With the deepening of people's research on nanodevices, electrical noise in nanodevices is getting more and more attention. Noise performance is also an important characterization of nanoelectronic device performance, especially in the widespread application of ultra-sensitive detection and sensing in nanoelectronic devices. The noise in the device directly determines the highest resolution of the detector.

Electrical noise occurs with the advent of electronic devices, which affects device performance, interferes with device operation, and distorts useful signals. However, noise does not only cause problems for people. With the further development of science and technology, people begin to use noise to measure because noise essentially reflects the transport mechanism inside the mesoscopic system.

Using shot noise, people successfully observed the electronic copper pairs in superconductors and fractional charges in two-dimensional electron gas systems under strong magnetic fields.

2 Several noise sources in nanometer devices 2.1 Thermal noise Thermal fluctuations of carriers are inevitable sources of noise at non-absolute zero temperatures. Thermal excitation causes a systematic energy state to fluctuate by the number of occupations, thereby generating thermal noise. Even if no voltage bias is applied, thermal noise is present, so it is called balanced intrinsic noise and is the most typical white noise. The thermal noise in the conductor is relatively simple. Its energy spectrum is given by the Nyquist formula. It is the Boltzmann constant. "The resistance of the conductor is the absolute temperature.

2.2 Shot noise Shot noise originates from the particle characteristics of carrier transport and hence its name. In 1918, Walter Schottky discovered shot noise in vacuum tubes and developed his theory. The shot noise in the conductor is the result of charge quantification. Unlike thermal noise, shot noise must be observed in an unbalanced system, such as a bias voltage, and is therefore called unbalanced eigennoise.

To explain the origin of shot noise, we consider a hypothetical one. At that time, there were source-drain voltages in which V+s was SET, Rs=Ri+R2, Coulomb blockade parameter=2%-"-, and its range was 0. Since the carrier in the superconductor is an electronic pair of copper, its charge is twice that of the basic charge. Therefore, due to the coherent action of coherent tunneling and Coulomb interaction of the electron pair in the superconducting single electron transistor, the shot noise is generally high. In normal conductors and semiconductor single-electron transistors 7.3.2 Carbon nanotubes Carbon nanotubes as a self-assembled nanomaterials with excellent electrical properties are mainly used as field-effect transistors and single-electron transistors in nanoelectronic devices.

In metal films and nanowires, there are two important sources of external noise: charge motion and defect diffusion.

In carbon nanotubes, since CC bond binding energy is very strong, it is much less affected by charge motion and defect diffusion. Therefore, in theory, the 1// noise in carbon nanotubes should be much smaller than that of metal nanometers. Line*, 19. But in fact the 1//noise in single-walled carbon nanotubes that people measure is very high, in the general conductor, in the formula*) "=1-15; And in the single-walled carbon nanotube, 119 Measured = 10u, which is 4 orders of magnitude higher than the former.

The main noise sources of carbon nanotubes are impurities or defects on the substrate material, fluctuations in the surface of the carbon nanotubes, and contamination of other adsorbent substances. To reduce 1//noise, floating carbon nanotubes are often used in device fabrication for two reasons. First, suspended carbon nanotubes do not grow directly in situ, but require the use of atomic force microscopy (1FM). Moving it to a pre-made metal electrode, the impurities on the nanotubes will decrease during the operation of the AFM; another reason is the most important, because the nanotubes are far from the substrate, so the impurities or defects on the substrate The effect of a charge trap center that can cause 1//noise on the nanotubes can be ignored. As shown, its substrate charge traps are far away from carbon.

The bound state in the potential well rises, so that it overlaps with the conduction bands of the semiconductors on both sides, the electrons will more easily enter the potential well, the potential well level rises further, and the conduction band and the middle of the semiconductors on both sides The bound state within the potential well overlaps even more. This leads to positive feedback, which increases the shot noise more than the Poisson value of 2e/. In recent years, people often use quantum dots to study strongly correlated systems, the so-called Kondo effect. . In quantum dot systems, shot noise in quantum dots is very sensitive to the external field. %027. When a microwave field, magnetic field, or gate voltage is added to a quantum dot system, shot noise is reduced or increased. Furthermore, the zero-frequency shot noise of a quantum dot is a non-monotonic function of the bias voltage, and its value reaches a maximum near the Kondo temperature. Therefore, the use of shot noise can also be used to detect the quantum temperature Kondo%. 4 noise measurement system due to the nano-device size is small, the operating voltage is low, the signal is weak, the noise is relatively weak, so to measure the noise, you must first The noise signal is amplified. During the measurement, the noise of the preamplifier itself (especially the noise of the first stage amplifier) ​​of the measurement system will also be added to the final output. We synthesize the noise measurement system preamplifier circuit into an ideal noiseless amplifier network, a voltage noise source, and a current noise source in parallel, as shown.

Ideal Noiseless Network Noise Test Network Equivalent Circuit Schematic The noise measurement system is generally composed of a preamplifier section and an analyzer section. The more intelligent measurement system also includes an automated control and data collection and processing system. Analyzers generally use spectrum analyzers and dynamic analyzers. There are two common noise measurement methods. One is the measurement method at both ends, and the other is the measurement method at the four terminals. 4.1 Two-end noise test method is a typical two-end noise test system. In order to further amplify the signal, we use two-stage amplification and then access the analyzer. Obviously, the noise of the measurement system itself, especially the noise of the amplifier, has a great influence on the measured sample noise. The noise of the first-stage amplifier has the greatest effect on the measured noise. Therefore, in the measurement process, the noise of the measurement system itself should be measured first, and then the noise can be subtracted from the sample to obtain more accurate results. In order to understand this system more intuitively, we equivalent the measurement system noise into a current noise source and a voltage noise source! *. For the noise source of the sample to be measured, s is the impedance of the sample to be measured, and the two stages of amplification are equivalent to one amplifier. The input impedance of the measurement system is in, so that the input noise of the entire system can be expressed by Equation 14).

Among them, rIV is a normalized cross-correlation function of sum. In order to calculate A/% from it, we need to know five parameters, namely: A/2, ""%, and two. So for any such noise test system, these five parameters must be found before being used for measurement.

First, use a large resistor % in series with a voltage source as the sample to be measured, and make "s" A"n and A satisfied!

Then, using 4 different resistors as the sample to be measured, the input resistance Zin of the quantity system is required to properly select these 4 resistor values ​​in order to simplify the operation. For example, when the % is infinite, Formula 16) simplifies to A. "2=2A!2, so it is possible to measure A! with an open circuit, and the other 3 resistors should be relatively low, and the difference between them should be large to facilitate calculation. To make the measurement more accurate, it should be more Measure the average and obtain the system's A", A! And Zin. The last thing to be sure is A! The normalized cross-correlation function rIV of A and A" generally uses an ideal capacitor to determine this correlation term. Since the ideal capacitor does not generate noise, this equation 14) can be written to determine the parallel noise voltage source AF, the series current noise source A! and the input resistance of the measurement system in.