During the two decades since the first production of isolated attosecond pulses (IAPs) in 2001 [1], IAP sources have become highly desired to open up the broad research area of ultrafast science [2–4]. In particular, the high-power IAP sources that support attosecond pump/attosecond probe experiments are powerful tools to explore ultrafast phenomena in the attosecond (as) region [5–7] that cannot be accessed by a conventional fs IR pump/as XUV probe scheme. One promising method to obtain a high-power, gigawatt-level IAP is an X-ray free-electron laser (XFEL) [8–11] that uses the accelerator technology in a large facility. However, the stability of ultrashort pulses from XFEL sources is still limited—even in a very recent report [12]—because of its amplification principle: self-amplified spontaneous emission [13]. Instabilities in the pulse duration, pulse energy, and spectrum shape also reduce the potential applications of XFEL sources to ultrafast dynamics research. Further, the accessibility of XFEL facilities is very limited because of their huge cost and space limitations. The development of laser-based, tabletop, high-power IAP sources is thus essential to eliminate the limitations of XFEL sources [12].

Generally, IAPs generated using laser systems have low output power because of the limited power of the laser system and the low conversion efficiency of high-order harmonic generation (HHG) [14,15]. Compared to XFELs, however, laser-based IAP sources have the advantages of temporal coherence, stability, and the ability to generate shorter pulse durations [16]. Thanks to their relatively low equipment cost, laser-based IAP sources are thus more suitable for laboratory-scale applications, if their low-power limitations can be overcome. With the recent development of a high-power, low-repetition-rate, near-IR laser system [17], and an enhancement effect found in HHG driven by waveform synthesis [18], the generation of intense IAPs through HHG [19] is expected to be possible. Previously, we reported using the HHG process to generate a high-power IAP using a high-energy driving-laser source with a multi-terawatt (TW), three-channel, optical waveform synthesizer at a 10 Hz repetition rate [20]. As an illustration of the results, we demonstrated the use of a custom-tailored optical waveform, obtained using parametric waveform synthesis, to generate an intense continuum spectrum covering the range 50–70 eV using HHG. By tailoring the synthesized waveform, we were able to obtain both an enhancement of, and a blueshift/redshift in, the generated continuum spectrum. In addition, the bandwidth of the cutoff continuum with sub-microjoule energy supports transform-limited (TL) pulse durations of sub-200 as [20]. However, for reasons we will explain later, it is not easy to precisely evaluate the pulse duration of the low-repetition-rate, high-energy supercontinuum.

Streaking methods [3] are most frequently used to characterize the temporal shape or duration of IAPs, with the frequency-resolved optical gating for complete reconstruction of as bursts (FROG-CRAB) method [21,22] used to retrieve the spectral intensity and phase information from the streaking results. However, the characterization of the IAP duration in an experiment with a low-repetition-rate laser source suffers the huge barrier of an extremely long data acquisition time. One reason for this is the low photon flux in the generated IAP, and a digital converter usually has to be used for photoelectron counting to acquire the photoelectron spectrum. When the trigger frequency is low, the data collection time increases in inverse proportion to that frequency. Furthermore, when the data acquisition time is longer, the stability of the generated IAP becomes more critical, which creates another barrier for IAP applications. To our knowledge, there has been no previous report of a successful streaking experiment at less than a 1 kHz repetition rate. Another reason is that there is no realization of IAP generation by a low-repetition-rate, multicycle laser. Consequently, streaking schemes have not yet been investigated to evaluate low-repetition-rate, high-energy IAPs.

To experimentally characterize the IAP duration from the generated supercontinuum, we achieved a groundbreaking as streaking measurement at a 10 Hz repetition rate. After retrieval, we found the shortest pulse duration of the IAPs generated by the synthesizer was 226 as, which confirms that the peak power of the generated IAPs exceeded 1 gigawatt (GW). To our knowledge, this is the first demonstration of a GW-scale IAP output in a tabletop environment.

For a high-energy, low-repetition-rate waveform synthesizer [20] that produces the three component pulses from the multicycle optical parametric amplifier outputs—the pump (800 nm, 20.3 mJ, 30 fs), the signal (1350 nm, 4.3 mJ, 44 fs), and the idler (2050 nm, 1.6 mJ, 86 fs)—the first important issue in applying the as streaking method is to remove the intense driving pulse after the HHG because the energy of this pulse is high enough to easily damage optical components such as metal filters and multilayer mirrors. In this work, we therefore designed a polarization-controlled power scheme to reduce the remaining energy of the driving pulses and purify the IAP from the synthesizer pulse. The detailed scheme is shown in Fig. 1. The gating pulse (800 nm, 30 fs, CEP-stabilized [23]) for the streaking is separated from the linearly polarized (p-polarized) pump pulse by a 45 deg, 2-inch (50.8 mm) mirror with a 20 mm diameter center hole. The polarization of the gating pulse is later converted to s-polarization through a half-wave plate (HWP). After passing through the delay-control devices, the gating pulse is recombined with the synthesizer output by another center-holed mirror. For this two-beam-path configuration, a Mach–Zehnder interferometer is easily constructed by introducing another He–Ne laser beam passing through the same path. By monitoring the interference fringes, we obtain the feedback signal that we apply to the delay-control device (piezo stage) to suppress the delay jitter between the pump and the gating pulse to less than 56 as in rms. The details are shown in Supplement 1. Then, by spatially varying the polarization using the periscope, we finally adjust the output of the waveform synthesizer to s-polarization, while the gating pulse remains ${{p}}$-polarized before the HHG gas cell.

