Generation of high-energy clean multicolored ultrashort pulses and their application in single-shot temporal contrast measurement

We demonstrate the generation of 100-{\mu}J-level multicolored femtosecond pulses based on a single-stage cascaded four-wave mixing (CFWM) process in a thin glass plate. The generated high-energy CFWM signals can shift the central wavelength and have well-enhanced temporal contrast because of the third-order nonlinear process. They are innovatively used as clean sampling pulses of a cross-correlator for single-shot temporal contrast measurement. With a simple home-made setup, the proof-of-principle experimental results demonstrate the single-shot cross-correlator with dynamic range of 1010, temporal resolution of about 160 fs and temporal window of 50 ps. To the best of our knowledge, this is the first demonstration in which both the dynamic range and the temporal resolution of a single-shot temporal contrast measurement are comparable to those of a commercial delay-scanning cross-correlator.

interaction phenomena in the relativistic region in the past decades [4][5][6][7][8][9][10][11]. In some of these important research areas, such as proton electron acceleration in thin solid targets [12][13][14] and electron generation in fast-ignition inertial confinement fusion [15,16], the temporal contrast of the driving laser pulse had been proved to be an extremely important parameter. For such an ultra-intense laser pulse, the prepulse and amplified spontaneous emission (ASE) noise prior to the main pulse are detrimental since their intensities might be high enough to ionize the target and thereby change its status [12][13][14][15][16], which makes high temporal contrast critically important for an ultra-intense laser pulse. Therefore, the characterization of temporal contrast of an ultra-intense laser pulse is the first important step towards temporal contrast improvement and precise analysis of these laser-matter interaction experiments.
In previous work, during temporal contrast measurement, the first and foremost process is the generation of a temporally clean sampling pulse [17][18][19][20][21][22]. For laser pulses with high repetition rates, a delay-scanning third-order correlator can be used to characterize the temporal pulse contrast [17,23]. In this case, a second-harmonic generation (SHG) process using a nonlinear crystal (such as BBO) generates a clean SHG pulse that is used as the sampling pulse. Then, in another nonlinear crystal, the third-order correlation signal is obtained through the sum frequency mixing (SFM) process between the SHG sampling pulse and the test pulse and detected using a highdynamic-range PMT. The dynamic range can be extended to 10 12 by adding gradient attenuators before the detector [22,23]. The temporal resolution can be controlled by step resolution of the delay-scanning stage and is also dependent on the group velocity mismatch between the sampling and test pulse in the crystal. However, the delayscanning third-order correlator requires thousands of pulses to accomplish the measurement, which is time-consuming and capable of only characterizing highrepetition stable laser pulses. Besides, since the temporal contrast improvement of the SHG process is not good enough because of a second-order nonlinear process, relatively intense postpulses following the SHG sampling pulse will lead ghost prepulses into the third-order cross-correlation signal, which will affect the measurement accuracy.
As most petawatt lasers are running in single-shot mode or at low repetition, measurement of the single-shot pulse temporal contrast is necessary. The crosscorrelator is the main method for high-dynamic single-shot temporal contrast measurement, where temporal contrast information is encoded into a line of signals in space using SFM in a nonlinear crystal with a large crossing-angle [18] . So far, the highest dynamic range with a single-shot measurement was obtained by the Qian group [21]. In their optical design, a 1040 nm clean sampling pulse is generated by using a typical optical parametric amplifier (OPA) scheme so as to single-shot characterize the temporal contrast of an 800 nm ultrashort laser pulse. The OPA system consists of an optical parametric generator (OPG) stage and an OPA stage using nonlinear crystals.
Since the final cross-correlator signal is collected through a fiber-array-based detection system, the temporal resolution is limited to about 0.7 ps, which is not precise enough for ultra-intense pulses with tens of femtosecond duration. The self-referenced spectral interferometry (SRSI) method was demonstrated recently by encoding the temporal contrast measurement into the frequency domain [24]. In this case, a resolution down to 20 fs can be obtained. However, the obtained best temporal contrast dynamic range has hereto been about 10 8 . Are there any methods that can simplify the generation of the clean sampling pulse? Is a new design possible to obtain high dynamic range and good temporal resolution simultaneously in a single-shot temporal contrast measurement?
