Image depth profiling is applied to the quantltative analysis of molecular beam epitaxlally grown gallium arsenide thin layers. The technique Involves the use of Ion implantation through a mask and subsequent analysla by secondary Ion mass spectrometry (SIMS). The proposed approach provides high accuracy results in the analysis of semlconductor thin layers. Semiconductor materials research has succeeded in steadily reducing the physical size of electronic devices. With the development of eptiaxial growth techniques, particularly molecular beam epitaxy (MBE) (1-3), layered structures of thicknesses on the order of 1-100 nm with precise composition profiles have been obtained. The fabrication of devices by MBE requires the ability to accurately add controlled quantities of electrically active dopants, a process which, in many cases, is not well understood. As experimentation continues, the need for accurate measurement of dopant concentration in such layers challenges the state-of-the-art of surface analytical techniques. Secondary ion mass spectrometry (SIMS) is currently recognized as one of the most sensitive techniques for the analysis of semiconductor materials (4,5), with detection limits in the parts per million range for most elements. The application of ion implant standards (6, 7) to SIMS has been developed to provide accurate quantitative concentration depth profiles. The method of solid-state standard addition by ion implantation has been shown to give excellent results (8). This method involves implanting the sample with a controlled dose of the element to be determined, profiling through the implantation zone, and then comparing the integrated signal obtained from the implant to the background signal arising from the dopant being determined. To ensure complete removal of the implanted species, the residual dopant signal is acquired at a depth of at least three to four times that of the implant peak (8). This places a restriction on the minimum sample thickness which can be analyzed by this method. Furthermore, dopant inhomogeneity in the sample presents difficulties to the conventional technique. Lateral (z-y) inhomogeneities can be dealt with only by the tedious process of running replicate analyses on several sampling areas. In-depth heterogeneity poses an even greater problem as all but gross inhomogeneities are masked within the implant region. Hence, effects such as diffusion toward the surface cannot be dealt with reliably. This problem can be circumvented in some instances by implanting with one isotope of the element of interest to serve as a standard and monitoring the residual component with another isotope. However, this is not always feasible, as in the case of beryllium which is monoisotopic. In order to reduce the thickness requirement to thinner layers, and to eliminate the problem of dopant inhomogeneity, we investigated the use of image depth profiling (9). This technique fully utilizes the advantages of the ion microscope (10) by providing simultaneous in-depth analysis of several features within the ion image. Therefore, if the field of view of the image contains both implanted and unimplanted regions, implant and background signal levels can be obtained simultaneously to give quantitative concentration data. The only thickness limitation then becomes that necesssary to contain the entire implant zone, while dopant heterogeneity can be visually assessed and determined quantitatively by proper choice of sampling areas within the ion image. This method was applied to two thin layer samples. First, a quantitative determination of the beryllium dopant concentration in a 1 pm MBE grown gallium arsenide layer was performed as a test sample. The technique yielded a precision of 24% and was accurate to within 13%, both comparable to the results achieved by the conventional solid-state standard addition method. Next a GaAs layer containing a diffused silicon dopant was quantitatively analyzed.
Read full abstract
Dec 1, 1982
Adam J Patkin + 2