 

Fig. 1. Scheme of the system for as streaking measurements. BS, beam splitter; HWP, half-wave plate; DM, dichroic mirror; HHG, gas cell for HHG; CV, convex lens with center hole; CC, concave lens with center hole; and eTOF, electron time-of-flight device.

Download Full Size | PDF

The pump pulse, signal pulse with idler, and gating pulse are focused separately using three lenses (focal lengths: 4.5 m, 3.5 m, and 4 m), while they pass freely through the entrance and exit holes of the HHG gas cell. Argon (Ar) gas is introduced hydrostatically into the 10 cm long interaction cell. Two beam splitters (BSs) are introduced 2.5 m away from the gas cell as the driving-pulse power limiter. The BSs are made of Si substrates with NbN (niobium nitride) coatings [24]. After two Brewster-angle reflections (74.8° at 800 nm) from the BSs, the ${{s}}$-polarized synthesizer output pulses are drastically reduced in power from tens of millijoules to several microjoules, while the HHG pulse and the ${{p}}$-polarized gating pulse are effectively reflected by the BSs. After the BSs, we used a spatial filter with two holes (2 mm and 1.5 mm in diameter) to transmit the IAP and shape the gating pulse, respectively. To optimize the focusing position of the gating pulse to overlap with the IAP at the end, a convex–concave pair of lenses with center holes is introduced after the spatial filter, with the IAP freely passing through the center holes of the lenses. We used a circular aluminum (Al) foil mounted on a mesh to remove the remaining energy of the synthesizer pulses completely from the HHG pulse, while the gating pulse can easily pass through the mesh (86% transmission) outside the Al foil. Another HWP with a center hole is also introduced between the convex–concave pair of lenses to change the gating pulse to ${{s}}$-polarization, which is the same as the IAP polarization. Both the IAP and the gating pulse are finally focused on and interact with a 0.13 mm thick neon gas jet at the entrance to the electron time-of-flight (eTOF) device. They are both focused by a multilayer Mo/Si concave mirror with a 300 mm focal length with a cross-angle of 0.003 radians. This mirror is specially designed to have a high reflection rate (13%) around the 60-eV region to extract the continuum cutoff region from the harmonic spectrum and to have a 52% reflection rate for the gating pulse. The details can be checked in Fig. 2(d). The estimated continuum harmonic energy is approximately 13 nJ at the interaction point in the eTOF device.

 

Fig. 2. (a) Experimental as streaking trace and (b) trace retrieved using FROG-CRAB with the PCGP algorithm; the retrieved FROG error is 3.5%. (c) Retrieved results for the pulse duration and phase of the IAP. Inset shows a comparison between the experimentally measured photoelectron spectrum and the FT spectrum from the retrieved pulse. (d) The eTOF-measured photoelectron spectrum of the generated supercontinuum is shown as the solid red curve; the supercontinuum measured directly by the spectrometer is shown as the blue-dashed curve; and the reflectivity of the Mo/Si multilayer concave mirror is shown as the black-dashed curve.

Download Full Size | PDF

Thanks to the high-energy continuum harmonics generated, for a sufficient flux of ionized electrons, the TOF signal can be observed directly with an oscilloscope instead of requiring a digital converter in counting mode. By getting rid of the digital converters, the data acquisition time is greatly reduced, even at the 10 Hz repetition rate. A typical photoelectron spectrum measured with an Al foil is shown as the solid red curve in Fig. 2(d). We obtained this photoelectron spectrum by averaging 100 shots for 10 s. The labeled photon energy corresponds to the eTOF measured photoelectron energy plus the ionization energy of neon with 21.56 eV.

