LNA noise matching (physical insight)

Any one can share insight on the following?

I am looking for physical insight or explanation on why noise figure is lower when the LNA transistor device sees the optimum noise source impedance?

Thank you!

Because NF of the transistors are measured with 50Ohm termination that is quite away from Zs(opt).Therefore NFmin appears lower than NF@50Ohm.

Maybe I did not make my question clear enough...I am still wondering how different impedances affect the noise power sources in the devices and LNA's noise figure.

It is easeir to understand this at low freq.
Circuits may exibit more noise current or more noise voltage. Thus input noise depends on the resistance of the source. For low source resistance the noise current will have little effect but noise voltage will be dominant. For high source resistance the noise current will dominate. For example, FET may have small noise with high source resistance because they have near zero DC input current.

So there is some optimum source resistance for which input noise will be minimum. For RF freq it is more complicated because of complex impedance character.

Alex, thanks for sharing an interesting way of looking at this.