Dear Kai,

the carbon solubility in Mo is expected to be quite low. Even at the large flashing temperatures you use the solubility is in the range of 200 mass ppm (approx. less than 1 C atom per 500 Mo atoms). See, e.g., eqn(8) or (9) in Ref.

Materials Transactions, Vol. 55, No. 12 (2014) pp. 1786 to 1791

Assuming that 1 monolayer of C (with graphene-like packing density) is formed due to bulk to surface diffusion upon cooling, the numbers listed above predict for 2 mm thickness the application of about 10000 cycles as the worst case scenario (i.e. C is only removed due to Ar ion etching, no temperature induced desorption and the Mo-sample holder interface is neglected).

Is the C-1s binding energy related rather to adsorbates (> 285 eV) or carbide (e.g. 282.7 for Mo2C)? Do you see a C-1s signal in case that only Ar ion etching is applied, i.e. without further annealing?

Regards, FM

If this is a new sample you might have to do routine procedures for several days.  And flashing should be done at some higher T than 2000 K.

Kai, I have experienced similar observations in other metallic systems as well - even after repeated heating and sputter cleaning treatments, always new Carbon on the surface. My experience is, that when you stepwise lower the heating temperature, you may reach a state where the near-surface region (in terms of the diffusion length) is depleted with Carbopn, and you will not observe any new carbon at the surface after some of those steps. I agree with FM (above) that several 100 cycles might be necessary to remove all the Carbon. Greetings, Dirk

Preparations still ongoing. Am still waiting here to collect some more experience. Interesting to me: the apparently (but not necessarily) contrary statements of TT & Dirk, i.e. using higher vs. progressively lower temperature. Using higher temperature would reduce concentration gradients near the surface which could be built up with Dirk's proposition.

Anyway, once we get where we want to be (or start to...) I'll report on progress :-) [Just hope it won't take such long that I won't remember this promise by then]

Dear Kai,

maybe it is useful to apply Ar ion etching during cooling down from the flashing temperature. Upon cooling, C starts to segregate to the surface and it is immediately removed by the etching so that further carbon can reach the surface which is also immediately removed, etc. By this procedure you can combine many of the individual cycles to a single preparation step.

Regards, FM

Hello Frank,

that is an idea I have never considered so far!! The process you describe, would that be a cross-over from (more) entropy driven at high temperature to (more) energy driven at lower temperature?

Ultimately ("of course", I'd like to say) we'd not want to sputter upon cooling, since that would likely deteriorate surface order. In the end, we want to grow crystalline alloy thin films on that surface, not just get rid of carbon...

Dear Kai,

to be more precisely: Sputtering upon cooling does not mean sputtering until room temperature. Since the equations I mentioned in my 1st answer predict a steep decrease of C solubility between 2000 K and e.g. ~ 1500 K, you may apply the Ar ion etching only within the few minutes until T has droped from 2000 K to 1500 K.

Now you stop Ar ion etching and no further defects are introduced. However, upon cooling from 1500 K to RT the sample is still in an annealing state for several minutes (e.g. between 1500 K and 1000 K) for healing defects.

For example: We also work with such transition metal substrates for growing graphene and boronitrene monolayers. Our setup is limited to about 1000 - 1100 K, (i.e. your "post-sputtering cooling" temperature would be still larger than our regular annealing temperature). And even with this soft annealing, we observe very sharp LEED patterns (as recorded with a combined EELS/XPD setup), see e.g. Figure 3 in

dx.doi.org/10.1021/la500633n | Langmuir 2014, 30, 6114−6119

Regards, FM

Thanks for being so explicit, Frank. Much appreciated!