| Author | Message | ||
     Christina Timofeyenko (Ctimofeyenko) |
I have seen three proposed products for the reaction of hydrofluoric acid (HF) with silicon dioxide (SiO2), SiF4, SiF6 and H2SiF6. How would these species alter the NIR absorbance spectra of a HF:NH4F:H2O water mixture? | ||
| hlmark |
Christina - This is actually a very complicated question, and the answer will be known only when someone actually does the experiment, but there are a few things we can predict about it: SiF4 and SiF6 have no hydrogens in their own structure and therefore we can expect that they will contribute little, if anything, to an absorbance spectrum in the NIR region. The H2SiF6, on the other hand has some hydrogens in tis own structure, and therefore we can expect some bands due to the presence of the hydrogen atoms in these molecules. However, that's by far not the end of the story. The presence of so many F atoms in a single molecule will create very electronegative centers, which in turn will have major effects on their ability to hydrogen-bond with the surrounding water. This is true for all three molecules. In fact, the effects due to hydrogen-bonding may even be greater than the intrinsic absorbtions of the hydrogen even in the H2SiF6. This will also be affected by the acidity (i.e., pH) of the solution as well as such characteristics as their ionic strength, which will depend on the concentrations of the various species involved, as well as the presence and concentrations of any other ions in the solutions. As far as I know, surprisingly little work has been done on the effects of pH and ionic strength on the spectra of water and aqueous solutions. Howard \o/ /_\ | ||
| Christina Timofeyenko (Ctimofeyenko) |
Howard-I quickly looked up the proposed reactants and it seems that SiF4 exists at ambient temperature and pressure in the gas phase. It also appears that SiF6 should have a -2 charge so H2SiF6 may be more probable. Just to give you a little more background: I am working on a calibration for a 100:1 Buffered Oxide Etch bath and I modeled the system, for simplicity sake, without concern for the resultant etch products present in the bath. The model seems to predict well before wafers are etched and then is horrible after wafers have been etched...I am trying to hypothesize the reason. Christina | ||
| hlamrk |
Christina - well, silicon chemistry isn't my forte. We're all just hypothesizing here (i.e., guessing) based on some of the known properties of the materials, to try to get a direction for follow-up experiments to understand what's going on. So, for example, as you say, pure SiF4 is a gas at standard conditions. But to draw an analogy, so is pure ammonia (NH3). But it's also known that ammonia is highly soluble in water, and it may be that SiF4 is also. Or it may be that SiF4 is somewhat soluble, and in different experiments different amounts of SiF4 evaporate from the solution, giving non-reproducible (i,e., "horrible") results. Or it may be that different amounts of the various products are formed, based on some etching condition that is not being controlled or monitored. That could be a physical condition (e.g., temperature) or a chemical one. Some reading in the literature of Silicon chemistry, and especially about the reactions with HF, might give some insight. Another possibility occurs to me: if one of the side reaction leads to formation of SiO2, as you mention, then I would expect that side-product to precipitate, possibly as very fine crystals that might remain suspended. This would give a variable amount of scatter to the solution, which might also create non-reproducible effects from one run to another, if the proper conditions aren't controlled. Good luck Howard \o/ /_\ | ||
| Gabi Levin |
Hi Fellows - let me try and throw in my nickel's worth - Let us start by asking: 1. What species did you model for? The HF or the NH4F which is the buffering agent I trust through the common F(-) ion. 2. The spectrum depends on the pH, which determines how much of the HF is dissociated and how much is not, this is a known. 3. The Buffer is 100:1, I assume 100 HF and 1 NH4F, so it does not have a very large buffering capacity - is that correct? 4. How did you model? Did you use PLS1? or MLR? 5. When you say that the models are breaking down - is it after a small amount of etching, or after a substantial amount of etching? 6. By how much did the pH change during the etching that causes the breakdown of the model? More general comments - SiF4 is a gaseous species, and I am short of detailed knowledge of its chemistry in aqueous environment. From the fact that it can be corrosive to metals in humid environment, I assume it does react with water, I don't know the details. The products of such reaction could be HF and SiO2, possibly appearing as particulate material, see later also, and Howard's comment. The SiF6 in contrast is very very stable, does not dissolve in water, and will not stay in the chemical system, so it is probably expelled out and will not affect the spectrum. The third product - The third compound is known as flourosilicic acid and according to data sheets it is a strong acid, and I assume that in the concentrations it is forming it is completely dissociated into 2H+ and SiF6(--) If it is fully dissociated it is not contributing anything of significance to the spectrum. My main suspicion is that: 1. The buffer having small buffering capacity, the pH changes even after relatively small amount of HF usage. 2. With the pH changing - the spectrum of the water changes due to changes in the over all number of hydrogen bonding. This is a fact!. We have done this measurement in etching solutions years back in IBM. We correlated very nicely to etching solution (composition was different in some ways) strength as function of bath usage by following the changes in the water OH band that result from pH changes. The last idea that Howard brought up - appearance of particulate matter that diffuses the light, is very important, it needs to be checked, if it happens, you shall create calibrations under applicable conditions. In the case where we worked in IBM this did not occur. I hope this helps you in any way. If you wish to communicate with me on a one to one basis, please don't hesitate to write to me [email protected] Msny thanks, Gabi Levin, Brimrose | ||
| Christina Timofeyenko (Ctimofeyenko) |
Hello-Thanks for all your input. 100:1 BOE is actually 100 parts NH4F to 1 part HF...in this case maybe the pH is not affected as much? I have modeled both HF and NH4F. I used the PLS1 algorithim. It is hard to say how many wafers are being etched because we have not been give that information and of course it changes from day to day. At this time we do not have access to any pH data. I will need to do some more digging in the area of Si chemistry. Thanks a lot, Christina | ||
| hlmark |
Christina - keep in mind that if the SiF6 is insoluble then it can also be "expelled" (to use Gabi's word) in the form of small crystals that will stay suspended and scatter the light. This will depend on concentrations, rates of reaction, etc. and not determinable only from the chemistry - you'll have to know something about the process, too. \o/ /_\ | ||
| Gabi Levin |
Hi Christina, If it is 100:1 NH4F to HF, then there is another problem - there is very little acid to react with SiO2, so the pH can change, but probbaly not as much. Yet, if there is little acid to react, there will be low amounts of the products. Now - I did some checks - SiF4 will react with water to yield HF and Si(OH)4 which does precipitate as small particles, if the concnetration is sufficient to cause precipitation. Since I don't have solubility data on this base, and the concnetrations are unknown, I can not tell you if it does precipitate. You have to pull a sampe of the solution and examine it. SiF6 will not, under any circumstances form crystals, it is a gaseous compund down to very low tempertaures, it will simply leave the chemical systems through the exhaust of the fume hoods they have above the tanks, and I happen to be familiar with the behviour of SiF6, as I did research back in the seventies on this compound. To really make progress you need to have samples of the solution at different "ages" i.e., different degrees of usage, and examine it visually. If you see any turbidity, this can be a real cause for your problems. The change in pH - you need to have this data to better understand the problem. If you wish to share more info with us, please be welcome Gabi Levin Brimrose | ||
| Christina Timofeyenko (Ctimofeyenko) |
Thanks Gabi for your post...interesting thoughts. We have been saving the samples because we do want to measure Si concentrations on the ICP and see if we find any correlations. At this point I have not noticed any turbity...but will keep an eye out for that. I think it will also be interesting to take pH measurements. |