Two-Dimensional Dopant Profiling of an Integrated Circuit Using Bias-Applied Phase-Imaging Tapping Mode Atomic Force Microscopy

Abstract

Tapping mode atomic force microscopy was used to spatially resolve areas of different doping type and density on a static rando m access memory integrated circuit. The application of a dc bias applied between cantilever and sample during imaging results in a change in the tapping-mode phase contrast that depends on the doping density of the imaged area. Our experiments demonstrated that this method allows for distinguishing between p- and n-doped areas as well as distinguishing between regions of doping den sities ranging from 1016 to 1020 cm-3. Methods for characterizing doping patterns in submicrometerpatterned semiconductor circuits are becoming increasingly important as device structures continue to shrink. During the last several years, a variety of techniques for two-dimensional doping profiling were introduced. Currently, scanning capacitance microscopy (SCM), scanning Kelvin force microscopy (SKFM), and nanospreading resistance probe (nano-SRP)1-5 show the most promise among these techniques for fulfilling the requirements specified by the National Roadmap for Semiconductor Technology (NRST).6 These requirements include: 20 nm spatial resolution, 10% uncertainty in the dopant concentration determination and a sensitivity range for dopant concentrations from 1 x 1014 to 1 x 1020 cm-3. Furthermore, it is desirable that the measurements are reproducible and nondestructive. The above mentioned scanning probe methods are at least partially capable of fulfilling these specifications, however each technique falls short for at least one of the requirements in the NRST. The greatest problems with SCM and SKFM are associated with inconsistencies in sample preparation. In the case of nano-SRP, the measurement results in damage to the sample surface.7 Recently, we introduced a new tapping mode atomic force microscopy (TMAFM)-based technique that can laterally resolve regions of varying dopant density and type. 8,9 This method uses TMAFM with an additionally applied dc bias between cantilever and sample. Depending on its polarity and magnitude, the bias introduces Coulomb forces between the cantilever and the sample surface. The relative strength of these forces is a function of the doping density. These minute variations in the Coulomb forces can be monitored as a change in the phase of the cantilever oscillation relative to the cantilever driving frequency (TMAFM uses a cantilever excited into resonance with an oscillating piezoelectric driver for the imaging process).10 Topographic and doping level dependent phase images can be acquired simultaneously with this method by operating the microscope in the so-called interleave lift mode in which each line in an image is scanned twice. The first scan produces a standard TMAFM height image where the amplitude of the cantilever oscillation is monitored as a feedback signal. The second scan retraces the same line at a user defined height above the sample following the previously determined topography profile. This procedure results in the tip sample distance remaining constant during the second scan when the doping dependent phase image is measured. This procedure assures that the measured phase contrast depends exclusively on the electronic properties of the sample surface and avoids distance dependent Coulomb forces. It also eliminates short range dispersion forces that can dominate the tip-surface interactions at distances near contact.