1. Introduction

Air breakdown induced by intense light has been a well-known phenomenon since the invention of the laser [1, 2]. During the past decades, laser induced air breakdown that leads to plasma generation under different focal conditions, including the case of shallow focusing leading to the filamentation of the propagating light, has been extensively investigated both theoretically and experimentally [2–5]. Recording of plasma emission images and time-resolved shadowgraphs has been used as the effective investigation method of the laser induced air breakdown both in weak and strong ionization regimes [6–8]. However, the phenomena and dynamics of strongly ionized air plasma induced by focused high intensity double femtosecond laser pulses have not been paid enough attention.

Investigation of air plasma generated by double femtosecond laser pulses can provide not only important insights to the laser-plasma or laser-shock wave interaction but also a special scheme for examining Thomson scattering. The latter is an important diagnosing tool for measuring the plasma properties [9–11]. Moreover, study of double femtosecond laser pulse air ionization also benefits some key technological applications that often employ double laser pulses to ablate the target, such as LIBS, and femtosecond laser micro-machining [12, 13].

In this paper, experimental results of both plasma emission images and time-resolved shadowgraphs are reported to reveal the interesting dynamic characteristics of strong air ionization produced by two focused and collinear femtosecond laser pulses. A fixed time interval of 10 ns and different pulse energy ratios apply to the double laser pulses. The polarization characteristics and intensity of the scattered second laser pulse by the first pulse generated shock wave are also studied. Coherent Thomson scattering enhanced by plasma instabilities is deduced to be the main scattering mechanism involved in our observations.

2. Experimental setup

Figure 1 shows the schematic diagram for recording the time-resolved shadowgraphs of air plasma induced by single or double femtosecond laser pulses. A commercial Ti:sapphire femtosecond laser amplifier system (HP-Spitfire, Spectra-physics Inc.) is employed, which can generate 50 fs, 2 mJ, 1 kHz laser pulses with a central wavelength of 800 nm. The output femtosecond laser beam with a smooth spatial distribution of a Gaussian shape is split into two beams which are respectively used as the pump beam to ionize the ambient air and the probe beam to record the time-resolved shadowgraphs of air plasma.

 

Fig. 1 Schematic diagram for recording the time-resolved shadowgraphs and plasma emission images of air plasma produced by femtosecond laser pulses (top view). The optical components surrounded by the dashed rectangles are employed only for certain cases explicitly indicated in the text. The distances from the CCD camera to the steering mirror B, the 4 × objective, the polarizer, and the filter stack are respectively ~1300 mm, ~912 mm, ~650 mm, and ~550 mm. PD: photodiode.

Download Full Size | PDF

In our experiments, the double laser pulses are obtained by adjusting the pulse dumping time of the regenerative amplifier, so that the single amplified laser pulse inside the amplifier resonator is converted into two successive output laser pulses. The two pulses are 10 ns apart, being determined by the regenerative cavity round-trip time, and totally collinear.

When the pulse dumping time is adjusted, the effective grating spacing of the pulse compressor, being optimized for compressing the second laser pulse, is kept unchanged. And the dumping time of the second laser pulse remains the same as that in single pulse operation mode. So the temporal duration of the second laser pulse is 50 fs.

The second order dispersion for each cavity round-trip of the regenerative amplifier we adopt is approximately 7130 fs². So the duration of the first laser pulse, which is dumped out one round-trip time earlier than the second laser pulse, may be estimated to be ~400 fs after passing through the compressor. It can be deduced that the double laser pulses used in our experiments contain no additional amplified spontaneous emission (ASE). And also it is obvious that adjustment of the dumping time does not change the extinction of the Pockel-cell-based optical switch in the regenerative amplifier. Therefore, the contrast ratio of double laser pulses will decrease little compared with the ordinary case of single laser pulse mode.

As the first laser pulse possesses negative second order dispersion because of over compensation by the pulse compressor, it may get compressed on its way to reach the air ionization region in accounting for the positive dispersion introduced by the optical components outside the laser amplifier system. Along the same thought, the duration of the second laser pulse after focused by the 10× objective will become longer than 50 fs. However, such a pulse broadening effect will not make the second pulse longer than 100 fs because it takes about 30 mm BK7 glass to broaden a laser pulse from 50 fs to 100 fs whereas the equivalent amount of glass along the pump beam path is obviously less than 30 mm in thickness.

