Paper 3.12: N. Bloembergen, H. Lotem and R.T. Lynch, Jr., “Line shapes in coherent resonant raman scattering,” Indian Journal of Pure and Applied Physics 16, 151–158, 1978.
In the mid-seventies there was considerable confusion and controversy about the correct treatment of damping terms in CARS spectroscopy. I maintained that the general density matrix approach described in papers 1.3 and 1.4 should yield unambiguous results. Rob Lynch systematically kept track of all the terms in the calculation of χ (2) and χ (3) , and rearranged them in such a manner that terms arising only in the presence of damping are clearly separated. In the absence of damping, only resonances between the ground state and an excited level occur in the denominators. Damping introduces extra resonances between pairs of excited states. Their numerators vanish if the only damping is the natural width caused by spontaneous emission decay to the ground state. For collisional or other types of damping, their numerators are still small, but non-vanishing. I decided to publish the elaborate algebraic results as an appendix to a paper on line shapes of resonant raman scattering, which was our contribution to a special issue of an Indian journal commemorating the 50th anniversary, or golden jubilee, of the discovery by Raman of the effect named after him. Our results agreed, except for a few minor algebraic errors, with the independent calculations by J.P. Taran and M. Denariez at ONERA near Paris, the French institute for studies of the upper atmosphere. At scientific meetings, we had a difficult time convincing chemists who carried out the bulk of resonant raman experiments of the correctness of our approach. Taran summarized the situation at the end of one international conference to me in the following words, “Perhaps we won this battle, but we have lost the war. ” This reinforced my resolve to demonstrate the physical reality of the “extra” terms explicitly exhibited in this note.
In the mid-seventies there was considerable confusion and controversy about the correct treatment of damping terms in CARS spectroscopy. I maintained that the general density matrix approach described in papers 1.3 and 1.4 should yield unambiguous results. Rob Lynch systematically kept track of all the terms in the calculation of χ (2) and χ (3) , and rearranged them in such a manner that terms arising only in the presence of damping are clearly separated. In the absence of damping, only resonances between the ground state and an excited level occur in the denominators. Damping introduces extra resonances between pairs of excited states. Their numerators vanish if the only damping is the natural width caused by spontaneous emission decay to the ground state. For collisional or other types of damping, their numerators are still small, but non-vanishing. I decided to publish the elaborate algebraic results as an appendix to a paper on line shapes of resonant raman scattering, which was our contribution to a special issue of an Indian journal commemorating the 50th anniversary, or golden jubilee, of the discovery by Raman of the effect named after him. Our results agreed, except for a few minor algebraic errors, with the independent calculations by J.P. Taran and M. Denariez at ONERA near Paris, the French institute for studies of the upper atmosphere. At scientific meetings, we had a difficult time convincing chemists who carried out the bulk of resonant raman experiments of the correctness of our approach. Taran summarized the situation at the end of one international conference to me in the following words, “Perhaps we won this battle, but we have lost the war. ” This reinforced my resolve to demonstrate the physical reality of the “extra” terms explicitly exhibited in this note.
This note announces the first examples of dispersion in a multidimensional frequency space. Independently tunable dye lasers are used with two or three light beams incident on the sample. The intensity generated at the frequency ω 4 − ω 1 + ω 2 − ω 3 shows resonances when ω 1 − ω 3 and ω 2 − ω 3 are resonant with two different raman vibrations, or when ω 1 + ω 2 is near a two-photon transition and ω 1 − ω 3 near a vibrational resonance. The latter situation occurs in CuCl crystal, where two collective excitations are simultaneously enhanced, the Z 3 excitation at the sum frequency ω 1 + ω 2 in the near ultraviolet, and the resonance at w 1 − w 3 of the phonon–polariton in the infrared. Two full-length papers on this subject were published soon afterwards, “Third-order nonlinear optical spectroscopy in CuCl ,” S.D. Kramer and N. Bloembergen, Phys. Rev. B14 , 4654–4669, 1976, and “Interference between raman resonances in four-wave difference mixing,” Haim Lotem, R.T. Lynch, Jr. and N. Bloembergen, Phys. Rev. B14 , 1748–1754, 1976. Papers 3.9, 3.10, and 3.11 describe early examples of the general features of third-order nonlinear spectroscopy with three incident light beams with independent frequencies, directions of propagation and polarization vectors. At the same time that these developments in condensed matter spectroscopy took place, analogous methods evolved elsewhere in atomic spectroscopy. S.D. Kramer and R.T. Lynch, Jr. were both graduate students, while Haim Lotem was a postdoctoral research fellow from Israel. After obtaining his Ph.D. degree, Steve Kramer joined the Oak Ridge National Laboratory and Bob Lynch joined the IBM development laboratory near San Jose. Steve Kramer now works at the Institute for Defense Analyses near Washington, D.C. Haim Lotem returned to Israel, where he joined the Nuclear Research Facility near Beersheba. Soon after the Lotem family arrived in the United States, their son Zohar was born. Since sadly no one of their immediate family from Israel could be present, I was asked to serve as godparent at the circumcision ceremony in their Brooklyn apartment. I reminded Haim that I was not Jewish but he replied that it did not matter. Just before the ceremony, the rabbi, or mohel, had a talk with me about how I knew the Lotems, etc., and soon asked about the origin of my family name. When he found out the Protestant Christian background of my family, he disqualified me from entering the room where the circumcision would take place, much to the consternation of the anxious parents. A qualified replacement was immediately found in the person of my graduate student, Eli Yablonovitch. He had been born in Poland and raised in Canada where he obtained a Bachelor of Science degree at McGill University in Montreal. He was a confirmed bachelor at the time and he stated that the last circumcision ceremony he had attended was his own. He carried out this extracurricular duty with good humor. We have kept in close touch with the Lotem family. They returned to the United States several times on sabbatical leaves. Eli Yablonovitch and I attended a conference in Israel a few years later. On the El Al overnight flight to Israel, I was woken up by a group of men who were involved in an agitated discussion: they needed a tenth person to form a minion. I informed them that they had woken up a disqualified person and pointed out Eli Yablonovitch to them, who again performed the task without fail.
