Antenna array can use interference to drastically phase shift their frequency, but at a substantial efficiency loss. For radar applications, the efficiency loss is unacceptable, but for IFF interrogation, it's still workable.
Moreover, the J-10B featured both a suspected PESA and AESA array. One of the primary differences between them was that the AESA array lacked IFF dipoles. If, say, you look at other aircraft, the IFF dipoles are often located elsewhere, but if you have the aircraft completely the same, either, one, the J-10B with AESA has no IFF capability, or two, the J-10B with AESA uses its AESA to generate IFF capability.
No. IFF is implemented as a separate array, or an entirely separate system. It is
not I repeat not a radar. It is a communication system.
The reason why IFF is implemented elsewhere is guess what, its because it is really a totally separate system. You can implement it anywhere as long as it faces the same direction as the radar.
The reason why you don't see the IFF on the J-16's and J-20's radar is because the IFF is implemented outside of the radar. The most direct implication is that a new IFF system was introduced between the first J-10B and the J-16. The J-10B without the IFF across its face, probably because a newer IFF system has been introduced to replace the old system.
There are PESA, like the later iteration of BARS, that does not have IFF dipoles across their face, so IFF dipoles have no connection to being a PESA either. The reason is that later, you got a new IFF system that can be implemented outside of the radar.
You can have mechanical parabolics with IFF dipoles across their face too, such as on the original F-15's radar. You also have slotted planars too, with and without the IFF dipoles. Again simple explanation, a newer IFF system that is implemented outside of the radar replacing the older one that has dipoles strewn across the array face.
I do not see the point of dramatically phase shifting down from X to L-band, and this will result in massive mutual coupling. This is a sort of interference where the wave from one emitter affects another emitter. In the case of an X to L-band shift, the expanded wave would affect a whole series of emitters, given the physical size of the wave (X and Y axis equal to the wavelength against the physical spacing of the elements). I do not think its acceptable because the whole array is interfering with each other. The reason why each element is spaced by half to slightly more of the wavelength is avoid mutual coupling in the first place.
It would be simpler and far more efficient to implement a separate L-band communication system to serve as the IFF.