DISCUSSION OF THE PAPER

Participant: I have a comment and a question. Without having seen your original spectrum of the epitaxial silicon/bulk silicon composite, what I suggest is that your detection of species such as chromium, manganese, iron and copper may in reality be the following: at mass 52, since you have a large carbon peak, it could be Sic2+; at mass 55 instead of manganese you may have AlSi; at mass 56 you may have Si2, and since you have a large sodium peak this may not be out of the question, and at mass 63, 65 the polyatomic peak SiC1. Now if your energy bandwidth is broad, you are going to be in trouble with polyatomic species, and therefore I suggest you look once again to the possibility of interferences.

The question is the following: In the profiling of phosphorus and silicon how do you discriminate between SiH at mass 31 and 31p especially at low concentrations?

Phillips: Your comments are well taken. Although I cannot give you a comprehensive answer to your question, the need for correlation experiments is recognized. As I indicated, mass analysis impurity data are being correlated with impurity data from flame emission spectroscopy measurements. Morabito: If you did not have this chemical technique you would not know from the SIMS technique what these impurities are because of these mass resolution problems. Although we have looked at phosphorus and silicon and it is not impossible to resolve SiH from P'. The technique there is to look at the PO species.

Dobrott: I have a few comments on applications with the IMMA. At that phosphorus level the SiH is nearly completely absent from the spectra provided the coldplate was run to gather most of the water. The SiH

will not get you into trouble until you get much below the 1017/cm3 region for phosphorus. The mass 52 which you identified however if that were

as Cr could be Sic 2

the case you would see a much stronger mass 40 peak and also the mass 50 peak of Cr would be absent. You are probably perfectly right that mass 56 corresponds to iron because if you had much Al you would mask the iron 54. However, again, the intensity of molecular species is down considerably from what the monotonic species is, so you could get some feel of iron. Now when it comes to the copper of mass 63 and 65, it is true that silicon 28, together with chlorine 35 and 37 would dominate at 63 and 65, and unfortunately as nature would have it, the isotope ratio is about the same as what you would expect for copper. But it is unbelievable to me that he could take an epitaxial grown silicon and have enough chlorine incorporated into it that it would combine with the silicon and still come out to anything detectable by the IMMA.

The only other comment that I have is more or less a question. It was stated that the concentration level of anything which was not detected was below 0.05 ppma. I do not believe that this value is quite correct since a considerable number of noble ions were present. I would say the correct value that I would put on that with confidence is less than 10 parts per million unless you did specific long term counting on those particular ions.

Phillips: One comment on chromium and titanium. Often times those two elements are intentionally added to the melt during sapphire growth for the purpose of minimizing the number of defects in the crystal as it is grown. It has been experimentally observed that the number of twins per unit volume can be reduced if either or both of those two elements are added in reasonably large trace quantities.