Time dependent description of resonance Raman scattering is performed for a general system, by a density matrix reformulation of damping theory originally used for resonance fluorescence of atom. Radiative damping in the intermediate states can be described in terms of "damping operator" in realistic situations. In large ergodic systems, the scattering reduces to luminescence if the relaxation time within the excited states is much shorter than the radiative lifetime. In a typical case, time resolved emission spectra consist of various stages (i) scattering complex where absorption and emission are inseparable, (ii) dynamical and then (iii) kinetic relaxations giving rise to the hot luminescence, which finally converges to (iv) the ordinary luminescence from relaxed excited states. Hot luminescence does not appear within a motionally narrowed line. The interpretation of spectra depends also on the method of observation.
Saturation effects in coherent anti-Stokes Raman scattering (CARS) spectroscopy are discussed. The discussion is limited to Raman-resonant CARS (${\ensuremath{\omega}}_{l}\ensuremath{-}{\ensuremath{\omega}}_{s}\ensuremath{\simeq}{\ensuremath{\omega}}_{21}$, where ${\ensuremath{\omega}}_{l,s}$ are the frequencies of the pump fields with powers Pl,s, and ${\ensuremath{\omega}}_{21}$ is the frequency of the Raman transition $|1〉\ensuremath{\rightarrow}|2〉$) with the possible addition of a one-photon resonance (${\ensuremath{\omega}}_{l}\ensuremath{\simeq}{\ensuremath{\omega}}_{31}$ is the frequency of the electronic transition $|1〉\ensuremath{\rightarrow}|3〉$). For these cases we show that the CARS polarization pCARS is proportional to the off-diagonal density-matrix element ${\ensuremath{\rho}}_{21}$. In order to determine ${\ensuremath{\rho}}_{21}$, we use Laplace transforms to solve the Bloch equations for the effective two-level system ∣1〉 and ∣2〉 when ${\ensuremath{\omega}}_{l}$ is far from resonance, or for the three-level system ∣1〉,∣2〉, and ∣3〉, when ${\ensuremath{\omega}}_{l}\ensuremath{\simeq}{\ensuremath{\omega}}_{31}$. The steady-state expression for pCARS in the former case gives ${\mathcal{p}}_{\mathrm{CARS}}\ensuremath{\propto}{P}_{l}{P}_{s}^{\frac{1}{2}}$ at low powers and ${\mathcal{p}}_{\mathrm{CARS}}\ensuremath{\propto}{P}_{l}^{0}{P}_{s}^{\frac{\ensuremath{-}1}{2}}$ at high powers. In the three-level system, we show that when the pressure is low and at least one field is weak, the slow time dependence of ${\ensuremath{\rho}}_{21}$ must be considered. When one field is strong and the other weak, the CARS spectrum is Stark split. When Pl is high, for example, ${\mathcal{p}}_{\mathrm{CARS}}\ensuremath{\propto}{P}_{l}^{0}{P}_{s}^{\frac{1}{2}}$ for ${\ensuremath{\omega}}_{s}\ensuremath{\simeq}{\ensuremath{\omega}}_{32}$ and ${\mathcal{p}}_{\mathrm{CARS}}\ensuremath{\propto}{({P}_{l}{P}_{s})}^{\frac{1}{2}}$ when ${\ensuremath{\omega}}_{s}\ensuremath{\simeq}{\ensuremath{\omega}}_{32}\ifmmode\pm\else\textpm\fi{}{V}_{13}$ where V13 is the one-photon Rabi frequency for the $|1〉\ensuremath{\rightarrow}|3〉$ transition. The Wilcox-Lamb approximation is used to reduce the three-level Bloch equations to rate equations containing one- and two-photon terms. When the fields are so weak that both one- and two-photon terms are small compared to the decay terms, the usual expression for pCARS is reproduced. If only the direct two-photon processes are important, the effective-two-system results are reproduced. When both fields are intense and nearly resonant, the steady state is rapidly achieved. The results for the case where one field is much stronger than the other are essentially the same as those for one strong and one weak field. When the fields are of comparable strength, ${\mathcal{p}}_{\mathrm{CARS}}\ensuremath{\propto}{P}_{l}^{\frac{1}{2}}{P}_{s}^{0}$ for ${\ensuremath{\omega}}_{l}\ensuremath{\simeq}{\ensuremath{\omega}}_{31}$ and ${\ensuremath{\omega}}_{s}\ensuremath{\simeq}{\ensuremath{\omega}}_{32}$, and the CARS spectrum is split into five components.
