Multi-Layer Multi-Semiconductor Characterization—Spectroscopic Toolbox for GaN HEMT

Transition from silicon to III-nitrides in power electronics poses a challenge for characterization of novel devices using silicon era methods. Unlike classic metal-oxide-semiconductor field effect transistors (MOSFETs), GaN high electron mobility transistors (HEMTs) are multi-layer and multi-semiconductor structures. Most existing characterization methods, intended to handle a single layer of silicon fall short of distinguishing between the different layers of a modern GaN device. To overcome this limitation, we propose to use light, which is able to excite each of the layers at its own band gap energy, enabling simultaneous probing of the whole structure, in-situ and ex-situ. Here, we present a set of different methods based on the Franz-Keldysh effect.[1-2] Our method utilizes the fact that at the presence of an electric field, a band-to-band transition commences at energies below the gap, and the resulting optical response is thus smeared to the red rather than having a step function shape. We illuminate the HEMT structure with a constantly increasing photon energy and collect an electric response. The acquired spectrum typically shows several transitions at different photon energies, each corresponding to the band gaps of a specific layer. The shape of each transition depends on the strength of the electric field at the corresponding layer. We utilize a simple mathematical model to convert the shape of each response into a linear curve, the slope of which yields the electric field in the layer, and its intercept with the energy axis equals the band gap. So far, we have developed several characterization methods based on the Franz-Keldysh effect. The first is based on channel photocurrent spectroscopy,[3] measuring the change in the channel current as a function the photon energy. The band edge response of each of the layers in the resulting spectrum is then analyzed using our model, revealing the built-in electric field. The results of our measurements are used to calculate the 2DEG sheet charge, channel mobility, and finally, to construct the band diagram of the device. Varying gate and drain voltages, we obtain a set of spectra to study the dynamics of a device under different operating conditions. Our second method utilizes a photovoltage spectroscopy measured on bare GaN HEMT wafers (prior to fabrication of contacts and device geometry).[4] Using the contactless Kelvin probe method, we obtain an equilibrium spectrum of the structure, followed by calculation of electric fields, band gaps, 2DEG charge and band diagram. In contrast to the photocurrent spectroscopy, here we measure the structure at zero bias only. However, we achieve powerful tool for in-situ metrology at the earliest stage of the fabrication process. The third method combines both photocurrent and photovoltage spectroscopies to find 2DEG sheet charge density and the channel mobility.[5] This method does not require a full spectrum, but rather only two points below and above the band gap of the AlGaN barrier. Finally, our forth method is similar to the first two, but may be applied to thick layers or bulk.[6] In this case, the mathematical model is more complex and requires parabolic fit. However, in addition to the band gap and the maximum electric field of the surface depletion region, we can also obtain the carrier concentration in the layer. The presented methods make up a powerful toolset for characterization of GaN HEMTs, but is not limited to transistors only and may be used on most of the multi-layer multi-semiconductor structures, at the end or during the fabrication process.<br/><b>References</b>: [1] W. Franz, Z. Naturforsch. A 13, 484 (1958). [2] L. Keldysh, Sov. Phys. JETP 7, 788 (1958). [3] Y. Turkulets, I. Shalish, J. Appl. Phys. 123, 024301 (2018). [4] Y. Turkulets, I. Shalish, J. Electron Dev. Soc. 6, 703 (2018). [5] Y. Turkulets, I. Shalish, Electron Device Lett. 40, 383 (2019). [6] Y. Turkulets, I. Shalish, J. Appl. Phys. 124, 075102 (2018).<br/><b>Acknowledgement:</b> Financial support from the Office of Naval Research Global through a NICOP Research Grant (No. N62909‐18‐1‐2152) is gratefully acknowledged. Approved for public release (DCN# 43-9231-22).