Thanks for that response.  It is certainly possible that the P is being
sputtered more than Ga. My initial thought was that if the GaP bonds are
being dissociated by the Ar+ ions, then both Ga and P will be above their
melting temperature at the target surface. Ga = 30C, P = 45C. GaP = 1540 C.
So while GaP is being sputtered, any elemental remains of Ga and P will be
evaporating, with Ga faster than P. That would result in a P-rich target and
substrate. Let me know if that reasoning makes sense. Also, I noticed a
distinct 'metallic' smell in the chamber which I am inclined to attribute to
gallium.



On Thu, Jan 29, 2009 at 11:57 AM, Edward Sebesta  wrote:

> The GaP is certainly decomposing during sputtering. Sputtering is
> knocking individual atoms off the target with Argon atoms. However the
> relative vapor pressures aren't relevant to the problem here. If a Ga
> atom is sputtered and its trajectory reaches your substrate there should
> be deposition of the atom and not evaporation.
>
> The issue here is the different atomic weights of Ga and P, (69.72 and
> 30.97 respectively) relative to Argon atom (39.95) which is used to
> sputter. Also, heavier elements have more electrons to result in
> inelastic collisions and reduce sputtering efficiency, if I remember
> correctly. You may be getting a much lower flux of Ga than P upon
> initial sputtering. I am guessing that the wafer you are using as a
> target is Ga rich. Continued sputtering might make it Ga rich enough
> that your depositions balance out. It would be interesting to see what
> would be the composition after you ran some conditioning wafers and then
> redeposited and remeasured your composition. Perhaps a higher power
> might cause the Ga to sputter with higher relative flux to P, but you
> would still have some imbalance.
>
> Evap methods for compound materials involve conditioning the alloy
> target until the composition of the melted puddle is skewed so that its
> relative rates of evaporation match the desired alloy composition. So if
> you have A/B alloy. And A has twice the vapor pressure as B, then you
> run the target until the e-beam puddle was 1/3 A and 2/3 B and the vapor
> pressure was balanced and as the target melted the puddle would keep a
> constant composition. Actually, it would be much more complicated if the
> vapor species were not monoatomic but perhaps A2 or something and
> include some AB. But the same principle would apply as conditioning the
> target until a balanced flux. Russel J. Hill, in the 1976 Airco Temescal
> book explains alloy deposition in detail.
>
> Ed