Ultra-intense ultrashort lasers draw great interest in laser–matter interaction research, such as laser acceleration, laser fusion, attosecond sciences, atomic molecular physics, materials science, nuclear physics and astrophysics[1]. With remarkable progresses in chirped pulse amplification (CPA)[2] and optical parametric chirped pulse amplification (OPCPA)[3], the laser peak power has already reached the order of petawatts (PW, 1015 W)[4–6]. In order to investigate the physical laws of extreme conditions under stronger lasers, many countries are vigorously developing 10 PW-level laser systems, and have proposed development plans of 100 PW-level laser systems[7–9]. The current PW laser systems are based on Ti:sapphire, but are limited by thefact that the size of Ti:sapphire is not large enough to support a peak power of 100 PW or higher. Meanwhile, for Nd:glass based laser systems at 1 μm, it is difficult for the pulse width to reach shorter than 100 fs ; therefore, very high energy is required to achieve 100 PW laser pulses. Compared with CPA, the OPCPA laser system is the most suitable program for 100 PW lasers, as it can support high energy and a short pulse width at the same time. Because the largest size of deuterated potassium dihydrogen phosphate (DKDP) crystals can be up to over 400 mm, the DKDP-based OPCPA system is the most promising candidate for 100 PW lasers. The most powerful pump laser that can provide more than 10 kJ energy is based on Nd:glass, and the broadest gain bandwidth of DKDP pumped by a frequency doubled Nd:glass laser is centered at 910 nm; hence, the seed source for this system must also be centered around 910 nm and the bandwidth should be more than 200 nm to support an ultrashort duration. Different from typical seed lasers with center wavelengths of 800 and 1053 nm[10,11], broadband 910 nm laser pulses are rarely obtained directly from commercial lasers. Therefore, nonlinear frequency conversion should be adopted, such as optical parametric amplification (OPA) and second harmonic generation (SHG) processes. For example, a 910 nm laser pulse has been successfully achieved with the OPA process and subsequent SHG process[12]. However, for the limited bandwidth of phase-matching in OPA and SHG processes, laser pulses with shorter pulse duration are extremely difficult to obtain directly. It is necessary to further broaden the spectrum to achieve few-cycle or even cycle-level pulses.