A theoretical model is presented to calculate the redeposition of ions and neutrals generated during sputter bombardment in a secondary ion mass spectrometry (SIMS) measurement. This model is compared with measurements of boron in silicon in an on-chip geometry, i.e. the analysis of low doping levels of an impurity in an area surrounded by a region that contains high amounts of this impurity. The present results indicate that the ion acceptance profile of the secondary ion beam and the redeposition of ions and neutrals generated during the sputtering form the limiting factors in small-area on-chip analysis. The non-ideal definition of the acceptance profile is solved by superimposing an electronic gate onto the optical gate of the instrument. For the analysis of B in Si and Si in GaAs and AI,Ga, -xAs, the use of Cs' reduces the background signal of the measurement compared with 0,'.
~~ ~~~
INTRODUCTIONSecondary ion mass spectrometry (SIMS) analysis of dopants, with its unparalleled sensitivity and depth resolution, has contributed in many ways to the progress of semiconductor technology by providing adequate feedback on the different processing steps. In many cases this analysis can be made on uniform test wafers, which provides ample material for the analysis. However, in a number of cases such a test wafer does not exist or one is interested in the effects of full processing on the dopant profiles in small areas. It then becomes necessary to perform a SIMS analysis on-chip in a well-determined (usually small) area.In the present paper we have investigated the implications and problems of such an on-chip analysis, bearing in mind that this sometimes implies measuring the wanted dopant profile in a small area surrounded by material containing very high amounts of the element of interest. To explore this in more detail, a special test sample has been made to simulate this situation and to make a quantitative evaluation of the mechanisms of interest possible. The test sample consists of squares, of various sizes, etched into a boron phosphor silicate Glass (BPSG) layer.
~~~~
EXPERIMENTALPreliminary experiments were performed on a uniformly implanted wafer whereby the field of view was continuously reduced. As expected, the reduction in sensitivity as the area is decreased was found to scale * Author to whom correspondence should be addressed. proportionally with the area being analysed when using the field aperture to reduce the analysed area. In fact it was found that the loss in sensitivity was less than expected based on area ratios, when changing to the small areas with using the transfer lens settings, because this increases the collection efficiency. This implies that a reduction of the analysed area with a 150 pm diameter down to 12 pm diameter generates no extra problems. The test sample (see Fig. 1) prepared consists of squares of Si, of size 1 pm-1 mm, etched in an SiO, layer (800 nm doped layer with 5% B and 5% P on a 200 nm undoped layer).'.' The squares themselves received a "B implantation at 40 keV o...