Low frequency noise measurements in InN films
|Jean-Marc Routoure 1, Laurence Mechin 1, Bruno Guillet 1, Arantxa Vilalta Clemente 2, Javier Grandal 3, Miguel Ángel Sánchez-García 3, Sara Martin 3, Fernando Calle 3, Pierre Ruterana 2|
1. Groupe de Recherche en Informatique, Image, Instrumentation de Caen (GREYC), bd Maréchal Juin, Caen, France
We report DC and low frequency noise measurements at room temperature in InN films patterned using “transmission line method” structures. Two samples with thicknesses 250 nm (sample A) and 345 nm (sample B) were investigated.
The InN films were grown using MBE on a 5 μm thick GaN template on top of a 300 μm thick Al2O3substrate. The width of the devices was 94 μm and the length varies from 38 μm to 210 μm. DC and Noise measurements were performed on more than 25 devices. A mapping of the two samples was performed as well as a study as a function of the length in each sample. The mean resistivity for sample A is 3.8x10⁻⁶Ω⋅m(+/-15%) and 12.7x10-6Ω⋅m(+/-4%) for sample B. These values are in the range of those reported in the literature.
The noise spectral density consists of a white noise contribution due to the thermal noise and in the low frequency range, mainly a 1/f contribution. The 1/f noise level varies with the square of the DC voltage V across the sample. A Hooge parameter αcan then be estimated according to the calssical empirical : SV/V²=α/(fxN) with SV the 1/f noise spectral density level estimated at the frequency f and N the total number of free carriers in the device. For samples with carrier concentrations in the range of 10¹8- 10¹9at/cm³ the Hooge parameter is found to lie between 4x10-5- 4x10-4. The Hooge parameter is a normalized figure of merit for the device quality. Values reported for different kind of materials goes from 10-7for pure and defect-less material up to more than unity. The values obtained in this instance are in the same order of magnitude than the Hooge parameter of standard CMOS silicon processed devices which may be an indication that the studied devices are of substantially good quality.
Finally, the study versus the length of the devices reveals that the metal-InN contact contribution is not negligible for either DC and noise measurements. More statistics are needed for an elaborate identification.
Acknowledgment: This work is supported by the EU under the Grant agreement N°: PITN-GA-2008-213238, Initial training network RAINBOW of the 7 RTD Framework
Presentation: Oral at E-MRS Fall Meeting 2009, Symposium A, by Jean-Marc Routoure
See On-line Journal of E-MRS Fall Meeting 2009
Submitted: 2009-05-25 17:14 Revised: 2009-06-07 00:48