Contrary to our expectations, in both investigated species a low-power laser like a Xe-lamp excitation produced IR-emission signal, which was high enough to allow the reliable registration by the FTIR technique. Figure 2 shows a typical IR-emission spectrum, which consists of two components. One of the components is the broad continuum emission with a blackbody-like spectrum that shows a sharp drop to the low-frequency region at 800 cm⁻¹. The several sharp characteristic peaks observed on the top of the broad curve are the features of the second spectral component. As in previous works, these peaks were correlated with bond vibrations as well as with appropriate infrared absorption and Raman spectrum (see below).

Table 1 contains spectral characteristics of different components in the emission spectra of CCl₄ and benzene. In CCl₄, emission signal I_T (integral (“total”) intensity of emission spectrum), I_B (emission intensity of (background) continuum emission component) and I_P (“pure” emission signal of discrete component) were linearly dependent on the excitation intensity and remained frequency independent. In С₆Н₆, emission signal exhibited a complicated property dependence on many factors. Particularly, signal I_T and I_B appeared to be frequency-de- pendent rather than power-dependent, whereas signal I_P displayed quadratic dependence on the excitation intensity and seemed to be frequency-independent. We noticed that in benzene signal intensity I_T and I_B could be decreased with the growth of exciting power instead of the increasing. For example, the irradiation of benzene at second harmonic of Nd:YAG laser (l₂) with a power of 4 mW had produced I_T or I_B signal practically of the same intensity as ones produced by a more enhanced exciting power of Xe-lamp beam. At the same time, the emission signal I_T or I_B observed under irradiation at main frequency (l₁ = 1.06 m) was decreased compared to ones for decreased intensity at l₂. It was particularly surprising because it is well-known that radiation at l₁

Notes: HWHM is the half-width of the spectral line at half-maximum; A.u. is an arbitrary units; I_T is the intergral (total) intensity of the IR emission spectrum; I_P is the “pure” intensity of the stimulated IR emission spectrum with the background washed off; I_B = I_T − I_P is the emission signal of continuum (background-like) component; G = (I_T − I_B)/(I_T − I_P) is the gain factor; The G values are the average of the five measurements.

absorbed by the overtones of CH stretching vibrations of benzene [22] and therefore, had to give rise for more intense infrared radiation. This is telling us, that the background emission signal is a complicated process and its origin is not entirely clear.

In Table 1 we compared signal intensities of the two spectral components in the infrared region between 1800 and 400 cm⁻¹, using coefficient G (G = I_T − I _B/I_T − I_P). In a certain sense, coefficient G is the gain coefficient of stimulated emission signal. As it can be seen, the relatively large signal gain G in CCl₄ and benzene was observed under Xe-lamp irradiation. In the case of laser irradiation, G coefficient in CCl₄ remained practically the same at the level of approximately 0, 8%, whereas G coefficient of benzene was varied from 0.4 to approximately 1.5% with changing of the illumination conditions. Closer examination of the power-dependence of both spectral components, I_B and I_P, with the use of cw-low-power sources revealed some interesting details in spectral behavior of CCl₄ and C₆H₆ species. Figure 3 shows plots of LIRE continuum and discrete signal powers of carbon tetrachloride and benzene vs. Xe-lamp excitation intensity. It should be noted, that the observed background signal I_B, common to both species, is somewhat larger in C₆H₆. This background is observed without adsorbed molecules. It was increasing almost linearly with the input power to a point where the signal started to decrease, as shown in Figure 3(a). The solid lines are Lorentzian fits to the point data. This feature reveals the saturation of the broadband pump by a Xe-lamp. At the same time, Ip signal intensity was progressively growing with the exciting intensity of Xe-lamp, as can be seen in Figure 3(b). The solid lines indicate quadratic fits to the point data, displaying a nonlinear power-dependence of signal intensity I_b in both organic liquids. These data clearly show that both signals, I_B and I_P are caused by a nonlinear process(es) and most likely these processes differ in origin. Strange as it may seem, there was no essential difference in the medium response between Xe-lamp and Nd:YAG laser excitation under given experimental conditions. The power-dependence of LIRES signal vs. laser excitation intensity at l₂ nm is presented at Figure 4. Given observations imply that the LIRES signal strongly depends on the intrinsic material properties and source’s parameters. Figure 4(a) shows plots of signal power I_B of liquid species as a function of input laser power. In CCl₄, the emission signal I_B increased without threshold and this is followed by a sharp decrease. In C₆H₆, a small plateau in signal intensity followed by a sharp linear increase and the onset of the nonlinear response. In both cases, the solid lines are sigmoid fits to the data points. As in the first case, these features reveal the power-limiting capacity and the square-low-non- linearity of the material response as well. Figure 4(b) shows plots of signal power of I_P discrete component for carbon tetrachloride and benzene vs. excitation laser intensity. In CCl₄, as in benzene, exponential growth in I_P signal intensity in the emission spectrum was observed. The solid lines are quadratic fits to the experimental data. The obtained results indicate that the material response produced by different low-power cw-sources is similar in appearance. It means that in both cases underlying nonlinear mechanism(s) could be similar and I_P signal may originate from pumped vibrational sublevels of electronic ground state. If our previous assumption concerning the nature of a discrete component in light-induced infrared [17] [19] [20] emission (LIRE) spectrum is correct, the two-wave mixing mechanism had to work under Xe-lamp and as well laser excitation. It is significant that liquid benzene and CCl₄ possess a key feature, requisite for this process, which is the third-order optical susceptibility. Indeed, cubic nonlinear properties of liquid benzene and CCl₄ substances are well-known. However, the heterogeneity of an exciting field, which is also necessary for two-wave mixing process, is not shared by laser for at least one-color beam excitation. In this case, the heterogeneous field may only arise from

the field scattered by the molecules and/or reflected from the surface. Due to the particular low-intensity laser beam ordinary Raman scattering obviously could not play a marked role in production of IR-emission signal. Therefore, other mechanisms have been proposed that may influence the spectral properties of the investigated species.

The liquid CCl₄ and benzene, as the transparent organic species, possess the Kerr-nonlinearity [23] , which is well-known to be dominant nonresonant nonlinearity. The Kerr-nonlinearity, as a rule, is associated with the self-focusing effect [24] -[27] . The latter occurs when light modifies a medium in such a way, that it creates intensity-dependent focusing lens. It increases the beam intensity enormously and enhances all nonlinear optical effects.

Self-focusing has been widely studied in many different media [28] -[32] . Such studies were usually performed using high-power laser pulses. Meanwhile, continuous-wave self-focusing at low-power level was seen in gas [33] and nematics, materials with a large cubic nonlinearity [34] -[36] . As the self-focusing properties of benzene and carbon tetrachloride are well-known, it is reasonable to assume, that this effect may also took place under our experimental condition. This is not incredible as it may be seen at first sight. Indeed, in early works it was shown that a low-power laser excitation of transparent organic substances, including CCl₄, changed refractive index across the beam width [37] -[39] . A variation of the refractive index may be as small as 10⁻⁸ [40] . Despite of very small value, such changes may produce a measurable lens effect. The lens may be created either through temperature dependence of the refraction index or through numerous other nonlinear effects including self focusing effect. The later could lead to remarkable lowering of the thresholds for other nonlinear process such as self-phase modulation or the stimulated Raman scattering [41] [42] . It should be noted, many experimental evidences exist that stimulated Raman scattering is initiated by self-focusing process [24] -[27] [36] [41] - [44] . As such, stimulated Raman scattering may take part in selectively populating of vibrational levels, consequently leading to the generation of mid-infrared emission. Indeed, the theoretical possibility to observe the process of generation of infrared radiation concurrent with stimulated Raman generation has earlier been considered by a number of authors [45] -[47] . According to this consideration, large-amplitude coherent vibrations will be produced under stimulated Raman scattering, and if these molecules have an electric dipole moment, infrared radiation will be emitted at the vibrational frequencies. We believe, despite the particular low-intensity of laser beam used in our experimental conditions, high Kerr nonlinearity of investigated organic species (liquid CCl₄ and C₆H₆) may lead to self-focusing process, whose role in production of mid-infrared emission may be not so negligible as one could imagined. Further work is being undertaken to understand this phenomena.

Of the great interest are the spectral features of discrete component in the light-induced infrared emission (LIRE) spectra which will be analyzed in detail. It is useful to begin with a consideration of benzene molecule, as the more complex specimen. This specimen is known to have D_6h point group symmetry and 30 normal modes with 20 basic frequency (10 modes are doubly degenerated). The D_6h has the centre of symmetry, so an exclusion rule exists between modes which are only Raman or IR-active. From 20 normal vibrations seven fundamentals are Raman-active and only four ones are active in infrared: three of them are in-plane vibrations (n₁₈, n₁₉ and n₂₀), the fourth is the out-plane fundamental n₁₁. However this molecule has a large number of infrared active combination, difference, and hot bands, throughout the mid-infrared region. Our observations shown, that all LIRE- active vibrations are localized in 2500 - 500 cm⁻¹ spectral region. No signal was observed higher than 2800 cm⁻¹. Similar observations were found for carbon tetrachloride, in which no bands were observed already above 1600 cm⁻¹. (Data are not present). Figure 5 and Figure 6 show the infrared emission spectra of benzene in the 2100 - 500 cm⁻¹ spectral region, where the strongest peaks are localized. The assignments to the peaks shown in the spectra of Figure 5 and Figure 6 including possible combination and overtone mode are listed in Table 2. This table presents the frequency values of the different bands observed during the illumination of benzene with use of both cw-lower-power sources, Xenon lamp and Nd:YAG laser. Table 2 does not include the smaller intensity peaks that occur between 2500 and 400 cm⁻¹. Peak assignments were based upon frequencies correlation, symmetry arguments and assignments from Herzberg’s [48] . Mode labeling follows Herzberg’s notation. As it follows from the experimental data (see Table 2), the bands with symmetry class a_2u, e_2u, e_1g, e_2g and randomly b_1u and b_2u are enhanced in the LIRE spectra of benzene. As it is well-known, bands with a_2u and e_1u symmetry types are IR-active, while the bands with e_1g and e_2g symmetry are inactive in IR, being Raman-active, or/and

Notes: v is variable intensity: vs and vw is very strong or very weak intensity; s, m, and w are strong, medium, and week intensity; sh means a shoulder; b means broad band; h means “hot” band; IR means infrared-active; R-Raman-active; TP-two-photon active; HR^1a-hyper-Raman active, Tw (this work)―LIRE-active; ^{*, **}strongly overlapping bands; ^aRef. [48] ; ^bRef. [52] ; ^cRefs. [49] [50] [55] ; ^dRef. [53] ; ^fRef. [54] .

LIRE-active at the same time.

Spectrum shown in Figure 5 (upper curve) was recorded under Xe-lamp excitation. Infrared absorption (the bottom curve) is presented for comparison. As could be seen, the major part of emission peaks correlated with bond vibrations and the appropriate infrared absorption spectrum. It was therefore assumed that emission peak could be assigned to the same molecular vibration transition as those associated with the absorption band. Indeed, three basic IR-active bands (1477 (n₁₃), 1038 (n₁₈), and 681 (n₄) cm⁻¹) seen in Figure 5, are observed at practically the same positions, however, they have different relative intensity as compared to absorption ones.

The strongest bands are measured near 898, 796 and 725 cm⁻¹ while the bands of a moderate intensity occur near 1521, 1102, and 1038 cm⁻¹, as it can be seen in Figure 5. It should be noted, that all strong emission bands are not present in absorption spectrum, where only one weak band at 771 cm⁻¹ is observed. The origin of the largest intensity peak at 898 cm⁻¹ is unknown. It is not also observed in ordinary Raman scattering. However, it may be revealed in IR-absorption spectra of some benzene derivatives, belonging to the D_3h symmetry group [49] -[51] . We found that, the frequency, as well as relative intensity, of this band strongly varied with changing the illumination conditions (see below). Two other intense peaks at 796 and 725 cm⁻¹ may be derived from out-plane H bending motion [52] .

In spectral 1650 - 1400 cm⁻¹ region, beside the fundamental n₁₃, three other bands are revealed: of a moderate intensity 1521 cm⁻¹ band and two closely spaced bands at 1438 and 1402 cm⁻¹_. The first one was identified as the IR-active n₄ + n₁₁ (e_1u) mode, which is also seen in IR-absorption spectrum (see Figure 5). The latter corresponds to the Raman-active combination mode n₆ + n₂₀ (e_1g), which has been observed with Raman scattering

and is both Raman and LIRE-active. The origin of the peak at 1438 cm⁻¹ is unknown. Besides the most enhanced bands there are many other additional features in this region, such as the small duplet peaks at 1637 and 1613 cm⁻¹, probably corresponding to the frequencies of IR-active combination mode n₁₄ + n₁₈ (e_1u) and Raman-active combination mode n₂ + n₁₈ (e_2g), respectively.

In 1970 - 1750 cm⁻¹ region, instead of two sharp features at 1958 and 1813 cm⁻¹ in IR-absorption, only very weak distinct bands at 1970 and 1946 cm⁻¹ and strongly overlapping bands at 1811, 1786, and 1758 cm⁻¹ are observed in emission spectrum. The weak doublet around 1970, 1946 cm⁻¹ was assigned to the splitting of 1964 cm⁻¹ band, which corresponds to IR-active n₁₁ + n₁₉ (a_2u) mode. The bands at 1811 cm⁻¹ precisely match the IR-active bands at 1813 cm⁻¹, which corresponds to n₁₈ + n₁₉ (a_2u) mode. Possible assignments for other bands are listed in Table 2.

It should be noted that LIRE spectrum (Figure 5) recorded under Xe-lamp excitation exhibits features that are not present in the spectra recorded with a laser source (Figure 6). For example, in the case of Xe-lamp, there are no emission bands in spectrum, that are higher than 1980 cm⁻¹, whereas Nd:YAG laser may produce several relatively intensive emission bands in this spectral region. For laser, frequencies of several bands may be shifted in the emission spectrum and relative intensity of the bands may be strongly changed as compared to Xe-lamp excitation. The linewidth in emission spectrum recorded with Xe-lamp is about 18 - 22 cm⁻¹ FWHM (Full width of half maximum), whereas a linewidth for laser excitation is slightly wider ~22 - 28 cm⁻¹ (curve a and c) and 26 - 30 cm⁻¹ FWHM (curve b). The observed distinctions between Xe-lamp and laser excitation may arise from a difference in a wave-driven excitation mechanism.

Although spectra recorded at the different excited laser frequencies are very similar, they are not the same (see Figure 6). Therefore, it is interesting to compare the spectra recorded with different input frequencies. Figure 6(a), Figure 6(b), and Figure 6(c) show emission spectra for C₆H₆ measured at three different input frequencies, where the large-intensity peaks are displayed in the low-frequency 920 - 600 cm⁻¹ spectral region. All these spectra are characterized by the presence of the single enhanced feature near 603 - 606 cm⁻¹, which occurs as a symmetric single peak at 603 cm⁻¹ or peak at 606 cm⁻¹ with the slight asymmetric distortion toward higher frequencies or the doublet peaks at 606 and 570 cm⁻¹ on the curves a, b and c, respectively. The 606 cm⁻¹ band precisely matches the frequency of n₁₈ (e_2g) mode (intramolecular C-C-in-plane bending motion), which is Raman-active and is not observed in IR. Also, spectra showed in Figure 6 exhibit the common features around 900, 850 and 760 cm⁻¹, with peak positions as well relative intensities being strongly varied with changing the exciting frequency. For example, the strongest enhancement of the band at 915 - 912 cm⁻¹ is observed in the emission spectra measured at exciting l₁ nm and l₁ plus l₂ nm laser beams, whereas, upon excitation with laser l₂ nm beam, it occurs as a small peak with frequency red-shifted as compared to considered above, but it precisely matches the corresponding frequency in the spectrum measured upon Xe-lamp excitation. This feature has no assignments, as already mentioned above. It may correspond to the calculated “hot” band of 1038 cm⁻¹ (a_2u) [53] mode either deformation vibration of out-of-plane of C-H mode appearing in the distorted molecules of benzene derivatives [49] [51] . Thus in benzene molecule, the LIRE spectra reveal the only LIRE-active bands. The frequencies of the bands around 848 and 760 cm⁻¹ are closely matched to corresponding frequencies of R-allowed n₁₁ (e_1g) and IR-active 771 cm⁻¹ (a_2u) modes, respectively. The relative intensity of these bands was strongly varied upon changing the input laser frequency, as can be seen in Figure 4.

In “fingerprint region” , we point out one more important common feature around 1480 cm⁻¹, which is presented as the peak of high or moderate intensity at 1485 cm^-1 respectively ( see curves a and c) and of very weak intensity at 1474 cm⁻¹ (curve b). This peak precisely matched with the fundamental frequency IR-active n₁₃ (e_1u) mode. Meantime the emission spectrum (curve b) measured at one-color l₂ nm laser beam demonstrates the strong intensity band at 1416 cm⁻¹ which is not observed in spectra measured at l₁ nm or at two-color l₁ + l₂ nm beams. This distinct band could be identified as the combination of n₆ + n₂₀ mode (e_1g class symmetry), which is Raman-active and inactive in IR.

In spectral region 1780 - 1600 cm⁻¹, there are no bands in the spectrum measured at l₂ nm (Figure 6(b)), whereas for the l₁ and l₁ + l₂ nm laser excitation (Figure 6(a) and Figure 6(c)), there are three additional distinct features with a different relative intensity observed near 1758, 1666 - 1664 and 1601 - 1598 cm⁻¹. In Figure 6(a), two closely spaced peaks at 1664 and 1598 cm⁻¹ are of the same intensity, whereas in the spectrum shown in Figure 6(c) the peak at 1664 cm⁻¹ displays the relatively large enhancement as compared to one at 1601 cm⁻¹. The 1664 cm⁻¹ peak closely matches the combination frequency of n₂ + n₄ (a_2u). The peak around 1600 cm⁻¹ corresponds to either the combination frequency of n₂+ n₁₈ (e_2g) (depolarized Raman-active vibration) or to the “quadrant” stretch vibrations of carbon ring (CC symmetric “quadrant stretching 1596 cm⁻¹ is forbidden in IR, but allowed in Raman [51] ). At the same time, laser excitation at l₂ nm produces the relatively strong single band around 1579 cm⁻¹ with a small shoulder in the side of high frequencies (see Figure 6(b)), which however doesn’t appear in spectra shown in Figure 6(a) and Figure 6(c). This complex band is precisely matched with the calculated frequency of the other “quadrant” stretch vibrations of carbon ring, which may be revealed by IR-absorption spectroscopy in distorted molecule of substituted benzene, but could not in the spectrum of unsubstituted benzene [51] . Thus, the LIRE spectra shown in Figure 6 demonstrate the presence of a number of forbidden and/or Raman-active lines, which are inactive in IR-absorption, but are Raman and/or LIRE-active. These observations reflect the structure alterations, which may be due to one or more of the following effects:

1) Removal of D_6h symmetry of free benzene molecules in the liquid state.

Because of the intermolecular forces, the structure may suffer distortion resulting thereby in lowering of their D_6h symmetry and hence in changes vibrational modes. However, although some mode is not IR-active in free benzene, it becomes active in the benzene molecules in the liquid state.

2) Cluster formation.

Formation of dimers or higher-order molecules in illuminated region will lead to the increased number of bands in the infrared emission spectra. In this case the mutual exclusion rules for the single molecules will be altered, and some modes forbidden in IR for the single molecule, can become IR-active for complex.

3) Self-focusing effect.

Due to self-focusing process, benzene molecules trapped in narrow “channel” inside illuminated region (data not shown) that may produce changes in IR spectra similar to those observed on compression or during liquid to solid phase transitions.

Finally, in the high frequency region (higher 1800 cm⁻¹), we point out the shift of out-of-plane bands as compared to IR-absorption. Thus, while absorption bands appear at 1811 (n₁₈ + n₁₉ (a_2u)), 1906 (n₁₁ + n₁₄ (a_2u)) and 1958 (n₁₁ + n₁₉ (e_1u)) cm⁻¹ for liquid benzene [48] , in the case of one-color excitation at l₁ nm and l₂ nm they are observed at 1825 and 1958 cm⁻¹ or at 1803, 1899, and 1946 cm⁻¹, respectively. In the case of two-color (l₁ and l₂) laser excitation, these bands are observed at 1811, 1888, and 1946 cm⁻¹. Also, we point out that the relative intensity of these bands depended considerably on the input frequencies. It is obvious that the differences in the spectra presented in Figure 5 are qualitative rather than quantitative and originated predominantly from the interaction of the laser beam with the sample rather than thermal excitation. The most important question concerning the origin of the observed spectral changes still remains open. Further spectroscopic experiments are certainly required to clarify the mechanism of underlying phenomena.

From our experimental data it is clear, that IR emission spectra contain important information concerning the internal properties of investigated species as well as light-matter interaction.