next up previous
Next: 2.1 The AVC Method Up: MINIMOS-NT Previous: 1. Introduction


2. Simulation of AVC Measurements

In the last decades the size of semiconductor device structures has been continually decreased and in parallel the number of devices contained on a chip has been increased approximately by a factor of 2 every 18 months which is known as Moore's law [4]. Inspite of predictions forecasting a slowdown or even a halt of further miniaturization the feature size will be continually reduced in the near future [5]. The decrease of the feature size increases the demand for methods for delineating the pn-junction position and for characterizing the doping distribution. Current state-of-the-art feature sizes are 0.18- 0.25 $ \mu$m [5]. The characteristics of semiconductor devices are determined mainly by the distribution of dopant atoms. These are intentionally added impurities which are used to change the equilibrium carrier concentrations. A doping characterization method has to be able to resolve a few percent of the feature size to be useful.

Several methods have been developed for this purpose. These are the Secondary Ion Mass Spectroscopy (SIMS)[6], the Spreading Resistance Probe (SRP) method [7], the Scanning Capacitance Microscopy (SCM) method [8], Capacitance Inverse Modeling [9], and the Auger Voltage Contrast (AVC) method [10]. These methods are all based on different physical principles and therefore have different advantages and disadvantages with respect to accuracy, necessary equipment, and cost.

The AVC method has been developed to meet the demand for a method for rapid one- and two-dimensional pn-junction delineation and doping profiling. This is of special interest for determining the depth of a pn-junction formed by implantation and diffusion.

In the AVC method a beam of high energy primary electrons is focused on the surface of the cross-sectioned test device and scanned across the surface. Because of their high kinetic energy the incident electrons generate electron-hole pairs in the semiconductor. A fraction of the secondary electrons has enough kinetic energy to leave the semiconductor. The kinetic energy of these secondary electrons conveys information on the doping at the location where the ionization took place.

The kinetic energy of the emitted secondary electrons is shifted due to the presence of a dipole layer on the surface caused by surface states. Theoretical studies [10] indicate that this effect is only important for small beam currents and that for higher beam currents the energy shift at the surface is equal to the energy shift in the bulk. Additionally, very small beam currents have to be avoided in actual measurements because the number of emitted secondary electrons which can be used for averaging is too small and the signal-to-noise ratio is not sufficient to produce reliable results.

Analytic approximations describing the effects caused by the incident high energy electrons during an AVC scan which are valid for small beam currents break down for current densities used in actual measurements. Therefore the differential equations for the potential and the carrier concentrations have to be solved numerically. To gain better understanding of the physics involved in AVC measurements and to facilitate the interpretation of the results of AVC measurements the device simulator MINIMOS-NT has been enhanced to be capable of simulating such measurements.

First the AVC method is described in detail and approximations for the measured surface potential are derived for idealized conditions. Then the influence of the semiconductor surface and the connected surface potential barrier and surface states is investigated. The models for simulating AVC measurements which have been implemented in MINIMOS-NT are presented in the third section of this chapter. Then the simulation of an AVC scan is described and the simulation of some simple devices is presented. Finally the limits of the AVC method are discussed and the method is compared to other doping characterization methods.




next up previous
Next: 2.1 The AVC Method Up: MINIMOS-NT Previous: 1. Introduction
Martin Rottinger
1999-05-31