We performed the as streaking measurements by changing the gating pulse delay using the piezo-controlled stage. The measured streaking trace is shown in Fig. 2(a). The delay step of the gating pulse is set as 100 as. For each delay time of the gating pulse, 200 shots of photoelectron spectra measured with the eTOF are collected and averaged by the oscilloscope. This streaking trace exhibits a clear modulation with a roughly 2.7 fs time period. This period corresponds to one cycle of the gating pulse, which has a center wavelength of 800 nm. This modulation lasts through the entire 20-fs scan range. It is related to the multicycle pulse duration (30 fs) of the gating pulse, and it also can be seen in the retrieved gating vector. We programmed homemade calculation software to retrieve the input pulses using the FROG-CRAB method with the principal component generalized projections (PCGP) algorithm [25,26]. The retrieved trace shown in Fig. 2(b) has a mean square error of 3.5%. From the retrieved results for the IAP shown in Fig. 3(c), we obtain an IAP duration of 226 as FWHM. By comparing the measured photoelectron spectrum and the Fourier transform spectrum obtained from the retrieved IAP profile, we find that the two spectra overlap quite well, as shown in the inset in Fig. 2(c). Thus, we conclude that this retrieved result is reliable. Accordingly, this experimentally confirms the generation of GW-scale IAPs (pulse energy of 0.24 µJ, measured with the previously reported procedure [19]) using a fully stabilized, three-channel optical waveform synthesizer [20]. As far as we know, this result is the first demonstration of an as streaking experiment at a 10 Hz repetition rate, and it provides conclusive evidence for the realization of a GW-scale, soft X-ray IAP source on a tabletop.

 

Fig. 3. (a) Simulated HHG spectrum ${-}{{1}}$ and ${-}{{2}}$, which correspond to a change of 0.8 fs in the delay of the idler pulse in the synthesizer. The inset in (a) shows the Fourier transforms of the pulses for spectrum ${-}{{1}}$ and ${-}{{2}}$. (b) Photoelectron spectrum of the IAP measured when the delay of idler pulse is changed by ${+}{1.7}\;{\rm{fs}}$ (from delay point A to B). (c) Retrieved IAP profiles with the idler pulse at delay point A (the 1st and 3rd scans) and (d) corresponding profiles at point B (the 2nd and 4th scans).

Download Full Size | PDF

As described previously, the cutoff of the generated spectrum can be redshifted/blueshifted by changing the synthesized waveform used to drive the HHG process. Figure 3(a) shows simulations of the shifted HHG spectrum when the delay time of the idler pulse in the synthesizer is changed by 0.8 fs [20]. The TL pulse duration is correspondingly changed from 236 as to 267 as, as shown in the inset in Fig. 3(a). During this TL pulse estimation, we used the same Hanning window to filter the cutoff region of the spectrum. Depending on the synthesized waveform conditions, the bandwidth of the measured photoelectron spectrum varies due to the combined effect of cutoff spectrum shifting and window filtering; in the experiment, this is done with the multilayer concave mirror and the Al filter.

In the experiment, when we changed the delay of the idler pulse in the synthesizer from an initial position A to a position B with a ${+}{1.7}\;{\rm{fs}}$ delay, the photoelectron spectrum measured by the eTOF device shifted from the red to the blue curve shown in Fig. 3(b). We applied four rounds of as streaking sequentially with the delay positions located in the order A–B–A–B for the idler pulse of the synthesizer. From these attosecond streaking results, we find that the retrieved IAP durations obtained in the 1st and 3rd scans, with the idler pulse delay at position A, are around 270 as, as shown in Fig. 3(c). Correspondingly, for the 2nd and 4th scans, with the idler delay at position B, the retrieved IAP durations are consistently 240 as, as shown in Fig. 3(d). These results provide strong evidence that stable and intense IAP generation has been achieved by the waveform synthesizer at a 10 Hz repetition rate. The duration of the output pulse can thus be tuned by tailoring the waveform synthesizer. In the current experiment, the shortest measured pulse duration (226 as) was limited by the bandwidth of the multilayer mirror, which is centered at 65 eV. It may be possible to overcome this bandwidth limitation by replacing this multilayer mirror with a grazing incidence toroidal mirror. The atto-chirp of the measured IAP is very small [27] because only the cutoff region of the spectrum is extracted by the multilayer mirror. This is also confirmed by the phase information retrieved for every streaking measurement.

We have demonstrated as streaking measurements to characterize a low-repetition-rate, intense IAP source. We experimentally confirmed the GW-scale IAP output (10 Hz, 0.24 µJ, 226 as) using the HHG from a waveform-tailored, high-power, multichannel synthesizer. The results prove that intense IAPs were generated, and the excellent stability of the synthesizer makes it possible to characterize the pulse duration by using the as streaking method at a 10 Hz repetition rate. The streaking experiment also proves that tunability of the pulse durations of the IAPs generated by the synthesizer is achievable. Our streaking system also can be applied to measure the sub-µJ low-repetition-rate continuum HHG above 100 eV [28] by employing appropriate optics. The tabletop-based, intense IAP source developed in this work constitutes a breakthrough in the power and repetition-rate limitations of IAPs in the HHG route. As complementary partners with the free-electron laser route sources, their synergy will pave the way for future applications [29–31] in attosecond science.