Search

The noise term involved in the impedence field method, for modelling the noise of devices, which may work under both space‐charge and hot carrier conditions, is shown to be proportional to the local noise temperature and to the local ac conductivity. This allows one to determine the noise source term experimentally, which is performed, as an example, for n‐type silicon at 77 K. Also are established the mathematical conditions, under which the impedance field method reduces to the salami method.

Taking into account the finite lenght of the gate of a JG FET and the variation of the mobility versus electric field, it is shown, through a self-consistent hypothesis, that the electrical and geometrical properties of the active conducting channel remain unchanged in the saturation range. Experimental results of noise measurements performed in the saturation range at 300 K and 77 K can easily be explained by this theory.

The differential relaxation time τ (k,F) defined in a previous paper for isotropic semiconductors in the hot‐carrier range is shown to depend on the orientation with respect to the field force F. τ (k,F) is expressed using the collision operator and the distribution function. In the Ohmic range τ (k,F=0) is found to be equal to the usual relaxation time τ (k) related to transition probabilities per unit time. Longitudinal τN(k,E) and transverse τn(k,E) differential relaxation times are numerically computed for p‐type germanium and are found to be actually different although of the same order of magnitude, namely 10−12–10−13 sec. It is proved that the distribution function for hot carriers may be a displaced Maxwellian only if τ (k) is k independent, which does not hold for most of the situations of physical interest. It is shown that the differential relaxation times involved in longitudinal D∥(E) and transverse D⊥(E) diffusion coefficients are τ∥(k,E) and τ⊥(k,E). D∥(E) and D⊥(E) are numerically computed...

The stationary Boltzmann equation is solved by using a matrix method. This allows determining directly the steady state regime for hot carriers in semiconductors. The main advantages of this new method are: the result does not depend on initial conditions, and the cost is much lower than that of iterative or Monte Carlo techniques, particularly at high electric field.

Noise measurements at 77 K performed on JG FET 2 N 4416 (Texas Instruments Corporation) show an important disagreement with theoretical expressions given by Van der Ziel. By analogy with results obtained with p-type germanium samples in the presence of a moderately high electric field, the “warm-carrier” hypothesis is made for the FET channel. Experimental results in the range 77 K–300 K fit the theoretical laws deduced from this assertion.

Transverse ( T n⊥ ) and longitudinal ( T n| ) diffusion-noise temperatures are compared in isotropic semiconductors in the presence of an applied electric field E . It is shown that T n| > T n⊥ and that the ratio T n| / T n⊥ tends towards infinity at intense electric fields. Differential mobility is mainly responsible for the strong anisotropy induced by an electric field in the noise temperature. The study of such phenomena in anisotropic semiconductors requires a knowledge of the variation of the mobility tensor with the vector E .

Properties of the mobility tensor and differential mobility-tensor components in the hot-carrier range are investigated in cubic semiconductors as regards to symmetry and independently from any scattering mechanism. It is shown that all the components can be deduced from only one diagonal and one nondiagonal component. General relations between components are established. Considering conduction of each valley brings additional information which is explicated for semiconductors with valleys lying along the

It is proved that the three diffusion coefficients, related to noise, spreading of a bunch of carriers, and gradient of concentration, are identical. This demonstrates in the hot carrier regime the validity of the relation kTn(E)/q=D (E)(dvd(E)/dE)−1 which is an extension of the Ohmic Einstein relation (Tn is the noise temperature, D is the diffusion coefficient, vd is the drift velocity). Therefore the noise and time‐of‐flight techniques appear as being two ways which are quite different, leading to the same diffusion coefficient.

Measurements of the transverse differential conductivity of hot carriers in semiconductors are performed, for the first time, using an experimental setup described in the present paper. This allows the determination of the transverse diffusion coefficients.

Two numerical methods for modeling the noise of one dimensional devices are presented. We briefly recall the technique very recently developped, to be used when the electrical variable of interest is the electric field, and we extend it to the case where the electrical variable of interest is the local potential. Numerical results are in quite good agreement with analytical ones for sclc diodes.

