It is right that you need liquid sample for ICP. But you can dissolve your sample in acid, aqua regia or any other solvent in which the compounds can be dissolved. ICP will give you far more reliable result compared to EDX. In case of EDX you will get a relative percentage of Ti, Si, Al, Ca, F and O. But the problem of EDX is that it will give you a very rough estimate of the atomic counts. Many experts don't consider the EDX data as reliable results. If you are taking EDX measurement at different parts of your sample you might get different atomic ratio of the elements. Even you can alter with the values by pre-defining the compound.
Not only XRF but all the three methods mentioned here are elemental analysis techniques. So you will get relative persentage of elements present in your compound. You need to show XRD patterns to justify the presence of each components along with some elemental analysis to justify the ratio.
From my experience for samples containing SiO2 it is better to use EDX. Dissolving the sample in HCl or H2SO4 may not provide the SiO2 content adequately