As shown in Fig. 1, a 10× objective (NA = 0.25) is employed to focus the pump beam. The time delay between the pump and probe pulses can be adjusted with the optical delay line. The probe beam is frequency doubled with a 2-mm-thick BBO crystal with Type I phase matching. As the frequency doubled efficiency is ~10%, the power of the second harmonic probe beam is ~20 mW in this case. The time-resolved shadowgraphs of air plasma are recorded with a 4 × objective (NA = 0.1) and a CCD camera (LU135M, Lumenera Inc.). A 400 nm band pass filter and some neutral density filters are inserted behind the imaging objective to block the air plasma emission and the residual 800 nm fundamental light mingled within the probe beam from entering the CCD camera. For the air plasma produced by double laser pulses, each shadowgraph is actually the superposition of two femtosecond time-resolved pictures respectively recorded by the two 400 nm probe pulses. When the probe beam is blocked and the band pass filter is removed, the setup in Fig. 1 also can be used to record the air plasma emission images. Throughout the experiments, the pump laser pulses are monitored by a photodiode with a rise time of 1 ns (DET210, Thorlabs Inc.) which is connected to a 500 MHz oscilloscope (DL9040, Yokogawa Inc.).

3. Results and discussions

In this paper, the pump laser always propagates from the left to the right in all the plasma emission images and shadowgraphs. The plasma emission image of air ionization induced by horizontally polarized double femtosecond laser pulses with a total energy of 0.33 mJ is presented in Fig. 2(b) , which is greatly different from that in Fig. 2(a) produced by a single horizontally polarized 50 fs, 0.33 mJ laser pulse. The pump laser pulses used in Figs. 2(a) and 2(b) are monitored by a fast photodiode with a 500 MHz oscilloscope (see Fig. 1) and the obtained oscillograms are shown in Figs. 2(c) and 2(d) respectively.

 

Fig. 2 Plasma emission images of air ionization induced by a single (a) and double (b) femtosecond laser pulses, being recorded by a black and white CCD camera with a protection window. (Frame size: 220 μm × 164 μm.) The distance between the protection window and the CCD sensor is ~6 mm. The pulse energy of the single laser pulse used in (a) is 0.33 mJ. The pump laser pulses used in (a) and (b) are monitored by a fast photodiode and an oscilloscope, and their oscillograms are respectively shown in (c) and (d). From (d), the first and second laser pulses’ energy is estimated to be 0.13 mJ and 0.2 mJ respectively and the first-to-second pulse energy ratio is 0.65.

Download Full Size | PDF

From Fig. 2(b), it is easily noticed that some extra bright spots exist outside the laser propagation path, such as those circled by the dashed ellipses. These spots disappear when the protection window of the CCD camera is removed, which is evidenced in Fig. 3(a) . Besides, it can be also found from Fig. 2(b) (and more clearly from Fig. 4(g) given in the following) that the bright dots generated due to the existence of the protection window actually constitute a rectangular lattice. So it is considered that the strong light forming the bright spots along the laser propagation path is backward diffracted by the pixel array of the CCD image sensor, striking on the protection optical window, and then is reflected back to the CCD image sensor again, generating the rectangular dot array. Light reflection due to the protection window may generate complex bright dot pattern in the recorded plasma emission images because of multiple reflections at the two surfaces of the protection window. The specific pattern of these sideward bright spots may also differ from case to case due to the slight variation in placing the CCD camera relative to the direction of the light forming the bright spots. Detailed discussions of this peculiar phenomenon will be presented in a separate paper [14].

 

Fig. 3 Plasma emission images of air plasma induced by the femtosecond laser pulse pair of the same parameters as those used in recording Fig. 2(b), except that the protection window of the CCD sensor is removed. In obtaining (a), no wavelength-selective filters are used after the imaging objective, which is in contrast to (b) that is recorded with a short wave pass filter of a high rejection within the spectral range from 750 nm to 875 nm. Frame size: 220 μm × 164 μm.

Download Full Size | PDF

 

Fig. 4 Plasma emission images (left column) and corresponding time-resolved shadowgraphs (right column) of air plasma induced by double femtosecond laser pulses with different pulse energy ratios. The total energy of the pump laser pulse pair is all set at 0.28 mJ. The plasma emission image and the corresponding shadowgraph in the same row are recorded for the same pulse energy ratio. The first-to-second pulse energy ratios are 3.2, 1, 0.2, 0.07, and 0.02 for (a, b), (c, d), (e, f), (g, h), and (i, j) respectively. For all the shadowgraphs a time delay of 13 ps exists between the pump and the probe beams. Frame size: 220 μm × 164 μm.

