You focus your beam to the smallest diameter possible (= crossover possition) with the intensity knob and then you enter the diffraction mode. Now you should see the Kikuchi pattern.

Dear Zefeng Yu,

if you would like to observe Kikuchi lines, then you have to set on standard selected area or convergent beam electron diffraction pattern. The problem is that Kikuchi lines are not always visible. Kikuchi lines are produced mostly by incoherently and inelastically scattered electrons. At smaller sample thickness the Kikuchi lines are not present because dynamical scattering, which generate Kikuchi lines, is insufficient. More precisely, Kikuchi lines are produced but have extremely low intensity which makes them invisible. As the sample thickness increase (and all three Laue conditions are satisfied) magnitude of dynamical scattering increase as well and Kikuchi lines becomes visible. In very thick crystals the absorption effect dominate and make Kikuchi lines invisible again. So, for observation of Kikuchi lines you need to turn your TEM to diffraction mode and have the sample with appropriate thickness. The best sample thickness is approximately around middle of transparency thickness. You can influence this thickness by accelerating voltage or orientation of crystal along zone axis.

In case you would like to observe Kikuchi bands, which creates a pattern known as Kikuchi map, then use the procedure mentioned by Michaela. Kikuchi bands are not so sensitive to sample thickness as Kikuchi lines.

Which type of sample are you using? In metals it is quite easy to use the methods described above, if you have nanoparticles most likely you do not have enough inelastic scatternig to see the lines on the inelastic bckground of the pattern. Try to use a large condenser aperture above 50mu to boost the intensity and first look at the screen before you put it on a camera. A lot of cameras are not good for diffraction and the limited dynamic range prevents to see the kikuchi lines next to the diffractino spots.