文摘
The decoupling of epitaxial factors influencing the dynamic instabilities of AlGaN/GaN metal–insulator semiconductor high-electron-mobility transistors is investigated. Three different sets of samples have been analyzed by means of dynamic instabilities in the threshold voltage (math-equation-construct">mage="true" class="math-equation-image">mathml="true" class="math-equation-mathml" style="display:none"><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><msub xmlns:w="http://www.wiley.com/namespaces/wiley" xmlns:wiley="http://www.wiley.com/namespaces/wiley/wiley" xmlns:cr="urn://wiley-online-library/content/render" xmlns="http://www.w3.org/1998/Math/MathML"><mi>Vmi><mi mathvariant="normal">thmi>msub>mml:math> shift). Secondary ion mass spectroscopy and steady-state photoluminescence (PL) measurements have been performed in conjunction with electrical characterization. The device dynamic performance is found to be significantly dependent on both the C concentration next to the channel, on the distance between the channel, and the higher doped C region. Additionally, we note that experiments studying trapping should avoid large variations in the sheet carrier density (math-equation-construct">mage="true" class="math-equation-image">mathml="true" class="math-equation-mathml" style="display:none"><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><msub xmlns:w="http://www.wiley.com/namespaces/wiley" xmlns:wiley="http://www.wiley.com/namespaces/wiley/wiley" xmlns:cr="urn://wiley-online-library/content/render" xmlns="http://www.w3.org/1998/Math/MathML"><mi>Nmi><mi mathvariant="normal">smi>msub>mml:math>). This change in the math-equation-construct">mage="true" class="math-equation-image">mathml="true" class="math-equation-mathml" style="display:none"><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><msub xmlns:w="http://www.wiley.com/namespaces/wiley" xmlns:wiley="http://www.wiley.com/namespaces/wiley/wiley" xmlns:cr="urn://wiley-online-library/content/render" xmlns="http://www.w3.org/1998/Math/MathML"><mi>Nmi><mi mathvariant="normal">smi>msub>mml:math> itself has a significant impact on the math-equation-construct">mage="true" class="math-equation-image">mathml="true" class="math-equation-mathml" style="display:none"><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><msub xmlns:w="http://www.wiley.com/namespaces/wiley" xmlns:wiley="http://www.wiley.com/namespaces/wiley/wiley" xmlns:cr="urn://wiley-online-library/content/render" xmlns="http://www.w3.org/1998/Math/MathML"><mi>Vmi><mi mathvariant="normal">thmi>msub>mml:math> shift. These experimental trends are also supported by a basic model and device simulation. Finally, the relationship between the yellow luminescence (YL) and the band edge (BE) ratio and the math-equation-construct">mage="true" class="math-equation-image">mathml="true" class="math-equation-mathml" style="display:none"><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><msub xmlns:w="http://www.wiley.com/namespaces/wiley" xmlns:wiley="http://www.wiley.com/namespaces/wiley/wiley" xmlns:cr="urn://wiley-online-library/content/render" xmlns="http://www.w3.org/1998/Math/MathML"><mi>Vmi><mi mathvariant="normal">thmi>msub>mml:math> shift is investigated. As long as the basic layer structure is not changed, the determination of the YL/BE ratio obtained from steady-state PL is demonstrated to be a suitable way to predict trap concentrations in the GaN channel layer.