Download Full Size | PDF

In Fig. 3(b), when a short wave pass filter (700FL07-25, Andover Inc.) with a transmission rate less than 1% for the spectral range of 750 nm - 875 nm is placed before the CCD camera with the protection window removed, the brightness of these two dots in Fig. 3(a) decreases dramatically. These two bright dots in Fig. 3(a) are thus considered to be not due to the air plasma emission which normally covers the whole visible light spectrum [15]; instead, their high brightness can only be attributed to the strong scattering of the second laser pulse by the first pulse induced air plasma or shock wave.

Also, in Fig. 3(b) three glowing regions with the middle one rather weak may be identified, which are mostly the emission of the plasma induced by the double femtosecond laser pulses. (The small bright spot on the right is likely to be the remaining 800 nm scattered light of the second laser pulse.) Such a spatial distribution of the air plasma is still substantially different from that induced by a single laser pulse shown in Fig. 2(a). For clarity, the experimental conditions and results presented in Figs. 2 and 3 are summarized in Table 1 .

Table 1. Summary of the experimental conditions and results in Figs. 2 and 3

View Table

Finally, it is noted that some concentric circular stripes exist around the large brighter spot in Fig. 3(a), which are expected to be resulted from the diffraction effect of the limited aperture of the imaging lens [16]. The appearance of these circular fringes in fact indicates that the region of the first pulse induced plasma that strongly scatters the second laser pulse is actually very small in space and much like a point source such that the aperture effect of the imaging system can be clearly brought out.

3.1 Dynamic process of air breakdown by double femtosecond laser pulses with different pulse energy ratios

In order to investigate the dynamic process of air breakdown by double femtosecond laser pulses, the pump-probe technique is used to record the transient states of the air plasma generated by double femtosecond laser pulses with various pulse energy ratios. The results together with the corresponding plasma emission images are presented in Fig. 4.

The plasma emission images in the left column of Fig. 4 are recorded using the CCD camera with its optical protection window but without any color filters. The total energy of the pump femtosecond laser pulse pair is fixed at 0.28 mJ. A half-wave plate is used to make the incident laser beam vertically polarized (in order to be able to record the strongest light scattering). The time-resolved shadowgraphs in Fig. 4 are recorded when the 400 nm probe pulse pair are 13 ps behind the pump pulse pair. So the first probe pulse can record the transient state of air plasma induced by the first pump pulse at a time delay of 13 ps; and the second probe pulse can record not only the air plasma induced by the second pump pulse at 13 ps time delay, but also the air plasma induced by the first pump pulse at ~10 ns time delay.

For Figs. 4(a) and 4(b) the first laser pulse’s energy is three times larger than the second one. In Fig. 4(b), a narrow dark channel can be seen inside an oval shock wave. This dark channel represents the transient state of air plasma [17] generated by the first pump pulse and recorded by the first probe pulse at 13 ps time delay. The oval shock wave in Fig. 4(b) actually evolves from this narrow dark channel and is recorded by the second probe pulse at ~10 ns time delay. The second pump pulse hardly ionizes the air because its focus actually locates in the micro vacuum zone inside the first laser pulse induced shock wave [18]. It is evident in Fig. 4(b) that the second pump pulse does not ionize the air outside the shock wave neither due to its low light intensity. In this case, no scattered 800 nm light appears in Fig. 4(a), which may result from both the lower intensity of the second laser pulse and less plasma density at both ends of the shock wave. (In Fig. 4(b), the shock wave seems widely open at its left and right ends.) Only an oval-shaped bright region is observed in Fig. 4(a), which is formed due to the air plasma emission induced by the first laser pulse.

