible to
150: neglect the coherent transient processes practically at any
151: reasonable excitation power.
152:
153: In the general case (neglecting the effects of propagation),
154: dependence of intensity of the light transmitted through a
155: saturable absorber $I_{out}$ on its intensity at the entrance $I$
156: can be presented in the form
157: \begin{equation}
158: I_{out}= \ae(I,t)I,
159: \end{equation}
160:
161: where the transmissivity factor $\ae(I,t)$ controls all the dynamic
162: and intensity-related characteristics of the saturable absorber.
163:
164: Under nonstationary conditions, when the light intensity varies
165: in time, the dynamics of transmission of the saturable absorber
166: is dedescribed by the equation
167:
168: \begin{equation}
169: \dot{\ae}={\ae_{eq}-\ae\over\tau}
170: \end{equation}
171:
172: Here, $\ae_{eq}$ is the equilibrium value of $\ae$
173: corresponding to the current light intensity, and $\tau$ is a
174: time constant (at low light intensities $\tau \approx T_1$. When the light intensity is not too high, the
175: dependence $\ae_{eq}(I)$ can
176: be presented in the form
177:
178: \begin{equation}
179: \ae_{eq}(I)=\ae_0-\ae_1 I, \hskip10mm \ae_1 I<<\ae_0.
180: \end{equation}
181:
182: As the light intensity increases, the nonlinear is usually bleached. In this case, the parameter $\ae_1$ is negative.
183: There are situations, however, when absorption of the optical
184: medium increases with increasing light intensity, rather than
185: decreases. This situation is typical for certain schemes of optical
186: orientation and is realized, in particular, in the alexandrite
187: crystal mentioned above, where, due to efficient excited-state
188: absorption, chromium ions in the excited metastable state may absorb
189: light, at the wavelength of excitation, stronger than those in the
190: ground state (Fig. 1b). For these {\it inverse}
191: saturable absorber, $\ae_1>0$.
192:
193: Solving Eqs. (1)--(3) for a stepwise change of the incident light
194: intensity, one can easily make sure that the time dependence of
195: the transmitted light intensity, in this case, contains two
196: components - an inertialess (stepwise) jump and an exponential component,
197: reflecting dynamics of establishment of steady-state
198: populations of the medium. The sign of the exponential
199: component, evidently, depends on the sign of $\ae_1$
200: (Fig. 2). In virtue of linearity of the Fourier-transform, the appropriate
201: response in the frequency domain
202: will also contain two components - a frequency-
203: independent ("white") pedestal and a Lorentz-wise peak in the range of
204: zero frequencies with a width of $\sim 1/\tau$. This dependence
205: can be easily obtained in analytical form by calculating the
206: response of the saturable absorber to the light beam with
207: harmonically modulated intensity
208:
209: \begin{equation}
210: I=I_0+I_1\exp(i\omega t),
211: \end{equation}
212:
213: >From Eqs. (1)--(3), neglecting the terms $\sim I_1^2$, we have:
214:
215: \begin{equation}
216: I_{out}(t)=I_0(\ae_0-\ae_1 I_0)+I_1\exp(i\omega t) K(\omega)
217: \end{equation}
218: \begin{equation}
219: K(\omega)\equiv \ae_0-\ae_1 I_0\bigg(
220: 1+{1\over 1+i\omega\tau}
221: \bigg)=|K(\omega)| e^{ i\phi}
222: \end{equation}
223:
224: Here, as usual, the complex intensities $I$ and $I_{out}$ describe
225: the amplitude-phase relations between harmonically oscillating light
226: intensities at the entrance and at the exit of the absorber.
227:
228: Figure 3 shows characteristic frequency dependences of the
229: amplitude $K(\omega)$, phase $\phi(\omega)$, and time delay
230: $\delta(\omega) = \phi(\omega)/\omega$ of intensity oscillations of the
231: light beam
232: transmitted through a usual (bleachable) saturable absorber (a)
233: and through an inverse saturable absorber (b).
234:
235: As is seen from these curves, the greatest time delay is
236: experienced by low-frequency spectral components of the light
237: intensity ($\omega << \tau^{-1}$), for which, taking into
238: account the condition
239: $\ae_1 I_0<<\ae_0$
240: from Eq. (3), we have
241:
242: \begin{equation}
243: \Delta t_{\hbox{max}}=\Delta t(\omega\rightarrow 0)=-{\ae_1 I_0\over \ae_0}\tau
244: \end{equation}
245:
246: This formula yields the time delay experienced by a smooth
247: pulse with the width much larger than the relaxation tine
248: $\tau$. As expected, the time delay may constitute only a small
249: fraction of the pulse width.
