This work presents a test structure targeted to measure MOSFET mismatches with a fast method. It relies on two single-spot voltage measurements in order to extract $\Delta V_{TH}$ and $\Delta \beta / \beta $ separately. The new methodology gives a theoretical increase in the measurement speed of 30x (23.17x in practice). The coefficient of determination ( $R^{2}$ ) of the linear regression analysis is used to compare standalone transistor measurements against the new proposed methodology. The correlation in the data demonstrates values not less than 0.94 ( $R^{2}\geq0.94$ ). The test structure can reproduce parameter correlations, and it is capable of extracting MOSFET mismatch design parameters, such as Pelgrom’s $A_{V_{TH}}$ , with an error of 2% and $A_{\beta }$ , with a negligible error. The experimental data presented herein are taken from measurements in prototypes fabricated in a 65nm CMOS bulk process. The whole circuit is composed of 16 2D addressable DUT device matrices, each having 256 same-size closely-placed MOSFET devices, totaling 4,096 MOS devices used in single-type (NMOS) transistor array.

中文翻译：

---

用于快速 MOSFET 器件失配表征的基于两个晶体管电压测量的测试结构

这项工作提出了一种测试结构，旨在通过快速方法测量 MOSFET 失配。它依靠两个单点电压测量来提取 $\Delta V_{TH}$ 和 $\Delta \beta / \beta $ 分别地。新方法理论将测量速度提高了 30 倍（实践中为 23.17 倍）。决定系数（ $R^{2}$ ) 的线性回归分析用于将独立晶体管测量与新提出的方法进行比较。数据中的相关性表明值不小于 0.94 ( $R^{2}\geq0.94$ ）。测试结构可以重现参数相关性，并且能够提取 MOSFET 失配设计参数，例如 Pelgrom 的 $A_{V_{TH}}$ , 误差为 2% 和 $A_{\beta }$ ，误差可以忽略不计。此处提供的实验数据取自在 65 纳米 CMOS 批量工艺中制造的原型的测量值。整个电路由16个2D可寻址DUT器件矩阵组成，每个矩阵有256个相同尺寸的紧密排列的MOSFET器件，总共4,096个MOS器件用于单型（NMOS）晶体管阵列。