Based on the results shown in Figs. 4(c)-4(h), the representative characteristics of air plasma induced by double femtosecond laser pulses can be properly described. As already discussed above, the oval-shaped plasma emission region in Fig. 4(c), a narrow dark channel and a shock wave appear in Fig. 4(d), are all generated due to the first pump pulse. For the case of smaller pulse energy ratio, although the plasma emission induced by the first pump pulse is not so obvious the first pulse indeed ionizes the ambient air because a shock wave appears in the corresponding shadowgraph (see Figs. 4(f) and 4(h)). The second pump pulse is strongly scattered by the ionized air just behind the ionization front of the shock wave. This is evidenced by the fact that the distance between the two brightest scatter spots in the plasma emission image is more than 96% consistent with the distance between the left and right ends of the shock wave front at the time that the second pump pulse arrives (indicated by the solid lines in Figs. 4(c) to 4(h)). As the first-to-second pulse energy ratio decreases, the vacuum volume inside the shock wave gets much smaller and the second pump pulse becomes more powerful, so the second pump pulse can ionize not only the compressed air shell just behind the shock wave front but also the ambient air outside the shock wave. It is seen from Figs. 4(d), 4(f) and 4(h) that the ionization region generated by the second pump pulse is divided into two parts by the micro vacuum volume inside the shock wave [18]. This explains the origin of the multiple plasma emission regions produced by double femtosecond laser pulses with proper pulse energy ratios, such as the case in Fig. 3(b).

For Figs. 4(i) and 4(j), the pulse energy ratio further decreases, and the ambient air is only weakly ionized by the first pump pulse. However, even in this case the air plasma channel induced by the second pump pulse is still divided into two parts (see Fig. 4(i)) because of the lower air density region generated by the expansion of the first pulse induced air plasma. In Fig. 4(i), no scattered 800 nm light is observed because the air breakdown and shock wave induced by the first pulse are too weak. Based on the results shown in Fig. 4 and the fact that the intensity of the scattered light is determined by the intensity of the second pump pulse as well as the electron density of the air plasma, strong scattered light is thus expected to appear for a moderate pulse energy ratio, such as the case in Fig. 4(e).

As indicated by the dashed line in Fig. 4, when the first laser pulse energy decreases, the center of the shock wave and thus the air breakdown region moves slightly towards the right side of the shadowgraph, i.e., away from the focal objective. This is mainly because for higher pump pulse energy, the position where the intensity of the focused laser pulse is above the air ionization threshold will be closer to the focal lens. For larger pulse energy ratio, the center of the shock wave is thus more adjacent to the focal objective than the focused Gaussian beam waist, so the intensity of the second pump pulse at the right end of the shock wave will be higher than that at the left end. This explains why in the plasma emission images such as Figs. 4(c), 4(e) and 4(g), stronger scattered light always occurs at the shock wave’s right end.

3.2 Polarization characteristics of the scattered light

In this section we will study the polarization characteristics of the scattered light to analyze its scatter mechanism. To do so, a half-wave plate is inserted in the pump laser beam for polarization adjustment. When the half-wave plate is rotated, the intensity of the scattered light is found to change accordingly. In other words, the scattered light intensity is sensitively dependent on the polarization of the pump laser. In particular, as shown in Fig. 5 which is recorded by the CCD camera with the protection window, the minimal and maximal intensities of the scattered light appear when the polarization direction of the linearly polarized pump pulse pair is set to be horizontal and vertical respectively. It indicates that the strongest scattering occurs in the direction perpendicular to the polarization direction of the pump laser.

 

Fig. 5 Plasma emission images of air ionization induced by double femtosecond laser pulses with horizontal (a) or vertical (b) polarization. The total energy of the pump pulse pair is 0.33 mJ and the first-to-second pulse energy ratio is 0.65. It can be noted that in Fig. 5(b), because of the strong scatter intensity, the scattered light from the left weaker scatter spot (see Fig. 3(a)) may also form an observable dot array; and due to the reflections from two surfaces of the CCD camera’s protection window, additional spots may appear, both of which are the reasons why the more complex spot pattern appears in Fig. 5(b). Frame size: 220 μm × 164 μm.

Download Full Size | PDF

Further experimental results on the polarization characteristics of the scattered light are shown in Fig. 6 . In this case, the vertically polarized pump pulse pair with a total energy of 0.33 mJ and a first-to-second pulse energy ratio of 0.65.is employed, and sum of pixel values of a particular bright spot (the one surrounded by the dashed ellipse in Fig. 5) is used to represent the intensity of the scattered light. A Glan-Taylor prism polarizer is placed before the CCD camera. During the experiment no alignment is made except changing the orientation of the polarizer. The “0°”/“90°” in Fig. 6 denotes that the polarizer is orientated to make the horizontally/vertically polarized light have the maximal transmittance. When the polarizer has an orientation of 0°, although the non-zero value of the intensity of the scattered light is present in Fig. 6, it actually represents the background intensity of the plasma emission image because the scatter spot is hardly observed. From Fig. 6 it can be seen that the intensity of the scattered light is maximum when the pump laser’s polarization is vertical and the polarizer is oriented to pass the vertical polarization.

