hello all !!!

I am reading Filter Basics: Anti-Aliasing, a tutorial by Maxim.
I am confused in the section of undersampling.
I am attaching the image of the pdf of the section where I am confused.

The tutorial says "Undersampling at 4MHz yields first-order sum and difference terms (f1 + f2 and f1 - f2) of 14MHz and 6Mz, and second-order terms (2f1, 2f2, 2f1 + f2, f1 + 2f2, | 2f1 - f2 |, | f1 - 2f2) | of 8MHz, 20MHz, 18MHz, 2MHz, 24MHz, and 16MHz. "

I am OK with the "first-order sum and difference terms".

I am confused as to how to justify the second order terms? where do they come from? any formula/expresssion? any book or website is welcome.