doi:10.1016/S0026-2714(00)00077-9
Copyright © 2001 Elsevier Science Ltd. All rights reserved.
The coherence of the gate and drain noise in stressed AlGaAs–InAlGaAs PHEMTs
References and further reading may be available for this article. To view references and further reading you must
purchase this article.
B. K. Jones
,
, a, C. N. Grahama, A. Konczakowskab and L. Hasseb
a Physics Department, Lancaster University, Lancaster, LA1 4YB, UK
b Faculty of Electronics, Telecommunications and Informatics, Technical University of Gdansk, 80-952 Gdansk, Poland
Received 27 March 2000;
revised 5 May 2000.
Available online 22 December 2000.
Abstract
Results of electrical noise measurements on stressed pseudomorphic high electron mobility transistors are reported. The DC characteristics were measured. The voltage stress between the gate and drain was carried out with the channel off and produced breakdown walkout. Different amounts of stress were applied to various devices with an unstressed control. The noise in the gate and drain currents, and the coherence between them, was measured in the range 0.1–5 kHz and analysed as white and 1/f noise. The mechanisms of the noise sources and their coherence are discussed. The coherence was found to be of a good quality or reliability indicator.
Fig. 1. The DC characteristics of device A11 (highly stressed) at 300 K. The Id–Vds and Ig–Vds curves before (
) and after stress (—) are shown at various values of Vgs. The gate current increases and decreases as the voltage is changed monotonically because it is dominated by the hot carrier current as seen in Fig. 2.
Fig. 2. The DC transconductance and gate current curves for device A11 (highly stressed) at 300 K before stress. The stress point is also shown.
Fig. 3. The evolution of the drain current of device A11 during stress. The stress conditions were Vds=6.0 V, Vgs=−1.0 V so that Vgd=−7.0 V.
Fig. 4. Gate current against the drain–gate voltage for devices to show the effect of stress. The channel was open during these measurements but, since the voltage drop is highest at the drain we assume most current flows at this end. The gate–source voltage is −0.30 V so that the current is dominated by hot carrier current. In order of stress intensity: A13, • A10, ▪ A12, × A11.
Fig. 5. Schematic circuit diagram of the two-channel noise measurement system.
Fig. 6. The drain current, drain 1/f noise (in arbitrary units, au) and the 1/f coherence between the gate and drain noise for device A11 (highly stressed) as a function of Vds at Vgs=−0.30 V: □ Id , drain 1/f noise, × 1/f coherence – (a) normal connections to device and (b) as above but with source and drain connections reversed.
Fig. 7. The drain and gate noise (in arbitrary units, au) and their coherence as a function of drain current for device A11 at Vg=−0.30 V: • drain noise, ▪ gate noise, × coherence – (a) White noise and (b) 1/f noise (this contains some of the same data as in Fig. (5a)).
Fig. 8. The 1/f noise (×) and white gate noise (•) (in arbitrary units, au) for the stressed device A11 (with source and drain reversed).
Fig. 9. The noise in the devices against the gate current in order of stress: A13, • A10, ▪ A12, × A11 – (a) The white gate noise (in arbitrary units au) and (b) The gate 1/f noise (in arbitrary units, au). The slopes are 2, 1.7, 1.0, 1.5 as the stress increases. The noise at constant current is variable with stress but this may be because of current variations.
Table 1. Device parameters before and after (→) stress

Table 2. Noise (in arbitrary units, au) and coherence values for the devices interpolated to a drain bias of Id=9 mA and Vds=4.0 V

Corresponding author. Tel.: +44-1524-593657; fax: +44-1524-844037; email: b.jones@lancaster.ac.uk