If I understand correctly, your question concerns the origin of harmonics in amplifiers. To be precise, even if the power at the input is limited to (virtually) one frequency, it can be divided at the output to several frequencies, where the lowest frequency equals the input frequenc, and the other frequencies are integer multiples of the base frequency.

The cause is non-linearity. If the amplifier would have a constant (independent of UE) gain a, so that UA(t) = a * UE(t), it would not generate harmonics. If the gain a depends on the amount |UE|, the output signal were still symmetric, i. e. UA(w * t + pi) = -UA(w * t), with the circular frequency w = 2 * pi * f, where f is the input frequency; in this case UA would contain exclusively odd harmonics. If the output signal is asymmetrical, it contains even harmonics, too.

While good amplifiers produce virtually no harmonics when used within the intended area of operation, even harmonics are usually caused by applying an input signal large enough to provide a "clipped" output signal. Since the clipping is usually not exactly asymmetric, the resulting output signal is asymmetrical, too.

Dear Joerg Fricke, thanks for your answer. It is really helpful. I have made a transistor with graphene I have applied some ac signal at its gate with certain frequency. When I am analyzing the signal of source-to-drain, I am getting two signals at 2nd Harmonics area apart by 120 Hz. Is it due to the power signal noise that makes 2nd harmonics signal into two parts... 

Dear Nowzesh Hasan, so you have for example base frequency = 100 Mhz, "harmonic" frequencies = 199.999940 MHz and 200.000060 MHz? Or is one frequency exactly the 2nd harmonics and the other 120 Hz below or above?

What happens to the difference of 120 Hz when you use a different base frequency?

Do you know the length of the "time window" which defines the part of the signal the FFT is applied to? If so, can you change it? If its length were 8.333.. ms = 1 / 120 Hz, the separation could be just an artefact because the distance between the frequency points in the output of the FFT were 120 Hz; so, if the actual frequency is located between two frequency points, its power would appear distributed to these points.

Hi Nowzesh Hasan,

in addition to the non-linearity of the amplifier, maybe you should also consider a kind of "amplitude modulation" due to 60Hz from power supply or from other noise source.

Given that: sin(w1t) sin(w2t) ) = 1/2*cos((w1-w2) t)  - 1/2*cos((w1 + w2)t) , this could expalain why you see 2 frequencies (w1-w2) and (w1+w2) apart by 2 * w2.

Best regards.