Spatially and spectrally separated multicolored femtosecond sidebands are observed when two femtosecond laser pulses with different wavelengths are focused into a glass plate with a small crossing angle. As a third-order nonlinear process, the generated firstorder CFWM signal has a cubic dependence on the intensity and its temporal contrast would be improved drastically, similarly to the effect of the self-diffraction (SD) and crossing polarization generation (XPW) process [30][31][32]. Furthermore, the temporal contrast of the first-order CFWM signal is better than that of the SHG sampling pulse which is the signal of the second-order nonlinear process. Consequently, the CFWM signal seems to be the perfect sampling pulse for a cross-correlator. However, the highest pulse energy of generated first-order CFWM signal was in a range of less than ten μJ [27] in previous work, which limits its application in a high dynamic range crosscorrelator due to foreseeable low correlation signal output.
In this paper, we greatly improve the output pulse energy of first-order CFWM signals to the 100-μJ-range, which is one-order higher than previously obtained highest first-order CFWM signal. We do this in a single-stage CFWM process with a thin glass plate in the first step. It should be noted that it usually takes one stage of OPG and at least one stage of OPA to obtain 100-μJ-level wavelength-shifted amplified signals using the OPA system. A compact single-shot cross-correlator was successfully constructed with a dynamic range of about 10 10 , an improved temporal resolution of about 160 fs, and a temporal window of 50 ps, where the high-energy clean first-order CFWM signal is used as the sampling pulse and a 16-bit sCMOS camera is used as the correlation signal acquisition device.

Generation of high-energy CFWM signals
Previously, the highest pulse energy obtained in the first-order CFWM signal is less than ten μJ, which is not high enough to serve as the sampling pulse for a high dynamic range temporal contrast measurement [27]. In previous work, two incident laser beams for CFWM were focused onto a glass plate using a spherical lens or concave reflective mirrors [25][26][27][28][29]. Self-focusing and filamentation may appear around the focal point of two high energy input beams, which will affect the generated signal. To avoid these problems and increase the input pulse energy, we use cylinder lenses to focus the two incident beams into a thin glass plate. Resultantly, pulse energies of generated CFWM signals are enhanced with increasing pulse energy of the two input beams. The proposed setup for high-energy CFWM signal generation is shown in Figure 1. After passing through a short-pass dichroic mirror with a ~800 nm cutoff, the incident ultrashort laser beam is split into two beams. Both the reflected beam with wavelength above 800 nm and the transmitted beam with wavelength shorter than 800 nm are focused by two cylindrical lenses with 500 mm focal-length onto a fused silica glass wedge with thick end of 1 mm and thin edge of 0.15 mm. A time-delay stage is used in the optical path of one of the beams to tune time delay for temporal overlapping of the two beams.  As in our previous work [30], the beam quality achieved is satisfactory (data not shown in this paper). The output power can be further increased and the tunable spectral range of the first-order signal can be extended if a 25 fs Ti: sapphire laser system with higher pulse energy and broader spectrum is used. Furthermore, the power stability can further be improved if external experimental conditions such as air-flow and vibration in the lab are better controlled.
Frequency-degenerated four-wave mixing processes such as XPW and SD can be used to improve the temporal contrast of ultrashort laser pulses [30][31][32]. In the same manner, the CFWM process is a frequency non-degenerated cascaded four-wave mixing process, where the generated CFWM signals provide an improved temporal contrast in comparison to that of the input laser pulses. To verify pulse cleaning ability of the CFWM process, the temporal contrast of both the input pulse and the first-order FDC signal are measured using a commercial third-order cross-correlator (Amplitude Technologies Inc., Sequoia 800). Results are shown in Figure 3. Obviously, the firstorder FDC signal is sufficiently cleaned with a temporal contrast improvement of about 10 6 in comparison to that of the input laser pulse. It should be noted that since the Sequoia is optimized for 800 nm central wavelength, 10 10 temporal contrast already marks the measurement limit for the first-order FDC signal with 146 μJ pulse energy.
As can be deduced from above results, the first-order CFWM pulse holds both a 100-μJ-level pulse energy and high temporal contrast, which means that it can be used as the sampling pulse in a single-shot cross-correlator.

