The 1∕f noise in carbon nanotube random network films has been experimentally and theoretically investigated. The authors have established that the percolation process is the primary physical mechanism influencing the noise level in such films. Using percolation theory, the authors give an expression for the 1∕f noise scaling that describes how noise is affected by design variables of nanotube macrostructures. As expected for percolation networks, the noise amplitude strongly depends on the film homogeneity. This shows that noise analysis could be used as a tool to evaluate the quality of films.

Low-frequency noise measurements are performed on classical Si bipolar transistors and on AlGaAs/GaAs heterojunction bipolar transistors (HBTs) mounted in a common-emitter configuration. Expressions for the spectral densities are derived taking into account a correlation between base and collector noise sources ib and ic. Values of emitter series resistances and of the base ideality factor are determined from these noise measurements. Then the spectral densities related to ib and ic as well as the cross-spectrum are extracted. In the case of classical Si transistors, the excess noise is attributed to the current noise source ib. The effect of the base series resistance is shown on the white noise. For the HBTs, the white noise is not reached. The excess noise is attributed to the correlated current noise sources ib and ic. From the analysis of the current spectral densities with base current the increase of correlation with bias is revealed. The extracted current spectral densities provide the foundation for the calculation of the coherence function associated to base and collector currents. This coherence function is presented in part II. The shape of the curves shows correlation phenomena to be mainly related to generation-recombination components. Also in part II, the spectral densities associated with the correlated and uncorrelated part of the collector current are extracted and studied versus bias and geometry.

The influence of carbon concentration on the low-frequency noise (LF noise) of Si∕SiGe:C∕Si heterojunction bipolar transistors (HBTs) is investigated. When carbon is incorporated into these HBTs, representative noise spectra of the input current spectral density SIB show significant generation-recombination (GR) components. On the other hand, for transistors without carbon incorporation, no GR components were observed. When only 1∕f noise component is observed, the 1∕f noise level is found to be independent of the carbon concentration and the associated figure of merit of the normalized noise magnitude KB has a very good value of ∼4×10−10μm2. In order to relate the 1∕f noise and the high-frequency performance of the transistor, we studied and modeled the figure of merit defined as the ratio fc∕fT (fc is the low-frequency corner frequency and fT the unity current-gain frequency). Then we performed a detailed analysis of the GR components associated with the presence of the carbon. We found that the observed Lorentzian spectra are associated with random telegraph signal (RTS) noise. However, no RTS noise was measured in carbon-free devices. It is believed that the RTS noise is due to electrically active defects formed by the addition of carbon, typically observed for concentrations above the bulk solid solubility limit in silicon. The RTS amplitude (ΔIB) is found to scale with the base current, to decrease exponentially with temperature, and to be independent of the carbon concentration. The mean pulse widths (tH,tL) of the RTS are found to decrease rapidly with bias voltage, as 1∕exp(qVBE∕kT) or stronger. Our results confirm that electrically active C-related defects are located in the base-emitter junction, and the RTS amplitude is explained by a model based on voltage barrier height fluctuations across the base-emitter junction induced by trapped carriers in the space charge region. The observed bias dependence of mean pulse widths seems to indicate that two capture processes are involved, electron and hole capture. These C-related defects behave like recombination centers with deep energy levels rather than electron or hole traps involving trapping-detrapping processes.

No abstract