 

Fig. 6 Polarization characteristics of the scattered light when vertically polarized double femtosecond laser pulses are employed to ionize the ambient air. The total energy of the pump pulse pair is 0.33 mJ and the first-to-second pulse energy ratio is 0.65. For each orientation of the polarizer two data points are obtained. The two data points obtained when no polarizer is used are too close to be distinguished, partially because saturation occurs for the selected pixels.

Download Full Size | PDF

Based on the experimental results shown in Figs. 5 and 6, it can be concluded that the strongest scattering is observed in the direction normal to the polarization of the pump pulses and the polarization of the scattered light is the same as that of the pump laser, which is essentially the intrinsic characteristics of the dipole radiation [19].

Moreover, it is known that in the direction normal to the pump laser propagation direction, light scattering is possible to be originated from dipole radiation of either free charged particles (Thomson scattering) or neutral molecules’ polarization (Rayleigh scattering), or radiation due to dust particles’ polarization (Mie scattering) [20, 21]. Because when a single femtosecond laser pulse is employed to ionize the ambient air no obvious scattered light can be observed in the plasma emission image (see Fig. 2(a)), it is thus inferred that Rayleigh scattering and Mie scattering hardly contribute to the light scattering observed in Figs. 2-5. Therefore, Thomson scattering is believed to be the main mechanism that causes strong light scattering in Figs. 2-5. (Note that, while theoretically for Thomson scattering one would not expect to see any scattering along the dipole axis, the 4 × imaging objective (NA = 0.1) used in our experiments actually collects the scattering light in a solid angle of 0.0315 sr, which may be the reason why certain amount of scattered light is still detected as shown in Fig. 5(a).)

3.3 Mechanism of the extremely strong scattered light

To further analyze the scatter mechanism, the scatter ratio is measured through using a calibrated silicon photo detector (818-SL, Newport Inc.) and a power meter (2832-C, Newport Inc.). During the measurement, the probe beam is blocked, and the CCD camera is replaced by the photo detector with all the filters and the polarizer removed (see Fig. 1). So the scattered power measured by the photo detector is only the scattered power in a specific direction which is collected by the 4× imaging lens. When the vertically polarized laser pulse pair’s energy ratio is set to 1:1 and the repetition rate of the pulse pair is 1 kHz, the ratio of the measured average scattered power from the right scatter spot in the plasma emission image (see Fig. 3(a)) to the average power of the second laser pulse is determined to be ~1 × 10⁻⁵.

For the incoherent Thomson scattering (ITS) that occurs when kλ_D$\gg$1, where k is the wave number of the electron density fluctuation, and λ_D is the electron Debye length [22], the scattered power may be calculated by simply adding the contributions from individual electrons because the correlations between the plasma electrons and ions are negligible. In other words, the ITS power is proportional to the total number of the electrons within the scatter volume.

For a non-relativistic plasma in thermodynamic equilibrium whose electron plasma frequency is much lower than the frequency of the incident laser beam, the ratio of the ITS power P_s to the incident laser power P_i in the direction perpendicular to both the wave vector and polarization direction of the incident laser can be expressed as [23]:

Based on the recorded time-resolved shadowgraphs, the scatter volume in the shock wave front is estimated to be πr²L = π(13 × 10⁻⁴)² × (16 × 10⁻⁴) = 8.5 × 10⁻⁹ cm³ and also A = πr² = 5.3 × 10⁻⁶ cm². The solid angle dΩ is 0.0315 sr because the scattered light is collected by a 4 × , NA = 0.1 objective. In order to make the scatter ratio equal to the measurement value of 1 × 10⁻⁵, the number of the electrons in the scatter volume should be up to 2.1 × 10¹⁶ and the corresponding electron density would be 2.5 × 10²⁴ cm⁻³. The particle number density of air under 1 atm is only 2.7 × 10¹⁹ cm⁻³. Therefore, even taking account that the air in the shock wave front could be several times denser than the standard air under 1 atm and multiple electrons can be ionized from one air molecule, an electron density of 2.5 × 10²⁴ cm⁻³ still cannot be obtained. Hence, ITS cannot be the main mechanism that induces the strong scattering observed in our experiments.