A numerical method for modeling the noise of one dimensional devices is presented, and is checked on a single space charge limited current diode, where analytical results are available.

Noise and conductivity measurements in p-Si at 77K, up to 2.5kV/cm electric field strength, shows that three noise sources are significant: diffusion, G. R.,and 1/f noise. The diffusion coefficient versus electric field is given. The 1/f noise is important up to 10 GHz. The G.R. noise cannot be explained assuming two energy levels (impurity level and the valence band).

A theoretical investigation and new experiments on the conductivity of lightly doped p‐Si (boron) at 77 K are presented. The conductivity is studied as a function of the electric field in the range 10

The theory of the fluctuations of the state occupancy function, recently developped, together with the differential relaxation time, allow getting the transverse diffusion coefficient, which is computed in p-type Silicon, as a function of the electric field, at 300 K, from 0 to 50 kV/cm.

A comparison is made between transient regimes obtained, on p-Si at 300 K, by a Monte Carlo simulation and by an iterative solution of the Boltzmann equation : the agreement is excellent. The transient regime : a) depends relatively few on the energy bandshapes ; b) Is strongly dependent on the initial energy, the increase of which may lead to a disappearance of the velocity overshoot ; c) Slowly depends on the shape of the initial distribution function. In every cases, the transient regime computed using the balance equations is in very satisfactory agreement with the previous simulations.

Let us consider a one port network biased by a d.c. voltage Vo with a d.c. current Io passing through.

Every electrical system biased by a dc voltage or a dc current exhibits fluctuations of the voltage and/or of the current at its terminals. These fluctuations are called the noise. The bias point is determined as the intersection of the characteristics of the system under consideration and of the output circuit. Therefore, the noise depends partly on the inner properties of the system. In the same way, the noise of the system is also governed by its inner properties.

The noise in single injection space charge limited current (SCLC) diodes has been studied by several authors [1],[2]. It is shown that the noise voltage spectral density can be expressed, at frequencies below the reciprocal transit time, as $$ S_V \left( f \right) = 8KT\frac{{dU}} {{dI}}, $$ (1) where T is the lattice temperature. With the usual definition of the noise temperature TN $$ S_V \left( f \right) \equiv 4KT_N \operatorname{Re} \left\{ Z \right\}, $$ (2) the equivalent noise temperature TN of the device, at low frequencies, is TN = 2T.

Noise in devices is due to local random events, namely scattering mechanisms undergone by the carriers, which produce fluctuations of the current flow, and hence fluctuations of the voltage at the ends of the device. The knowledge of the noise is important mainly for two reasons: From a fundamental point of view, it gives information concerning the transport mechanisms (basic limitations such as the scattering mechanisms with phonons, etc., or technological limitations due to impurities, defects, etc.); from an applied point of view, noise is a fundamental limitation to the sensitivity of the device, so that it is important to study whether it can be lowered or not.

The experimental excess noise current in n-Si at 77 K is compared with the theoretical generation-recombination (GR) noise. The discrepancy between the theory and the experiment shows that the usual formalism describing the GR noise is not valid in that case. The observed noise is either hot carrier diffusion noise, eithe hot carrier GR noise, that is GR noise with electric field dependent parameters.

Most of the experimental methods intending to exhibit quasi ballistic regimes involve the measurement of velocity overshoot. We suggest an experiment in which the variation of the noise would be measured. For this purpose we set up a new theoretical method for modeling the noise current Sig in the gate of FETs, which shows that SiG is related to the noise temperature of the active channel under the gate.

We present noise and conductivity measurements of electrons in GaAs at 300 K in the frequency range 220 MHz - 10 GHz. The noise is white. Partition noise is shown to occur above 2 kV/cm. Ue give a method to determine the differential conductivity versus the electric field, and we deduce the diffusion coefficient versus the electric field, up to 3 kV/cm.

International audience