The quantum mechanical calculation of the nonlinear susceptibilities presented in Paper 1.1 applies only to purely parametric, non-dissipative processes. It yields real values for the nonlinear susceptibilities. I realized in the fall of 1962 that the general scheme for calculating complex susceptibilities in magnetic resonance, based on a density matrix description of the material system, would also be applicable in optics. These ideas are worked out explicitly for two- and three-level systems in this paper. It mentions the interference of one- and two-photon transitions between the same initial and final states This gives rise to an imaginary part of χ (2) , the lowest order nonlinear susceptibility. The imaginary χ (3) describes both raman processes and two-photon absorption processes. I was fortunate to keep my graduate student Ron Shen, who obtained his Ph.D. degree from Harvard University in 1962, for another year as a postdoctoral fellow. His approach to physics was very similar to that of Peter Pershan and myself. He combined experimental and theoretical skills in an efficient manner. We were able to connect the results of quantum mechanical analysis with the experimental results of raman lasers and parametric processes. This paper has not attracted much attention and is not often cited. It appears that too many details are hidden in the general formalism. There are a great number of papers published during the past three decades on two- and three-level systems subjected to electromagnetic fields at two or more frequencies. There are so many variants, depending on the detuning, the damping constants, and the amplitudes, or the “Rabi frequencies”, of the applied fields that different terms are selected from the general expression in each particular situation. If our paper accomplished little else, it certainly contributed to our own understanding of nonlinear phenomena. Circumstances conspired so that Ron Shen and I were able to collaborate again on some joint papers during the academic year 1964–1965. Ron had been appointed as an assistant professor at the University of California in Berkeley and I had independently made plans to spend a sabbatical leave there. I wanted to work with Charles Kittel on solid state problems in the physics department, and to give a course on nonlinear optics in the electrical engineering department which had a lot of activity in quantum electronics. My course was attended by two of Prof. John R. Whinnery's students, Eric Ippen and Robert V. Shank. These two later became well known for the generation of femtosecond pulses which opened up many new avenues of research. Ron Shen stayed in Berkeley as a professor of physics, and he continues to lead a very productive research group in nonlinear optics. He wrote a well-known textbook, Principles of Nonlinear Optics , published by Wiley in 1984. He and his family have also remained life-long friends of ours.
Lineshapes of three-level resonance Raman scattering and two-level resonance fluorescence, homogeneously broadened by collisions, in the presence of arbitrarily strong coherent radiation, are studied by means of a tetradic (density-matrix) scattering theory. In the case of medium coupling strength V L to the radiation field (where T 2 -1 <or=V L << tau c -1 , with T 2 and tau c being the coherence damping and collision times, respectively) the Mollow lineshape (1977) for resonance fluorescence is rederived, with proper T 2 and population cross-relaxation effects included. In particular it is shown that relaxation through intermediate non-resonant levels can considerably alter the lineshape in resonance fluorescence. In the case of strong coupling (V L >or approximately= tau c -1 ) the collision rates are modified by the applied field (optical collisions). It is shown that, whereas in self attenuation spectra optical collision rates can be obtained only from line intensity or line wing measurements, in resonance scattering these rates are directly obtained from linewidth measurements.
The theory of coherent anti-Stokes Raman scattering (CARS) is extended to include the effect of pump fluctuations. The intensities and spectra of lines in resonant CARS are calculated to all orders in fields assuming a phase-diffusion model for waves at the two pump frequencies. The bandwidth of the two lasers enters in a much more complicated way than following a simple scaling of T1 or T2. Various resonances in CARS spectra due to dynamic splitting of the energy levels are discussed for a range of detunings, field intensities, and bandwidths. In contrast to the usual spectra in strong fields, the Rabi sidebands appear as dispersion-shaped structures. The laser linewidth is shown to change dramatically the CARS line shape. The case of no saturation is also treated, thus allowing for the inclusion of more general line shapes and fluctuations of the pump waves and the nonlinear susceptibility tensor ${\ensuremath{\chi}}^{(3)}$. Gaussian statistics of the pump field are shown to lead to enhancement factors in CARS intensity, similar to those appearing in the context of multiphoton absorption processes.