250:
251: Figure 4 shows normalized light pulses at the entrance (solid
252: lines) and at the exit (dashed lines) of the usual (a) and inverse
253: (b) saturable absorber for several ratios of the pulse width
254: $\delta$ and relaxation time $\tau$. As seen from the figure,
255: the pulse propagating through a bleachable absorber generally
256: experiences positive time delay, whereas in an inverse absorber
257: the delay is negative. Note once again that all these results have been known
258: for several decades (see, e.g., paper \cite{18}, in which the shape of
259: the pulse transmitted through a saturable absorber is calculated).
260: Let us now turn back to the experiments on observation of "slow
261: light" under conditions of coherent population oscillations \cite{5,6}.
262:
263: \section{DISCUSSION}
264:
265: Comparison of the results presented above with the
266: experimental data of \cite{3,4,5,6} did not allow us to find any
267: noticeable distinctions between them. Specifically, in full
268: agreement with the model of saturable absorber, the amplitude
269: of the light intensity modulation at the exit of the absorber
270: shows a Lorentz-wise peak in the region of low frequencies
271: $\omega \lesssim 1/\tau$
272: (Fig. 3 from \cite{3}), and the phase shift of
273: the intensity modulation, at these frequencies, is positive or
274: negative depending on the type of the absorber (Fig. 3 from \cite{5}
275: and Fig. 2 from \cite{6}). A positive time delay is experienced also
276: by Gaussian pulse transmitted through a bleachable absorber
277: (cf. Fig. 4 from \cite{5} and Fig. 4 of the present paper).
278:
279: Thus, all the observations of \cite{3,4,5,6}, including "slow" and "fast"
280: light, can be interpreted in a fairly trivial way, in the framework
281: of the simplest model of saturable absorber. This model,
282: evidently, does not imply either burning of a narrow spectral
283: hole or any modification of the light group velocity in the
284: medium.
285:
286: The authors of \cite{3,4,5,6} interpret their experimental results in a
287: more complicated way by representing the harmonically
288: modulated light beam in the form of three spectral components,
289: with the central one playing the role of the pump and the
290: sidebands playing the role of the probe. Such an approach was
291: has been repeatedly used to describe interaction of a modulated
292: monochromatic light with a saturable absorber and is, in
293: principle, quite correct \cite{14,21}. First of all, however, the authors
294: of \cite{5} analyze the response of the absorber only to one of the two
295: side components and do
296: not take into account their inevitable influence upon each other
297: (see, e.g., \cite{21}). As follows from the model of the CPO effect,
298: when the probe beam contains two components symmetric with
299: respect to the pump frequency, each of them contributes to the
300: other one and, in turn, experiences the influence of the latter to
301: its own amplitude and phase. Without considering this fact, the
302: suggested theoretical model is, in our opinion, inadequate. It is
303: also incorrect to apply the standard notion of group velocity to an
304: absorbing medium with a strong spectral dependence of
305: absorption (under these conditions, the spectrum and shape of
306: the light pulse should exhibit strong changes, and the notion of
307: group velocity needs to be redefined).
308:
309: However, the main drawback of papers \cite{3,4,5,6} is that
310: the "dip" detected in the space of {\it difference} frequencies
311: (beat frequencies between "pump" and "probe") is
312: interpreted as a dip in the {\it optical} spectrum, which was not
313: really observed and, under conditions of the described
314: experiments, could not be observed. To detect the dip, spectral
315: width of the pump beam (as well as of the probe) should
316: be smaller than that of the dip. As applied to the case of
317: ruby crystal with a 37-Hz width of the "dip", the frequency of laser
318: light should be stable to within 14 (!) decimal digits (at least for
319: the times of the order of inverse width of the dip). In the
320: experiments \cite{3,4,5,6}, the widths of the laser sources evidently
321: exceeded the widths of the "burned" features by many orders of magnitude. Under
322: these conditions, position of the above "narrow dip", having no
323: preferences in a wide spectral range, becones physically
324: uncertain and loses any sense.
325:
326: All the aforesaid is also true for other studies in which the
327: retarded dynamics of a saturable absorber
328: was attributed to the hole-burning effect under conditions of CPO
329: \cite{7,8}.
330:
331: A different experimental situation was realized in \cite{22}, where as
332: the pump and the probe were used quasi-monochromatic beams of
333: independent laser sources, and the dip in the homogeneously
334: broadened spectrum of excitonic absorption of a semiconductor
335: structure with quantum wells was really detected in a classical
336: way as was implied in early theoretical papers on the CPO effect
337: (see, e.g., \cite{14}). Based on the results of the amplitude and phase
338: measurements, the authors concluded that the group velocity in
339: the system under study, due to a high steepness of the dispersion curve
340: in the region of the above dip, should be 9,600 m/s. This
341: conclusion is, in our opinion, not sufficiently well grounded. First of all, in
342: view of the already mentioned mutual influence of symmetric (with
343: respect to the pump frequency) spectral components of the light
344: pulse, the character of its interaction with the medium should
345: crucially depend on its phase structure. In particular, as has shown
346: our preliminary analysis, for a certain phase profile of the probe
347: pulse, it proves to be insensitive to the action of the pump
348: and incapable of detecting the dip. This fact indicates
349: again the nonlinear nature of the "hole" detected in the CPO effect and incorrectness of direct application of the Kramers-Kronig
350: relations to this spectral feature. And, lastly, as was
351: already mentioned, it is inconsistent to apply the notion of the
352: group velocity to the absorbing medium with a strong spectral
353: dependence of imaginary part of the refractive index.
