Comment on ‘Evidence of slow-light effects from rotary drag of structured beams’

The paper Wisniewski-Barker E et al (2013 New J. Phys. 15 083020) is intended to distinguish experimentally between two mechanisms of pulse delay in ruby and to provide evidence in favor of the slow-light model. The proposed test is based on the idea of monitoring time delay of a ‘dark pulse’ or ‘intensity null’, rather than that of some Gaussian-like pulse. We show that, because of certain experimental inconsistencies, the results of the measurements do not allow one to prefer one of the models and, thus, are interpreted inadequately. In this comment, we propose and realize a simple modification of the experiment Wisniewski-Barker E et al (2013 New J. Phys. 15 083020), which allows us to unambiguously resolve this dilemma. We show that the effect of pulse delay in ruby is perfectly described by the simple model of pulse reshaping and does not require invoking the coherent population oscillation-based slow-light effects.


Introduction
The recently published paper [1] is intended to provide evidence of slow-light effects in ruby crystal. The laser pulse delay in ruby, according to [1], may result not only from trivial reshaping of the pulse in a saturable absorber, but also from steep dispersion of the phase index in the vicinity of the narrow spectral dip produced in the absorption spectrum of the crystal through the effect of coherent population oscillation (CPO). To distinguish between these two mechanisms, the authors proposed to study behavior of the 'intensity null' in a strong laser pulse, which should essentially differ for the two models of pulse delay.
We completely agree with the main idea of the paper that, in the case of reshaping, the transmitted light intensity should vanish at zero intensity of the incident light. We also agree that, ideally, the way proposed by the authors to check it (with a dip in the pulse center) may work. But what is crucially important for such an experiment is that the light intensity in the tested point should really be zero (to within the experimental accuracy), because, otherwise, the residual incident light will probe the varying transmission of the medium, and the experiment will lose its original sense. In addition, if, as the authors assert, 'the two mechanisms are practically indistinguishable when the input is a single pulse', it is difficult to believe that they will become distinguishable when a single pulse is replaced by two successive pulses. It is clear (and can be easily confirmed by appropriate calculations [2]) that the dip of a two-humped pulse with no real 'intensity null' will be inevitably delayed in a bleachable absorber without any slow-light effects. Thus, as seen from the presented photometric curves, the experimental observations of [1] qualitatively correlate with the model of reshaping, and the delay of the pulse dip cannot serve as an 'evidence of slow-light effects'.
It should be also noted that the employed visual representation of the results additionally disorient the reader, because what we see is the drag of bright regions, while what we want to observe is the drag of a thin dark line, which is, evidently, not the same.
So, we have to admit that the main goal of the experiment has not been achieved because of inappropriate realization of a sound idea. In this paper, we present, as we believe, a possible simple implementation of such an experiment.
Perhaps the authors are right that it is not easy to create a dip with the 'intensity null' in the temporal profile of a laser pulse, but it is much more important that this dip is not needed to realize the authors' idea. The 'intensity null' in the incident pulse is actually needed to check whether the transmitted light intensity at this moment is zero or not, rather than to measure the dip shift. Figure 1 shows a fragment of figure 1 from [1] illustrating transformation of the pulse with 'intensity null' expected for the models of optical bleaching (a) and group velocity reduction (b). The two pictures are indeed essentially different. Note, however, that the second (left) half of the pulse after 'intensity null' does not play any role in this experiment because it cannot affect the measured value of light intensity at the point of interest A (just for causality reasons). Therefore, this part of the pulse may be readily removed together with all the problems of creating 'intensity null' in temporal domain. As a result, we come to the experiment with a single pulse presented by right sides of the pictures (highlighted in figure 1). Our task is to check whether the output pulse intensity I A at point A (where the input intensity is blocked) vanishes or not. Below we describe such an experiment and its results.

Experimental
Schematic of the experimental setup is shown in figure 2. The laser beam with the wavelength λ = 532 nm was preliminarily focused and modulated by a chopper in its waist (f ≈ 30 Hz).  Figure 1. Distortion of the incident pulse passing through a bleachable absorber (according to [1]) for the models of pulse reshaping (a) and group velocity reduction (b). Figure (a), unlike appropriate picture in [1], is drawn neglecting possible contribution of fluorescence, which can be easily removed experimentally. The two models can be easily distinguished by measuring output pulse intensity at point A (I A ), which should be essentially nonzero in the case (b). digital phosphor oscilloscope Tektronix DPO5104. Filter F was placed in front of the photodiode to eliminate fluorescence of the crystal. Figure 3 shows results of the measurements. As seen from figure 3(a), distortion of the pulse with sharp edges perfectly corresponds to its expected behavior in a bleachable absorber [2]: The leading edge shows a step-wise increase with subsequent region of exponential bleaching. At the same time, the trailing edge remains undistorted, with no intensity signal after the incident light is blocked (after the 'intensity null'). Inset in figure 3(a) illustrates with higher accuracy that no trace of the self-pumped CPO-based slow light is observed.
Additional measurements with longer edges of the pulse (just to decrease the bandwidth of the light pulse, which, according to the CPO-based slow-light model, may be of importance) also did not show any deviation from the simplest model of saturable absorber ( figure 3(b)), and intensities of the input and output light came to zero (within our experimental accuracy) simultaneously (inset in figure 3(b)).

Conclusions
In summary, we have shown that consistent experimental realization of the idea proposed in [1] confirms unambiguously the simplest model of pulse delay in a saturable absorber (known and accepted since the 1960s [3]) and leaves no room for the alternative (CPO-based) model. It should be emphasized that this simple model [2,3] has never had any difficulties in describing dynamics of pulse propagation in a saturable absorber and did not need invoking any additional, more sophisticated, models. At the same time, the CPO-based model of self-pumped slow light [4] has never been substantiated theoretically and was repeatedly refuted in many earlier publications [2,[5][6][7][8][9]. We believe the authors of [1] will now agree that there are no physical grounds to connect pulse delay in ruby (as a particular case of bleachable absorber) with the CPO-based slow light.