For high performance clock, optical lattice is introduced to generate periodic trap for capturing neutral atoms through weak interactions. However, the strong trapping potential can bring a large AC Stark frequency shift due to imbalance shifts for the upper and lower energy levels of the clock transition. Fortunately, it is possible to find a specific “magic” wavelength for the lattice light, at which the first-order net AC Stark shift equals zero. To achieve high stability and accuracy of a neutral atomic optical clock, the frequency of the lattice laser must be stabilized and controlled to a certain level around magic wavelength to reduce this shift.#br#In this paper, we report that the frequency of lattice laser is stabilized and linewidth is controlled below 1 MHz with transfer cavity scheme for ytterbium (Yb) clock. A confocal invar transfer cavity mounted in an aluminum chamber is locked through the Pound-Drever-Hall (PDH) method to a 780 nm diode laser stabilized with modulation transfer spectroscopy of rubidium D2 transition. It is then used as the locking reference for the lattice laser. This cavity has a free spectral range of 375 MHz, as well as fineness of 236 at 780 nm, and 341 at 759 nm. Because neither of the wavelengths of 759 nm and 780 nm is separated enough to use optical filter, they are coupled into the cavity with transmission and reflection way respectively, and the two PDH modulation frequencies are chosen differently to avoid possible interference.#br#The stabilization of the 759 nm lattice laser on transfer cavity can stay on for over 12 hours without escaping or mode hopping. To estimate the locking performance of the system, a beat note with a hydrogen maser-locked optical frequency comb is recorded through a frequency counter at 10 ms gate time for over 3 hours. This beat note shows that the frequency fluctuation is in a range of 10 kHz corresponding to a stability of 2×10-11 level with 0.1 s averaging time, but goes up to 150 kHz corresponding to a stability of 3.6×10-10 at 164 s averaging time. The long-term drift can be the result of air pressure fluctuation on the transfer cavity, or the bad stability of the optical comb in the measurement process. However, current locking performance is still enough for the requirement of 10-17 clock uncertainty.#br#In conclusion, we succeed in realizing frequency stabilization and control for the lattice laser of Yb clock with the transfer cavity scheme. The result shows that the short-term stability is around 10-11 level, though a mid-long-term drift exists. However, the stability of 3.6×10-10 over 164 s can still promise a 10-17 uncertainty for the Yb clock. And, it can be reduced if the averaging time is long enough. The work can be further improved by installing the transfer cavity into vacuum housing for better stability in future.