354:
355: \section{CONCLUSION}
356:
357: The results of the above treatment allow us to conclude that the
358: experimental data of the papers \cite{3,4,5,6}, in which the "slow" or
359: "fast" light was allegedly observed under conditions of the
360: coherent population oscillations, can be easily interpreted in the
361: framework of the simplest model of saturable absorber. In our
362: opinion, the light pulse delay detected in these
363: experiments have nothing to do with the hole- (or anti-hole-) burning
364: in a homogeneously broadened absorption spectrum, as well as with
365: reduction of the group velocity of light in the
366: medium. These delays are
367: caused, in fact, by self-induced distortion of the pulse
368: propagating in the medium with intensity-dependent absorption. As for the
369: CPO-based hole-burning effect, it can be observed only under
370: conditions of quasi-monochromatic pump with its spectral width
371: smaller than the inverse population relaxation time. To observe
372: modification of the light group velocity in the medium under
373: condition of the coherent population oscillations, the probe
374: pulse spectrum should have approximately the same width. To
375: the present day, to the best of our knowledge, no experiment of this kind has been performed. One may assume
376: that, upon narrowing of the pump and probe pulse spectra, the
377: contribution to the pulse delay related to the refractive-index
378: dispersion feature
379: of the medium in the region of the pump frequency will become
380: noticeable on the background of the effects of retarded
381: absorption considered in this paper. However, taking into
382: account that this effect can be observed only in the
383: absorbing medium, and the time delay of the pulse may
384: constitute only a small fraction of its width \cite{9}, this "slow light"
385: will not differ from the "slow light" obtained using a saturable
386: absorber, and the prospects of its application for practical
387: purposes seem highly doubtful.
388:
389: The authors are grateful to E.B.Aleksandrov for useful
390: discussions.
391:
392: \begin{references}
393:
394: \bibitem{1} Casapi A., Jain M., Yin G.Y., and Harris S.E. //Phys. Rev. Lett. 1995. V. 74. P.
395: 2447.
396: \bibitem{2} Hau L.V., Harris S.E., Dutton Z., and Behroozi C.H. //Nature. 1999. V. 397.
397: P. 594.
398: \bibitem{3} Hillman L.W., Boyd R.V., Krasinski J., and Stroud C.R., Jr. //Opt. Commun. 1983. V .45.
399: P. 416.
400: \bibitem{4} Malcuit M., Boyd R.W., Hillman L.W., Krasinski J., and
401: Stroud C.R. Jr. //J.Opt.Soc.Am. B. 1984. V. 1. P. 73.
402: \bibitem{5} Bigelow M.S., Lepeshkin N.N., and Boyd R.W. //Phys.Rev.Lett. 2003. V. 90. P. 113903.
403: \bibitem{6} Bigelow M.S., Lepeshkin N.N., and Boyd R.W. //Science. 2003. V. 301. P. 200.
404:
405: \bibitem{7} Baldit E., Bencheikh K., Monnier P., Levenson A., and Rounget V. //LANL, arXiv: ccsd-00004377,
406: 2005.
407: \bibitem{8} Zhang Yun-Dong {\it et al.} //Chinese Phys. Lett. 2004. V. 21. P. 87.
408:
409: \bibitem{9} Boyd R.W., Gauthier D.J., Gaeta A.L., Willner A.E. //Phys. Rev. A.
410: 2005. V. 71. P. 023801.
411:
412: \bibitem{10} Matsko A.B., Strekalov D.V., and Maleki L. //Optics Express. 2005. V. 13. P. 2210.
413: \bibitem{11} Lamb W.E. Jr. //Phys.Rev. A. 1964. V. 134. P. 1429.
414: \bibitem{12} Schwarz S.E. and Tan T.Y. //Appl.Phys.Lett. 1967. V. 10. P. 4.
415: \bibitem{13} Baklanov E.V. and Chebotaev V.P. //Zh. Eksp. Teor. Fiz. 1971. V. 61. P. 922.
416: \bibitem{14} Sargent M. //Physics Reports. 1978. V. 43 P. 223.
417: \bibitem{15} Song J.J., Lee J.H., and Levenson M.D. //Phys.Rev. A. 1978. V. 17. P. 1439.
