room temperature. The silicon secondary ion yield changes by several orders of magnitude as the interface is penetrated. At all points in the specimen except at the Si02-Si interface the sodium profile shows a count rate which is essentially at the background noise level of the instrument.

In figure 3 we see the results of cooling the same specimen to -180°C and this analysis demonstrates that the drift of sodium is retarded. The instrumental conditions were kept exactly the same in both cases. Here we see that when the specimen is cooled the sodium does not move rapidly to the Si02-Si interface during analysis, but is retarded; in fact, the sodium concentration is two orders of magnitude higher throughout the oxide in the spot analyzed on the cooled specimen than it is in the spot analyzed at room temperature. Although this technique does not yet give the true sodium profile, it does give some encouragement to continue on this path of analysis, and table 2 describes the next set of specimens which were prepared. The first specimen, JR1C, the asgrown oxide control sample, was cut up and used as a base material for all of the sodium ion implants which ranged from 3.2 × 1014 to 1.4 x 1015 atoms per square centimeter total dose; some of which were annealed at 200 or 900°C. The results obtained on these samples are tabulated in table 3. In specimen JR1C analyzed at 25°C with the 0 beam, 100% of the sodium appears at the surface of the oxide; at -180°C with the 0 beam, 97.5% of the sodium appears at the surface. Therefore, we see that in the control specimen essentially all of the sodium is present at the surface of the oxide as contamination. The

next experiment on JR1C using the 02+ beam

demonstrates that we still have a fair amount of the sodium mobilized both at 25°C and at -180°C. At 25°C, 10% of the sodium is at the surface and 89% is at the Si02-Si interface. Going to -180°C on the same sample, we see that 54% is on the surface of the oxide, 26% at the interface, and 20% is included within the bulk oxide. So we have retarded the movement of the sodium although we do not feel that we have a perfect profile yet. Working with specimen JR1A1, which was implanted to 3.2 × 1014 atoms per centimeter and annealed at 200°C, we see that there is a very small amount of the sodium at the surface in both the -180°C and the 25°C specimens. Most of the sodium has gone to the interface. We interpret this as being caused primarily by the ion induced channels from the implantation process and by the fact that the 200°C treatment does not anneal out these channels. We still

have an inducement for the sodium to move, even though we have cooled the sample. JR1A2 has been annealed at 900°C for 30 minutes, and in this case we see at -180°C that 29% of the sodium is on the surface. 65% is at the interface. Essentially the same appearance was seen at 25°C, showing that there must have been some prior movement of the sodium at room temperature and then at 900 or 200°C before we did the analyses.

In order to show the differences a little bit more clearly we have made up differential plots between the work done at room temperature and the work done at low temperature. Figure 4 shows the results obtained with a differential plot of JR1C, our control specimen. This plot demonstrates that cooling holds some Na at the surface of the Si02, retards its diffusion through the Si02, and greatly reduces the pileup at the Si02-Si interface. In figure 5 we are looking at the JR1A1, which is an implanted and low temperature-annealed sample. We see that there is a sizeable portion of the sodium present which is retained at the surface by cooling the specimen during analysis. In the sodium profile through the sample there is a large decrease in the peak right before going into the silicon substrate and some reproducible small peaks as you sputter into the silicon itself that must be caused by interface states and the differences in sodium activity between 25°C and -180°C. In figure 6 we are looking at the differential plot of JR1A2, which was annealed at 900°C for 30 minutes. Here we see somewhat the same results as with JR1A1 except that there is a wide peak of the sodium at the surface compared to the specimen analyzed at room temperature and a very deep valley when we get close to the interface. The integrated sodium intensities of the two JR1A2 specimens were 26,000 and 29,800 counts, compared to 54,000 and 44,000 for the JR1A1 specimens. This difference may be due to the loss of sodium from the surface during the high temperature annealing process. Such loss caused by diffusion of sodium to the surface and partial escape into the furnace atmosphere would also account for the increased surface concentrations of sodium in the JR1A2 specimens, 22 to 29% compared to 6 to 7% of the JR1A1 specimens. That the sodium concentration at the surfaces of the JR1A1 specimens is lower than would be expected for the shallow 200 A range suggests that the sodium ions may be driven into the Si02-Si interface by surface charging during the ion implantation.

A run on the JR1A1 specimen with 0 primary beam at -180°C showed a high but continuously decreasing sodium concentration through the oxide, and no sodium at the Si02-Si interface. As with the control, the highenergy 0 beam apparently makes the sodium more mobile in the oxide. We see in our third example, JR1A2, a fairly uniform distribution of the sodium throughout the Si02, with the peak reduced at the interface and a peak showing at the surface. Therefore, we can conclude that when a specimen which has little implantation damage is analyzed at -180°C, we can get a much better profile of the sodium which we would say does exist on the surface of the Si02 as contamination from the growth technique, and throughout the oxide from the implantation, than we can at room temperature.

SUMMARY

In summary the following points have been established.

3. The interactions of primary beam energy and species, ion implantation dose and energy, annealing temperature, and a variety of other factors not yet known, combine to affect the Na profiles obtained and to improve or degrade the benefits of cooling the specimens.

ACKNOWLEDGEMENTS

We would like to acknowledge the contributions of Mr. Ronald Baxter of Leeds and Northrop Company, North Wales, Pennsylvania in the design and initiation of these experiments, and the sponsorship of the Defense Nuclear Agency under subtask Z99QAXTD033, work unit 51, work unit title, "Radiation Resistant MOS Technology," technical monitor, Mr. H. L. Hughes, of the U. S. Naval Research Laboratory.

REFERENCES