Two imaginary frequencies are found after dimer calculation

[ichthyornis]

I have calculate the transition state between two stable ones with dimer method. But the resulted transition state haves two imaginary frequencies. I have recalculated the transition state with a more precise parameter (EDIFFG = -0.01), but the imaginary frequency is not changed.

So, what could I do to obtain the real transition state now? Does it means the transition state is a second order saddle point?

Thanks in advance.

[graeme]

I can't tell from this information. The frequency calculations can be very sensitive to your choice of parameters. It is essential to use accurate forces (ediff=1e-8) and small finite displacements. Make sure that your frequencies do not change if you make your displacements smaller or your forces more accurate. I would only start worrying about second order saddles if you have trustworthy frequencies.

Another issue may be very soft or zero modes. If your system can translate or rotate, these low modes should be disregarded.