摘要： Mode-locked (ML) fiber lasers are becoming key components in industry, because they provide high performance while being relatively inexpensive. The generation of high-repetition rate pulses is gaining interest but it remains a challenge to achieve a GHz repetition rate with fiber lasers. On the other hand, whispering-gallery-mode (WGM) microresonators allow the strong confinement of light and miniaturizing an ML laser into a WGM microresonator will enable us to achieve a high pulse repetition rate, a small footprint and on-chip integrability. In this work, we explore the possibility of a passive ML laser with the system shown in Fig. 1(a). There are two key technologies involved; one is saturable absorption (SA) and the other is laser gain in a microresonator. First, we verified numerically that we could realize an ML laser with our proposed carbon nanotube (CNT) coupled erbium (Er) doped microresonator, and the result is shown in Fig. 1(b). We modelled our system with a nonlinear Schr?dinger equation, where we took gain and SA (also loss (Q), Kerr effect, and dispersion) into account [2]. The result shows that a self-starting ML is possible with microresonator parameters. Next, encouraged by the numerical result, we experimentally demonstrated SA in a microtoroid. We grew CNTs selectively on a silica microtoroid by chemical vapor deposition and investigated the SA behavior (Fig. 1(c)). By performing a pump-probe like experiment we successfully characterized the SA behavior in a microresonator system. This is the first demonstration of SA in a WGM microcavity system [1]. We then fabricated an Er-doped active microtoroid by using the sol-gel method. The sol-gel method is one way of forming silica from a metal alkoxide precursor. Although sol-gel has been widely used, it is still a challenge to fabricate high-quality defect-free film on a Si wafer that allows us to fabricate an ultrahigh-Q WGM microresonator. Figure 1(d)-(f) shows the fabricated films. We found optimum parameters and obtained a clean film as shown in Fig. 1(g). By using this film, we fabricated an Er-doped WGM microresonator, where the Q was 1.2×106 at 1480 nm (Fig. 1(h)). We obtained clear up-conversion luminescence when we pumped the cavity at 1480 nm, which constitutes the first step towards ML lasing in such a WGM microresonator. In summary, these results provide a promising approach for realizing an on-chip high repetition rate ML WGM microresonator.