There are spikes plotted in the datasheets phase noise plot (Fig.4, p10).
The amplitude is not acceptable for some regions spectrum masks.What is the source of those spikes and how can they be avoided?
The spikes are spurs as a result of the PFD (phase frequency detector) frequency of the PLL, in this case 1MHz. In a Integer-N PLL spurs will appear at the PFD frequency. They can be reduced by reducing the loop bandwidth.
The ratio of loop bandwidth (7k5) to PFD (1MHz) seems rather extreme already.
What loop filter topology was used for the measurement?
The data sheet plot is taken with ~7.5kHz loop bandwidth
Retrieving data ...