Single-shot temporal contrast measurement
The third-order correlator has been the dominant method for single-shot temporal contrast characterization in the past decades since its first application in 2001 [18]. The dynamic range, temporal resolution and temporal window are three key parameters for a single-shot temporal contrast measurement. The single-shot measurement is based on the time-to-space encoding in which the temporal intensity measurement is transformed to a spatial intensity measurement with a spatial detector as shown in the Figure 4.
Consequently, in order to improve the dynamic range of a temporal contrast measurement, a simple method is to obtain the maximal correlation signal and then introduce an attenuator to the strongest main pulse. So far, the highest single-shot dynamic range is 10 10 where a point attenuator is used to weaken the main pulse by about four orders of magnitude [21]. To improve the temporal window, a larger size of nonlinear crystal and a larger beam diameter with a relatively larger crossing-angle are typically used during the final SFM process. The best result to date is about 50 ps [21].
Several other methods have been applied to further extend the time window. For example, a pulse replicator was used to extend the temporal window to about 200 ps [20]. As for the temporal resolution of a single-shot third-order cross-correlator, there are three limitations. The first limitation comes from the group velocity mismatch (GVM) in the nonlinear crystal. The GVM can be expressed as τ = * ( where l is the length of the SFM nonlinear crystal, and vs and vp are group velocities of the sampling pulse and test pulse, respectively. Angles θ and α denote crossing-angles of the test pulse and the sampling pulse, respectively, with respect to the propagation direction of correlator signal in the nonlinear crystal. The second limitation arises from the pixel size of the cross-correlation signal detector. Finally, the third limitation is the spatial resolution of the 4f imaging system, which is used to map the correlation signal from the nonlinear crystal to the detection system. To date, temporal resolutions for single-shot cross-correlators are in the range of hundreds of femtoseconds [21]. When all three key parameters are taken into consideration, the best results of single-shot cross-correlators to date was obtained by the Qian group in 2014 with a fiber array detection system, where one OPG and one stage of OPA were used to generate the sampling pulse [21]. Figure 4 The time-to-space encoding for a single-shot cross-correlator. In this study, employing advantages from the setup of the Qian group [21], we construct an improved and simplified cross-correlator for single-shot temporal contrast measurement by using the first-order CFWM signal as the sampling pulse. Figure 5 illustrates the optical schematic setup of the novel cross-correlator. The femtosecond laser input pulse is split into two by using a beam splitter with a 20:80 reflection/transmission (R:T) ratio. The reflected signal is applied as the test pulse. The stronger transmitted signal is used as the input pulse for high-energy CFWM signal generation in a thin glass plate wedge. The generated first-order CFWM signal and the test pulse are firstly filtered by two soft-edge irises to obtain an almost top-hat intensity distribution, and then are spatially expanded by two separate beam expanders (GBE02-B, THORLAB), to be focused into a wedge-designed nonlinear BBO crystal by using two 200 mm focal-length concave cylindrical reflective mirrors with a large crossingangle. The cross-correlation signal is produced by the SFG process in the BBO crystal and imaged onto a 16-bit sCMOS camera by using a 4f lens pair. In order to achieve a high dynamic range measurement, we also attenuate the main peak of the crosscorrelation signal by using a strip-shaped density filter, which leads to about 4 orders of magnitude attenuation. Furthermore, a coated wedge is placed right behind the BBO to introduce attenuated signals for the generation of a reference replica signal below the main cross-correlation signal, whose attenuated main pulse still saturates the camera.
Moreover, to avoid scattering noise from the test and sampling pulses, a short-pass filter with a cutoff at 700 nm is placed in front of the sCMOS camera.  In the equipment of this study, a strip-shaped density filter with an attenuation of 10 000× is placed right behind the BBO crystal to weaken the strongest main signal.
Even after passing through the density filter, the main pulses are still too powerful for the detector, while the addition of another density filter would make the main pulse too weak to be detected. Therefore, we use a wedge glass plate, which can provide a replicacorrelation signal with an intensity attenuation of about 70×. The replica, located at a different region on the sCMOS chip, helps calibrate the intensity of the prepulse, postpulse and the attenuated main pulse, which still saturate the sCMOS detector. The wedge is not always necessary, as in the case where the peak signal after the stripshaped density filter does not saturate the detector. Finally, the preprocessed crosscorrelation signal is imaged onto a 2048×800 pixel area of the sCOMS and captured with an exposure time of 1 ms which is a single-shot measurement for 1 kHz repetition laser pulses. An example of the single-shot result captured by the sCMOS is shown in the Figure 7 in which Ⅰ,Ⅱ, and Ⅲ are the main pulse after passing through the stripshaped density filter and two prepulses that saturate the sCMOS detector, respectively.
