Why does IM2 contribute to the noise floor at LNA input ???

Of course, for FDD(Frequency division duplex) direct-onversion receivers,
The IM2(2 order intermodulation) contaminates the desired Rx signal due to Tx leakage, as shown below :





Nevertheless, that is because the IM2 is near the down-converted baseband Rx signal in frequency domain.


But~!!
Some documents indicate that IM2 will contribute to the noise floor at LNA input.
I am confused.



1.




At LNA input :

Rx signal is at RF frequency
IM2 is at baseband frequency
Is it possible for IM2 to contribute to the noise floor of Rx signal ????
I don’t think it should be possible.
I think the IM2 only contributes to the noise floor of down-converted baseband signal.



2.




IM2 is due to nonlinearity. That is to say, even though the IIP2 of LNA is very poor, at LNA input, there should NOT be IM2 made up of Tx and RX.
(I leave blocker/jammer considerations aside)
In that case, why does IM2 appear at LNA input ?



3.




At LNA output, there is IM2. Nevertheless, IM2 is at baseband frequency.
If we put a DC block between LNA output and mixer input,
the Rx signal is NOT impaired, but IM2 is blocked.
In that case, why do we have to consider the IIP2 of LNA ??
The strong IM2 due to poor IIP2 will be blocked by DC block, isn’t it ?



Thanks a lot~!!

The strong IM2 may cause LNA to saturate, so as to degrade the RX LNA signal

But I wonder why IM2 appears at LNA input.

IM2 made up of Tx and RX is due to LNA's and mixer's nonlinearity.
(I leave blocker/jammer considerations aside)
In that case, why does IM2 appear at LNA input ?


Besides, if we put a DC-block at LNA input,
the IM2 near baseband frequency will NOT saturate LNA theoretically.


Are my assumptions incorrect ?

Thanks a lot~!!