418: \bibitem{16} Lee J.H., Song J.J., Scarparo M.A.F., and Levenson M.D. //Opt.Lett. 1980. V. 5 P.
419: 196.
420: \bibitem{17} Gires F. and Combaud F. //J.Phys. 1965. V. 26. P. 325.
421: \bibitem{18} Selden A.C. //Brit.J.Appl.Phys. 1967. V. 18. P. 743.
422: \bibitem{19} Selden A.C. //J.Phys. D : Appl.Phys. 1970. V. 3 P. 1935.
423: \bibitem{20} Hercher M., Chu W., and Stockman D.L. //IEEE J. Quant. Electron. 1968. V. QE-4. P.
424: 954.
425: \bibitem{21} Kramer M.A., Boyd R.W., Hillman L.W., and Stroud C.R. Jr.
426: //J.Opt.Soc.Am. B. 1985. V. 2. P. 1444.
427: \bibitem{22} Ku P.-Ch., Sedgwick F., Chang-Hasnain C.J., Palinginis Ph., Li T., Wang H.,
428: Chang Shu-Wei, and Chang Sh-L. //Opt. Lett. 2004. V. 29. P. 2291.
429: \end{references}
430:
431:
432:
433: FIGURE CAPTIONS
434:
435: Fig. 1. Simplified energy-level diagrams of Cr$^{3+}$ ion in the
436: ruby (a) and alexandrite (b) crystals. Under conditions of
437: sufficiently strong optical pumping in the $|a\rangle$ -- $|b\rangle$ channel, a
438: fast relaxation of electronic excitation from state $|b\rangle$ to metastable state
439: $|c\rangle$ redistributes populations of the $|a\rangle$ and $|c\rangle$ levels (within the
440: time intervals of the order of $\tau$) and changes the absorption
441: of the crystal at the wavelength of excitation.
442:
443: Fig. 2. Time dependence of intensity of
444: the light beam transmitted through the usual (b) and inverse (c)
445: saturable absorber upon step-wise change of the incident light
446: intensity (a).
447:
448: Fig. 3. Typical frequency dependences of the amplitude
449: $K(\omega)$, phase $\phi(\omega)$, and time delay $\Delta t(\omega)=\phi(\omega)/\omega$ of
450: intensity oscillations of the light beam transmitted through the
451: usual (a) and inverse (b) saturable absorber.
452:
453: Fig. 4. Normalized light pulses at the entrance (solid lines) and
454: at the exit (dashed lines) of the usual (a) and inverse (b)
455: saturable absorber for several ratios of the pulse width $\delta$
456: and relaxation time $\tau$. Calculations were performed for
457: the pulse propagating in the presence of a background
458: illumination (under conditions close to these of \cite{5}).
459:
460: \begin{figure}
461: \epsfxsize=400pt
462: \epsffile{fig1.ps}
463: \caption{Fig. 1. Simplified energy-level diagrams of Cr$^{3+}$ ion in the
464: ruby (a) and alexandrite (b) crystals. Under conditions of
465: sufficiently strong optical pumping in the $|a\rangle$ -- $|b\rangle$ channel, a
466: fast relaxation of electronic excitation from state $|b\rangle$ to metastable state
467: $|c\rangle$ redistributes populations of the $|a\rangle$ and $|c\rangle$ levels (within the
468: time intervals of the order of $\tau$) and changes the absorption
469: of the crystal at the wavelength of excitation.}
470: \end{figure}
471:
472: \begin{figure}
473: \epsfxsize=400pt
474: \epsffile{fig2.ps}
475: \caption{Fig. 2. Time dependence of intensity of
476: the light beam transmitted through the usual (b) and inverse (c)
477: saturable absorber upon step-wise change of the incident light
478: intensity (a).}
479: \end{figure}
480:
481: \begin{figure}
482: \epsfxsize=400pt
483: \epsffile{fig3.ps}
484: \caption{Fig. 3. Typical frequency dependences of the amplitude
485: $K(\omega)$, phase $\phi(\omega)$, and time delay $\Delta t(\omega)=\phi(\omega)/\omega$ of
486: intensity oscillations of the light beam transmitted through the
487: usual (a) and inverse (b) saturable absorber.}
488: \end{figure}
489:
490: \begin{figure}
491: \epsfxsize=400pt
492: \epsffile{fig4.ps}
493: \caption{Fig. 4. Normalized light pulses at the entrance (solid lines) and
494: at the exit (dashed lines) of the usual (a) and inverse (b)
495: saturable absorber for several ratios of the pulse width $\delta$
496: and relaxation time $\tau$. Calculations were performed for
497: the pulse propagating in the presence of a background
498: illumination.}
499: \end{figure}
500:
501:
502: \end{document}
503: