1. Introduction

Diode-pumped alkali lasers (DPAL) have been extensively subjected to research and development for the past two decades due to their potential to compete with conventional high power laser systems or even replace them by employing alkali vapor [1,2], which is also beneficial to various applications in the fields of quantum physics, precision measurement, and nonlinear optics [3–6]. Moreover, the four-wave mixing (FWM) in alkali vapor involves a nonlinear optical process where in interactions between two or three wavelengths produce new wavelengths, while the pulse compression and quantum memory storage are enhanced by recording the amplitude and phase of coherent optical signals [7]. The FWM process can also be used to study the interference effect and the Autler-Townes splitting in alkali vapor through the coherent blue light [8].

Furthermore, FWM is an efficient and robust method for both frequency up-conversion and mid/far-infrared (mid/far-IR) generation in atomic media [9]. In a previous study conducted by Akulshin et al., they have utilized extended cavity diode laser (ECDL) sources of continuous wave (CW) coherent light operating at 780 and 776 nm to resonantly excite rubidium (Rb) atoms in the 5D_5/2 level (i.e., 5²S_1/2→5²P_3/2→5²D_5/2). Experimental results showed that directional infrared emissions at 5.23 µm, and 1.37 µm were generated in Rb vapors; these infrared emissions were originated in the population-inverted (between 5²D_5/2 and state 6²P_3/2) medium. Moreover, a collimated blue light at 420 nm (6²P_3/2→5²S_1/2) was obtained and a new pathway (5²P_3/2→5²D_5/2→6²P_3/2→6²S_1/2→5²P_3/2) for parametric FWM was established [10]. Meanwhile, Kitano and Maeda have observed the emissions associated with 420 nm yoked superfluorescence (YSF) emitted from Rb vapor after exciting Rb atoms from 5S to 5D states using an ultrashort pulse laser, and the infrared radiation at 5.23 µm happened [11]. Subsequently, they have investigated the distortion mechanism and transverse properties of the emitted YSF from measured wavefronts [12]. Brekke and Herman used a 778 nm CW laser (2hν) to generate a coherent light at 420 nm by the parametric FWM process in Rb vapor [13], the frequency of the 420 nm light was observed to shift as the excitation laser was tuned, and the tuning range is over 1 GHz. Moreover, the population inversion between 5²D_5/2 and 6²P_3/2 states was investigated by Akulshin et al. using the mid-IR emission at 5.23 µm produced from the amplified spontaneous emission (ASE) process and the isotropic blue fluorescence at 420 nm [14]. Another study reported that the parametric FWM is responsible for generating a spatially coherent blue light in alkali vapors, which is a process that involves population inversion and ASE. The relationship between the FWM and the ASE was evaluated using the directional mid-IR radiation in the forward direction (5.23µm) [15]. Furthermore, several existing studies have reported that the FWM process similarly occurred in cesium (Cs) vapor [16]. Sulman et al. have investigated the generated pulsed blue beam (7P→6S) and infrared beam through experimental results after exciting Cs atoms to 7D or 9S states using two-photon absorption [17]. Furthermore, Zhang et al. demonstrated the interaction between the seeded and self-seeded nondegenerate FWM in Cs vapor that was conducted in a diamond-type system [18]. It is of great importance to enhance the output power and efficiency and to narrow the linewidth of the blue light generated by FWM in an Rb vapor cell using a ring cavity, which was first demonstrated by Offer et al. using CW pump lasers [19] and developed by Brekke and Potier [20]. Except the collimated blue light generated by FWM, the infrared emissions were seen in alkali vapor, which can be obtained by the stimulated electronic Raman scattering (SERS) or stimulated hyper-Raman scattering (SHRS) processes in potassium [21,22] (pulsed excitation) and sodium [23] (CW excitation) vapors etc., both SERS and SHRS belong to the stimulated Raman scattering (SRS) effect; these results have been verified experimentally.

From those literatures mentioned above, mainly the lower electronically excited states of alkali are involved, and emissions generated are in the wavelength of mid-infrared and visible (or ultraviolet, UV). If the alkali atom is excited to high-lying Rydberg states, far-infrared radiation could be realized. These works inspire us that alkali vapor can even generate Terahertz (THz) emission between Rydberg states. The expansion of FWM was demonstrated in the THz field that utilized two CW laser beams at 780 and 515 nm to pump Rb atoms to the 10²D_5/2 state; then UV light at 311 nm (11²P_3/2→5²S_1/2) and THz radiation at 3.28 THz (10²D_5/2→11²P_3/2) were realized [24], subsequently, the energy of THz radiation was measured [25]. Further, THz technology exhibits various applications, such as imaging, communication, sensing, and biomedicine [26–29]. At present, The THz source based on electronic technique has the advantage of high power output, but only works in the frequency lower than 1 THz, so the THz source with high frequency (>1THz) mainly utilizes optical technique. The alkali vapor THz source has many advantages, such as simple structure, adjustable frequency, narrow linewidth, high gain coefficient, the THz radiation from the transition between Rydberg states in all the normal alkali elements (Li, Na, K, Rb, Cs) distributes in the range of 0.1 to 10 THz, just as shown in Table 1. Because of the high gain coefficient from the alkali atom, the THz small signal will be amplified when its frequency falls in the Raman gain region. Until now, the study of the THz character in Cs vapor is sparse, so it is necessary to investigate the mechanism of THz generation in Cs vapor.

Table 1. Alkali vapor THz frequency

View Table

The THz emission is involved in the FWM process, the collimated visible or UV light from FWM can reveal the character of the THz emission. In our previous study, we have generated a collimated visible light using single-visible-wavelength laser-induced FWM processes in Cs vapor, and the internal quantum efficiencies of the THz emissions were estimated [30]. For this study, we have experimentally investigated the FWM process in Cs vapor that was pumped by two-color lasers, and the intensity of the collimated UV light under different wavelengths of pump lasers was obtained. Furthermore, we have studied the SERS and SHRS effects and evaluated the impact of THz Stokes emission and phase matching on the UV FWM process. THz frequency could be adjusted by tuning the wavelengths of the pump lasers, and 1.0 THz pulse energy was estimated.

2. Experimental setup

The experimental scheme is illustrated in Fig. 1(a). A Nd:YAG pumped OPO laser (“Laser Vision”, ∼852 nm, name as P1) and a Nd:YAG pumped dye laser (Contimuum ND6000, ∼550 nm, name as P2) are used in this experiment, repetition rates of both lasers are 10 Hz. The pulse energy of P1 is 0.84 mJ, and the linewidth is 0.22 cm⁻¹ (FWHM, full width at half maximum); P2 has a pulse energy of 1.21 mJ and a linewidth of 0.07 cm⁻¹ (narrowband). These two beams were collinearly aligned through a dichroic mirror and synchronously arrive in the Cs vapor cell. The Cs vapor cell was manufactured using quartz glass with a length of 75 mm and a diameter of 25 mm. It was ensured that there is plenty of Cs metal in the cell. The operation temperature of the cell was controlled at 200 °C with precision of ± 1 °C in an oven with a temperature controller. The collimated UV light was transmitted together with the pump lasers after the FWM processes, while an optical filter eliminated P1 and P2, and only the collimated UV light was coupled into the spectrometer by a fiber and recorded on an intensified charge couple device (ICCD).

 

Fig. 1. Scheme of the experimental setup and schematic diagram of FWM processes in this work. (a) The experiment scheme of UV FWM. (b) UV FWM process with THz generated by ASE mechanism. (c) UV FWM process with THz generated by SERS mechanism. (d) UV FWM process with THz generated by SHRS mechanism.

Download Full Size | PDF

The schematic diagram of UV FWM processes in Cs vapor was presented in Fig. 1(b) to (d). In Fig. 1(b), P1 resonantly excites the Cs atom from the 6²S_1/2 state to the 6²P_3/2 state; subsequently, the wavelength of P2 will be subjected to fine adjusting until it resonates with Cs in the 6²P_3/2→11²D_5/2 transition via the ASE process. Then, the Cs atom on the 11²D_5/2 state will transit to the 12²P_3/2 state under an approximate radiation of 1.0 THz. After undergoing frequency mixing among P1, P2, and THz ASE, a UV light with a wavelength of approximately 335 nm is obtained. If the wavelength of P1 that resonates with Cs 6²S_1/2→6²P_3/2 transition is kept constant and the Cs atom in the 6²P_3/2 state is non-resonantly excited to the virtual level (vl) by P2 (Fig. 1(c)), a THz radiation of approximately 1.0 THz (vl→12²P_3/2) is observed during the SERS process. Subsequently, the collimated UV light is observed after the FWM process. When a Cs atom is excited from the ground state to the vl via non-resonant two-photon excitation, a radiation of around 1.0 THz is released during the SHRS process (Fig. (d)). Similarly, the collimated UV light is generated from the FWM process.

3. Results and discussion

3.1 SERS process

When the P1 pulse resonated with the 6²S_1/2→6²P_3/2 transition, the strongest fluorescence of the Cs D₂ line was observed, and plenty of Cs atoms were expected to populate on the 6²P_3/2 state. When the P2 pulse was set at 550.20 nm (off resonant with the 6²P_3/2→11²D_5/2 transition), a series of collimated UV and blue lights were detected by the spectrometer as shown in Fig. 2. This phenomenon indicated that these collimated UV and blue lights were generated by the FWM process. The strongest collimated UV light is the UV1 light which corresponds to 12²P_3/2→6²S_1/2 transition; the other collimated lights can be expressed as UV2: 11²P_3/2→6²S_1/2, UV3: 10²P_3/2→6²S_1/2, UV4: 9²P_3/2→6²S_1/2, UV5: 8²P_3/2→6²S_1/2, and B1: 7²P_3/2→6²S_1/2.

 

Fig. 2. A series of collimated UV and blue lights observed in the experiment.

Download Full Size | PDF

To further prove that these collimated UV and blue lights that are generated from the FWM process, the time-resolved spectra were obtained by using ICCD time-gating. The UV1 time-resolved spectrum was depicted in Fig. 3(a). The UV1 exhibited an excellent synchronization with P1 and P2. Subsequently, P1 and P2 was set to arrive in the cell at different times (Fig. 3(b)). Here, UV1 gradually disappears as the delay time between P1 and P2 becomes longer, which is in accordance with the FWM process.

 

Fig. 3. (a) Waveforms of UV1, P1, and P2 when P1 and P2are synchronized. (b) Variation inUV1 intensity at different delay times between P1 and P2.

Download Full Size | PDF

The resonant wavelength of the 6²P_3/2→11²D_5/2 transition is 550.288 nm. When P1 is resonant with the Cs D₂ line and the wavelength of P2 is 550.200 nm, the Cs atoms on the 6²P_3/2 state were excited to a vl that is higher than the 11²D_5/2 state; therefore, the THz radiation produced during the transition from the vl to the 12²P_3/2 state is involved in the FWM process, and a reasonable relationship with the frequency was expressed as:

where ω(UV1), ω(P1), ω(P2) and ω(vl→12²P_3/2) are frequencies of UV1, P1, P2 and THz respectively. After the wavelength of P2 was adjusted to 550.400 nm, the Cs atom transited from the 6²P_3/2 state to a vl lower than the 11²D_5/2 state. After the completion of the FWM process involving P1, P2, and THz radiation, collimated UV1 light was observed, as shown in Fig. 4(a). When the wavelength of P2 was tuned to 550.420 nm, another collimated UV light at 334.882 nm was collected and expressed as UV1′ (12²P_1/2→6²S_1/2) (Fig. 4(b)); the generation of UV1′ can be expressed as:

These experimental results verified our speculation that UV light was generated by the FWM process that involved THz radiations (Fig. 1); this indicated that the frequency of THz radiation could be adjusted by varying the pump laser wavelengths, and the generation of THz radiation occurred most likely through the SRS process. When the wavelength of P2 was tuned to 550.400 nm, THz Stokes light 1 (vl→12²P_3/2) would be enhanced by the near-resonant 11²D_5/2→12²P_3/2 transition, and the strong UV1 light was generated via FWM process, in fact, collimated UV1′ light should have the opportunity to be generated by FWM process with the coupling of THz Stokes light 1′ (vl→12²P_1/2), however, THz Stokes light 1′ is much weaker than THz Stokes light 1, so it is too hard to detect collimated UV1′ light. After adjusting the wavelength of P2 to 550.420 nm, both THz Stokes light 1 (vl→12²P_3/2, near-resonant with 11²D_5/2→12²P_3/2 transition) and THz Stokes light 1′ (vl→12²P_1/2, near-resonant with 11²D_3/2→12²P_1/2 transition) are generated, then collimated UV1 and UV1′ lights are generated by FWM processes respectively, different from the stable UV1 light, although UV1′ light was detected successfully, the signal was unstable and weak, because SERS effect was not strong for the generation of THz Stokes light 1′.

 

Fig. 4. UV spectra observed at different wavelengths of P2. (a) UV1 spectrum under the condition of P1 resonated with Cs D₂ line and P2 at 550.400 nm. (b) UV1 and UV1′ spectra under the condition of P1 resonated with Cs D₂ line and P2 at 550.420 nm.

Download Full Size | PDF

To prove the existence of the SRS process, the intensity variation of UV1 was recorded for a wavelength range of 550.000 to 550.600 nm in P2, as shown in Fig. 5. By smoothing the experimental data, three peaks (Peak1, Peak2, and Peak3) and two valleys (Valley1 and Valley2) were observed. Valley1 corresponds to P2 at 550.289 nm, which is in accordance with the resonant wavelength of 6²P_3/2→11²D_5/2 transition (550.288 nm). Furthermore, Cs atom in the 6²P_3/2 state would be excited to the 11²D_5/2 state; subsequently, a transition occurs from the 11²D_5/2 state to 12²P_3/2 state, with the THz ASE emission (Fig. 1(b)), the condition of both P1 and P2 resonantly pumping Cs atoms satisfies ASE mechanism. For Valley2, the wavelength of P2 is 550.390 nm, which is approximately resonant with 6²P_3/2→11²D_3/2 transition (550.386 nm). Taking Valley1 as an example, for a resonant pump, the intensity of UV1 decreases because P2 is absorbed strongly by 6²P_3/2 Cs atoms, and plenty of these atoms will populate on 11²D_5/2 state, THz ASE (11²D_5/2→12²P_3/2) is generated, but the absorbed pump laser will not completely convert to ASE, then only the little residual pump laser is involved in FWM process, and the efficiency of frequency up-conversion is decreased, that causes the intensity of UV1 reduced in the condition of resonant pump. During the variation in the UV1 intensity, Peak1 and Peak2 correspond to P2 at 550.226 nm and 550.342 nm, respectively, along with their values of ω(P1) + ω(P2) are 29901.6 cm⁻¹ and 29897.9 cm⁻¹, respectively; their deviations from the 11²D_5/2 state (29899.5 cm⁻¹ = 334.357 nm) are +2.1 and −1.6 cm⁻¹, respectively, which implies that the strong THz radiation is obtained during the vl offset 11²D_5/2 state due to a non-resonant pump; this result is consistent with the SERS mechanism [31,32], and SERS is suppressed obviously when pumping Cs atoms at the resonant level [33], as depicted in Fig. 1(c). Similarly, at Peak3, P2 excites the Cs atom in the 6²P_3/2 state to the vl offset 11²D_3/2 state, and THz radiation is generated through the SERS process. When Cs atoms are excited to 6²P_3/2 state by P1 and P2 non-resonantly pumps Cs atoms (6²P_3/2) to a vl that is near 11²D_5/2 state, then the THz Stokes lights are generated via the SERS process and intensity variation of collimated UV lights is consistent with the “SERS-FWM” mechanism.

 

Fig. 5. Variation in the intensity of collimated UV1 light with the wavelength of P2.

Download Full Size | PDF

3.2 SHRS process

In addition to SERS, SHRS is another type of SRS, and its mechanism diagram was shown in Fig. 1(d); P1 excites Cs atoms to a vl′ that is near the 6²P_3/2 state, while P2 continues to pump Cs atoms to vl and the THz radiation associated with the vl→12²P_3/2 transition is generated by the SHRS process. The SRS effect will impact the intensity of the collimated UV light from the FWM process [22]. For an in-depth investigation, the intensity of UV1 light was collected while scanning wavelengths of both P1 (851.600–852.700 nm) and P2 (550.000–550.600 nm), as shown in Fig. 6. When the wavelength of P1 that resonates with the Cs D₂ line is fixed, the variation in UV1 with P2 is illustrated by the black dashed line and is marked as the Cs D₂ line in the picture; this result is consistent with the FWM process involving SERS (Fig. 5), and the red or orange regions on the black dashed line correspond to Peak1 and Peak2 in Fig. 5.

 

Fig. 6. Intensity variation of collimated UV1 light with the wavelengths of P1 and P2.

Download Full Size | PDF

In Fig. 6, P1 and P2 excite the Cs atom to a vl; if this vl is near the 11²D_5/2 state, the obtained UV light will be strong and tend to concentrate in the red to yellow regions. If the deviation between vl and the 11²D_5/2 state becoming large, the intensity of UV1 decreases quickly; this suggested that the obtained THz radiation is strong when vl deviates the 11²D_5/2 state within a certain range, which is in agreement with the characteristic of SHRS. Just as shown in Fig. 6, the variation of ∼1.0 THz signal intensity could be speculated by UV1, the intensity in green regions is taken as a lower limit, it is speculated that ∼1.0 THz signal could be adjusted from 1.01 to 1.11 THz, around 0.1 THz detuning range.

If we consider vl as a reference value that corresponds to Peak1 (vl = 29901.6 cm⁻¹) or Peak2 (vl = 29897.9 cm⁻¹) in the SERS mechanism and mark the combination of P1 and P2, meeting the value of ω(P1) + ω(P2) at approximately 29901.6 cm⁻¹ and 29897.9 cm⁻¹, respectively, as oblique lines (made up of short dots) shown in Fig. 6 with the signs of Ridge1 and Ridge2; the Cs D₂ line belonging to the SERS condition could be used as a boundary; consequently, in view of ω(P1) + ω(P2) being a fixed value, the SRS effect on Ridge1 and Ridge2 will be symmetrical around the Cs D₂ line. Since the strength distribution of the THz Stokes light generated by the SHRS process is symmetrical [34], according to the FWM process in the collinear condition in Eq. (1), the intensity of UV1 should also be distributed symmetrically. From the experimental results shown in Fig. (6), only Ridge2 is consistent with the expectation above; but the strength distribution is evidently asymmetric between each end of Ridge1. The possible reason may be phase mismatching in the FWM process, which is a result of the refractive index associated with the wavelength of the pump laser being changed dramatically at the resonant position in alkali vapor.

The refractive index of Cs vapor was calculated using the standard Sellmeier equation [35]

The pump lasers are linearly polarized in the experiment, so q=0 [38], and <l‖r‖l’> is the reduced dipole matrix element [39]. Where the braces denote a Wigner-3j symbol (Wigner3j), and curly braces denote a Wigner-6j symbol (Wigner6j).

For relaxation parameters that are not introduced, Sellmeier equation is not suitable for the calculation of the refractive index around the resonant points, which means the results near the resonant points in Eq. (3) is imprecise. When using a strong pump, many Cs atoms will be populated in the 6²P_3/2 state and almost equal to the number of Cs atoms in the ground state. Considering the broad linewidth of P1, there is a spectrum component that resonates with the Cs D₂ line when adjusting wavelength. Supposing that the population inverses completely, the refractive index will also inverse symmetrically [40]. Figure 7(a) shows the refractive index (n₁-1) of P1 under the condition that all atoms are populated on the 6²S_1/2 or 6²P_3/2 states by using Eq. (4). In these two situations, the initial and final states of the transition are opposite to each other, thereby causing the oscillator strength to change from positive to negative; therefore, the two types of refractive indices are symmetric. In the experiment, large amount atoms populate the 6²P_3/2 states; therefore, the refractive index of P1 should be supposition of these two parts together. After calculation, the Cs vapor atom density in the 6²P_3/2 states (N_6P) is 9.11 × 10¹⁴ cm⁻³, and that in the 6²S_1/2 states (N_6S) is 9.13 × 10¹⁴ cm⁻³; further, the distribution of the refractive index with P2 was calculated, as depicted in Fig. 7(b).

 

Fig. 7. Refractive index of the pump laser in Cs vapor. (a) The distribution of refractive index n₁-1 with P1. (b) The distribution of refractive index n₂-1 with P2.

Download Full Size | PDF

In the FWM process, the phase-matching condition involving the refractive index should satisfy the expression:

If Cs atoms are excited to vl which is a fix value, the frequency of THz Stokes light will not vary, therefore, n₃ is a fixed value, while the wavelength of UV1 is invariable, n₄ is also a fixed value, and phase matching is decided by the variation of the refractive index with P1 and P2. During the SERS process, the phase matching condition of Peak1 can be easily satisfied, while the wavelength of P1 is resonant with the Cs D₂ line (11732.3 cm⁻¹), and n₁ is 1.00. For P2, the wavelength is 550.226 nm (18169.3 cm⁻¹), here the refractive indexes of P1 and P2 with Peak1 are considered in the vacuum condition, the values of n₁ and n₂ are both 1.00, therefore, n₁[λ(P1)]⁻¹ + n₂[λ(P2)]⁻¹ = 29901.6 cm⁻¹. This calculated result is taken as a reference value after comparing the phase matching of all the points about P1 and P2 on Ridge1 in Fig. 6, the situation of phase mismatching is obtained, and the value of phase mismatching is overlay on the wave number of P1 (Fig. 8). The sides at the top located between Ridge1 and Ridge1 phase matching exhibit significant phase mismatching; however, for the sides at the bottom, these two lines coincide better and the phase matching is quite satisfactory. This result is in accordance with the strength variation in UV1 on Ridge1 (Fig. 6), and successfully explains the reason behind the asymmetric distribution of the UV1 strength on Ridge1.

 

Fig. 8. Phase matching on Ridge1 and Ridge2.

Download Full Size | PDF

For Ridge2 (Fig. 6), the phase matching is satisfactory in the SERS process; at this moment, the wavelength of P2 is 550.342 nm (18165.6 cm⁻¹), also we take the refractive index (n₂) in the vacuum condition, then the reference value is obtained: n₁[λ(P1)]⁻¹ + n₂[λ(P2)]⁻¹ = 29897.9 cm⁻¹. We use the same data processing method for Ridge2 (Fig. 8), both the top sides and bottom sides between Ridge2 and Ridge2 phase matching coincide better, so phase matching is quite satisfactory. The distribution of the UV1 strength on Ridge2 in Fig. 6 is symmetrical, this demonstrates that the explanation of phase matching is correct again.

In the SHRS process, the SRS effect and phase matching affect the FWM process, resulting in the strength of UV1 to be varied. In Fig. 5, the wavelength of P1 is resonant with the Cs D₂ line in the SERS process, where in the SRS effect varies with the wavelength of P2 and directly determines the intensity of the THz Stokes light; under this condition, the FWM process is dominated by the SRS effect, and the phase matching is secondary.

3.3 THz pulse energy evaluation

Since the lacking of a direct THz measuring equipment at present, so THz pulse energy was evaluated with the help of the collimated UV light, but this method could only estimate the minimum energy of THz pulse. Optical filters were used to remove the residual pump lasers, and prisms were employed to separate UV1 from the other collimated UV lights. The stray lights from pump lasers and collimated UV lights would disturb the energy measurement of the UV1 light, so the spectrometer was used to find out the spread direction of the UV1 light, in the other direction which is perpendicular to the direction of UV1 light, the energy probe was moved at different positions to measure UV1 pulse energy, the knife-edge measurement of UV1 was depicted in Fig. 9. After eliminating the interference from stray lights, UV1 pulse energy was obtained to be around 10 nJ, this experimental result has not been corrected with the optical loss from windows of cell, optical filters and prisms. According to Manley-Roew relationship, there must be at least one THz photon involved in FWM process to generate one UV1 photon, so the minimum number of 1.0 THz photons can be obtained by the number of UV1 photons, and the minimum 1.0 THz pulse energy is estimated to be 12 pJ. In our previous work, the THz emission quantum efficiency is estimated to be 1‰–1% [30], so 1.0 THz pulse energy is evaluated to be on the order of nJ. THz radiation in this work is a pulsed signal, and its pulse-width should be larger than that of UV, and smaller than those of P1 and P2. Pulse-widths of UV1, P1 and P2 are 3 ns, 5ns and 6 ns respectively, therefore, the pulse-width of THz is expected to be 3∼5ns. So the peak power of THz reaches the magnitude of Watt, alkali vapor THz source has the potential to be used in the techniques of absorption spectrum and spectral imaging, which play a role in the biomedicine area.

 

Fig. 9. Knife-edge measurement of UV1.

Download Full Size | PDF

4. Conclusion

The collimated UV and blue lights were generated by utilizing two-color two-photonics excitation of Cs atoms, after investigating time-resolved characteristics of pump lasers and collimated light, it was demonstrated that the collimated light was generated by FWM process. By varying the wavelengths of the pump lasers, the vl could be adjusted; it has been found that THz radiation was involved in the FWM process and the frequency of THz emission could be adjusted. THz radiation can be generated by the SRS effect, and the mechanism based on ASE, SERS, and SHRS was analyzed specifically. When Cs atoms are excited to the 6²P_3/2 state due to P1 resonating with the Cs D₂ line and P2 non-resonantly pumps Cs atoms (6²P_3/2) to vl (with the deviation of about +2.1 cm⁻¹ and −1.6 cm⁻¹ from 11²D_5/2 state, respectively), the collimated UV1 light reaches its maximum. This implies that THz radiation is strong at the position of the vl offset 11²D_5/2 state; this characteristic is evidently consistent with the SERS mechanism. In the SERS process, the SRS effect dominated the FWM process. If P2 resonantly excites Cs atoms (6²P_3/2) to the 11²D_5/2 state, THz radiation with the 11²D_5/2→12²P_3/2 transition will be realized in the form of ASE, as demonstrated by the decrease in UV1 intensity. Cs atoms that are non-resonantly pumped to vl by P1 and P2 together, we can observe the strong collimated UV light only when the vl deviated from 11²D_5/2 state in a certain range; at this situation, THz Stokes light is strong, and satisfies the SHRS mechanism. In SERS process, the SRS effect is pretty strong in Peak1 and Peak2; When taking the corresponding vl values as a reference, while establishing different combinations of P1 and P2 to meet this reference value, the strength distribution of the THz Stokes light generated by SHRS process should be symmetrical; this result should also be observed for UV light. However, the strength distribution of UV1 was observed to be asymmetric between each side of Ridge1, but for Ridge2, the strength of UV1 distributed symmetrically and satisfied the expectation mentioned above, according to the variation in the refractive index of the pump laser in Cs vapor, the asymmetric phenomenon was explained successfully by phase matching. The SRS effect and phase matching affect the FWM process together. At present, the conversion efficiency of THz emissions is not optimum, and it is difficult to detect the THz signal, we only rely on the collimated UV light to evaluate the character of THz emissions, and 1.0 THz pulse energy is estimated to be on the order of nJ. In the next step of our plan, we will improve the conversion efficiency of THz emissions, and directly measure the THz signal to evaluate the character of THz emissions and THz influence on the FWM process.