High-Frequency Extraction of the Extrinsic Capacitances for GaN HEMT Technology

Abstract

A novel extraction methodology is proposed to distinguish between the extrinsic and intrinsic capacitances of wide GaN HEMTs. This approach is based on the experimental observation that the real parts of the impedance parameters of such devices increase at high-frequency. The mathematical analysis clearly shows that this so far uninvestigated behavior can be attributed to the extrinsic capacitances

A reliable and efficient small‐signal parameter extraction method for GaN HEMTs

“…As can be seen, the intrinsic capacitances shown in Figure 10A are almost constant within the frequency range of measurement, while the others fluctuate above and below their average values as frequency changes, which also proves the reliability of the extracted extrinsic and intrinsic elements. 19,21 The fluctuations of R i , G ds , and R gd are obviously larger than those of G m and τ, because the extracted values of R i , G ds , and R gd are more sensitive to test data. We can obtain the final values of the intrinsic elements by averaging their values over the entire frequency range.…”

“…Numerous research works on SSECM of GaN HEMTs have been done in the past decade. [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24] Chigeava et al 10 established a GaN HEMT 14-element SSECM by measuring "cold field-effect transistor (FET)" S-parameters (V ds = 0 V) to extract extrinsic parameters directly under a large gate voltage. Jarndal et al 11 proposed a GaN HEMT 22-element SSECM, in which a search algorithm was used to optimize the parameters.…”

GaN HEMT noise modeling based on 50‐Ω noise factor

“…As illustrated in Figure , the tested device exhibits a positive derivative of Re( Z ij ) versus the frequency (PDRZ) under “cold” pinch‐off condition , due to the extrinsic capacitance contributions. Therefore, the extrinsic capacitances have been obtained by increasing their values from zero until the PDRZ effect disappeared from the de‐embedded data, as proposed in . After removing the extrinsic capacitance contributions, the extrinsic resistances and inductances have been obtained from the slopes of the straight lines approximating ω 2 Re( Z ij ) and ω Im( Z ij ) versus ω 2 , respectively (see Fig.…”

An accurate parasitic parameters extraction method based on FW‐EM for AlGaN/GaN HEMT up to 110 GHz

“…However, the complicated parasitic parameters network will dramatically increase the difficulty of parameters extraction. Most of the papers have to use optimization method, which have challenges like multivalues problem, nonphysical results, and difficult to separate the parasitic capacitance and intrinsic capacitance . It is need the guessed starting values of the parasitic parameters and a family of scaling devices be available for measurement to get the optimal values for device by Sergio .…”

“…Most of the papers have to use optimization method, which have challenges like multivalues problem, nonphysical results, and difficult to separate the parasitic capacitance and intrinsic capacitance. 8 It is need the guessed starting values of the parasitic parameters and a family of scaling devices be available for measurement to get the optimal values for device by Sergio. 9 Zlatica 10 used artificial neural networks for constructing a temperature-dependent model representing the small-signal scattering (S-) parameters of a GaN HEMT over a wide bias range and a board frequency range, while this method need tremendous measured data and high computer configuration.…”