When kλ_D$\le$1, the correlations between plasma particles have to be taken into account and coherent Thomson scattering (CTS) occurs. Theoretically, the total electric field of the scattered light from the scatter volume is the sum of the scattered electric field of individual electrons. The time-averaged scattered power in solid angle dΩ can be expressed as [23]:

For stable plasma, the CTS power still scales as N just like the case of ITS. Similar to Eq. (1), when CTS occurs in a non-relativistic stable plasma, in the direction perpendicular to both the wave vector and polarization direction of the incident laser, we have $P_{s}d\Omega /P_i=[3{\sigma }_{T}Nd\Omega /(8\pi A)][S_T(\overrightarrow{k})/2\pi ]$, where S_T($\overrightarrow{k}$) is the integrated spectral density function which always has a small value [23]. For example, when T_e ≈T_i and ZT_e/T_i < 3 (Z is the charge number of the ions; T_e and T_i are the temperatures of electrons and ions respectively.),

In our experiments, when we obtain a scatter ratio of 1 × 10⁻⁵, the first laser pulse induced plasma behind the ionization front of the shock wave is most likely unstable, and furthermore, the second pulse has enough intensity to re-ionize and perturb the pre-ionized air, i.e., the leading part of the second pulse drives the instabilities in the scatter volume and these instabilities enhance the scatter of the trailing part of the second pulse. It is thus believed that in our case plasma instabilities lead to the strong scattered light whose power may scale as N².

Because it is difficult to derive equations that explicitly describe the scatter ratio for a strongly unstable plasma with multiple instabilities, we have to return to the basic expression for the scattered power of Eq. (2) and give simplified qualitative analyses about the mechanism of the strong scattering observed in our experiment by introducing the assumption of $E_s^2/2={\overline{(\overrightarrow{E_j}·\overrightarrow{E_l})}}_{j\ne l}$. The following analyses only indicate the possible scatter mechanism because the above assumption represents an extremely strong coherent state that the scattered electric fields from any two electrons in the scatter volume are coherent. Under this assumption, according to Eq. (2), we obtain (CTS Power)/(ITS Power) = N-1 ≈N, and using Eq. (1), we get P_CTSdΩ/P_i = 3σ_TN²dΩ/(8πA), where P_CTS is the CTS power. Based on the above analyses, in order to get a scatter ratio of 1 × 10⁻⁵, N² should equal to 2.1 × 10¹⁶, so N = 1.4 × 10⁸ and the corresponding electron density would be 1.6 × 10¹⁶ cm⁻³. Accounting for the air compression behind the shock wave front, and the re-ionization process of the pre-ionized air by the second laser pulse, such an estimated electron density is much more reasonable than that estimated using Eq. (1). Therefore, CTS enhanced by plasma instabilities is considered to be the main scatter mechanism in our experiment.

Besides, as already mentioned above, the regular array of bright spots in the plasma emission images, which is recorded by using the CCD camera with a protection window, is generated due to the diffraction of the scattered laser pulse by the pixel matrix of the CCD sensor. It is well known that the diffraction or interference effect is much more obvious for coherent light than for incoherent light. The clearly recorded spot array in the plasma emission images implies that the scattered light should be mainly composed of coherent light, which comes from the coherent Thomson scattering. This is thus another evidence that supports our analysis of coherent Thomson scattering taking place at the double laser pulse induced micro air plasma.

4. Conclusions

In conclusion, it is shown that when double femtosecond laser pulses with different pulse energy ratios are employed to ionize the ambient air, the air plasma generated by the first laser pulse may significantly affect the second pulse induced air plasma. For certain pulse energy ratios, the micro vacuum zone inside the first pulse generated shock wave can cause the air plasma induced by the second laser pulse to be divided into two parts in space, which leads to the observed multiple plasma emission sections. Noticeably strong scattering of the second laser pulse occurs at the ionization front of the first pulse induced shock wave. Coherent Thomson scattering enhanced by plasma instabilities is believed to be likely the main mechanism of such scattering. The investigation of air plasma produced by double femtosecond laser pulses not only can help understand the characteristics and basic dynamic processes of double pulse air breakdown, but also may benefit the researches of laser-plasma or laser-shock wave interaction. Some applications such as LIBS and femtosecond laser micro-machining, in which double laser pulses are used to ablate targets, can also benefit from the studies of the interesting phenomenon of double laser pulse induced air breakdown.