Meanwhile, 1, 2 and 3 denote replicas of Ⅰ,Ⅱ, and Ⅲ provided by the coated wedge, respectively. By summing up the intensity of pixels along the vertical direction and subtracting the background noise, we obtain the temporal contrast of the test pulse within the measurement temporal window. And for a single-shot measurement, the temporal window is about 50 ps, limited by the aperture of the nonlinear crystal, the width of the input beam, and the crossing-angle. By changing the delay-time between the sampling pulse and test pulse by about 40 ps, we performed another shot of measurement. Thus, the total integrated temporal window can be extended to about 90 ps as shown in Figure   8, where both shots of measurement has a 10 ps overlap region to help joint the results.
And so on, we can obtain about 130 ps temporal window with three shots of delayshifted measurement. Actually, since only about 2/3 part of the detector is used by the correlation signal, we expect to obtain wider single-shot temporal window in the near future with larger beam diameter and wider nonlinear crystal.
The noise aroused from SHG signals of the sampling pulse and test pulse need to be analyzed. By blocking either the sampling pulse or test pulse, the intensity of noise  In more detail, we find that even though the prepulses E and F are detected by both and t = 30.14 ps, respectively, arise because of the back and forth reflection when the main pulse passes through some optical components. For the sampling pulse generated by the second-order nonlinear process, postpulses with intensity ratios of 10 -6 or 10 -5 still exist and eventually lead to the ghost prepulse with the same intensity ratio through the correlation process. In comparison, the sampling pulse in our equipment is generated by the third-order process, and its intensity can be expressed in the time domain as: where FD, 1, 2 refer to the first-order FDC signal and the two input beams for CFWM, respectively. As a result, even if the pulses have postpulses with the intensity of 10 -3 , the intensity of the postpulses of the sampling pulse would in theory be weakened to about 10 -9 magnitude. Therefore, ghost prepulses will also be at about a 10 -9 level, which is much lower than the temporal contrast of about 10 8 in this experiment, and renders them undetectable. As a result, the high energy CFWM signal is a better sampling pulse with cleaner pulse trailing edge in comparison to the SHG signal.
Finally, we study the temporal resolution of our equipment. A femtosecond pulse (about 40 fs) from a Ti: sapphire regenerative amplifier [21] is measured using our experiment setup and the Sequoia-800. The result implies about 160 fs temporal resolution with our equipment, which is the FWHM of the measured correlation trace width. The resolution of Sequoia-800 is about 250 fs, which is consistent with the result in reference [21]. The measured correlation traces in a linear plot are shown in Figure   8(b). It is obvious that our measurement curve is narrower than that that of Sequoia-800. Therefore, it can be concluded that our equipment has a better temporal resolution than that of the commercial third-order correlator Sequoia-800. In our measurement, Three pixels of the sCMOS detector are needed to capture the signal. Thus, the theoretical temporal resolution limited by the imaging system is about 120 fs, which is close to measurement results. If the correlation signal could be mapped onto the sCMOS detector with several-times magnification, the resolution of the measurement could be further improved.

CONCLUSION
As high as 100-μJ-level powers of the first-order CFWM sidebands with shifted central wavelength and high temporal contrast are generated in a thin glass plate based on the CFWM process, which usually needs at least one stage of OPG and one stage of OPA when an OPA system is used. The first-order sideband is used as the sampling pulse in a compact and economical single-shot cross-correlator for temporal contrast measurement. The proof-of-principle experiment results have demonstrated that the single-shot correlator has a dynamic range of about 10 10 , a temporal resolution of about 160 fs, and a temporal window of 50 ps. To the best of our knowledge, it is the first experiment demonstrating that a single-shot cross-correlator is comparable to the commercial delay-scanning cross-correlator in terms of both the dynamic range and the temporal resolution. Furthermore, higher dynamic ranges can be obtained by increasing the correlation signal energy and sensitivity of the detector, and higher temporal resolution can be obtained by using a magnification 4f mapping setup, and wider temporal window can also be obtained by using larger input beam diameter and wider nonlinear crystals in the future.