I'd say from my limited understanding, R-600 or R-600a would be ideal choices for the auxiliary cooler for the initial separation.
would you recommend using R-600a or even R-600 as the refrigerant with the highest boiling point?
i intend to use hydrocarbons (290/170/1150), just like gomeler
Hydrocarbon Refrigerants that were mentioned:
R600 n-Butane -0.5C
R600a IsoButane -11.7C
R290 Propane -42C
R170 ethane -88.6C
R1150 ethylene -103.7C

I think between R600 and R600a, I'd go with the R600 (remember earlier I mentioned that R114 would be an excellent warmer boiling candidate to replace the R123/R22 combo).

In the CFC days, a polycold charge consisted of:
R114 +3.6C
R13 -81.5C
R14 -129.9C (hmm this looks colder then I remember -- maybe check other MSDS's)
Argon -185.7C (no it wasn't condensed, just absorbed)

I think a charge made from R600, R170, R1150 comes the closest to the first 3 CFC choices, and would be the simplest combination that would work for -100C operation. I would go with this first, and see what you get. Using only 3 components will make it a lot easier to tune. After you have gotten all that you can get from this combination, go ahead and try adding a small amount of R290 to see if it buys you anything.

Don't expect too much load capacity at -100C, since you will be operating close to 0 gage pressure, and not have a tremendous amount of mass flow. On the other hand you should see very good performance at around -90C to -95C.

Might have the term incorrect for the aux cooler, but it was a pre-stage HX meant to max out the condenser and to help increase efficiency. Feel free to correct me if I am incorrect, been a while since I've thought about this stuff.
I call the first HXC (before the 1st phase sep) an Auxiliary Condenser since it is essentially just an economizer, or what I believe some of you also call an SLHX. All the following HXC's are considered to be Cascade Condensers. Hey gomeler I read your article on autocascades... very nice