I do not know anything about your protein and which purification methods have been tried so far. From the limited information you provided, I would say that this would NOT be the best method to separate the 2 proteins (20 kDa and 35 kDa). First of all, your 35 kDa protein will likely go through the 30 kDa cut-off membrane (unless it forms dimers or has a non-globular shape etc...). But of course if you have enough of the protein, you can try it... My suggestion is to try ion-exchange chromatography, if the two proteins are not related (i.e. the 20 kDa is not a degradation product of the 35 kDa one), they are likely to have different surface charge and separate / elute in different ioninc strength. Another possibility is size-exclusion chromatography (Superdex 75 or similar), but these two proteins are pretty close in size, so the separation will likely not be perfect (again unless for example the 35 kDa forms dimers/oligomers; you will get fractions more rich in one of them and fractions more rich in the other, but neither completely pure...). But it all depends, how much of the "contaminant" protein is present, how much of the desired protein you have to start with and so on... Also, if you everything else has failed and you want to try the concentrators for purification in the end, you might be more lucky with 10 kDa cut-off, the 20 kDa protein might go through, while the 35 kDa might not. Also keep in mind, that if you use the concentrators, you always lose a portion of your protein - it gets "stuck" to the membrane. Hope this helps...
a 30K concentrator is usually designed to retain proteins larger than 30 kDa with an efficiancy of more than 90%. However I have the experience that proteins of about half the mass are also well retained. In your case, you will always find proteins down to 15 kDa with a 30K filter.
You could also test a 50K filter. Maybe the bands at 20 kDa are disappearing. However, I don't know how efficient your protein of interest ist retained by a 50 kDa filter.
Maybe, it would be better to polish your protein with size exclusion chromatography.
It also depends on your purpose. If you plan to do proteomics then run a low concentration (say 5% or 7.5%) SDS-PAGE maybe a easy way to do. Also, in this way you tend to have less contamination.
Concentrators are not a good method of purification. Typically for protein to pass through the concentrator it has to be 1/5 of the MW cut off. At best using a 30kDa MW cut off filter you could only filter through protein of 6kDa or less. If your contaminants are all lower molecular weight than this then it might work. However, this is not a good method of purification.
A size exclusion chromatography would serve you well for the separation of proteins, get a crude extract of the protein and load the crude extract onto the column. If HPLC is not available, you can make your own column using a burette and sephadex beads of appropriate size. Good luck!
If you prefer to concentrate your protein of interest which is 35KDa with a MW cut of 30KDa by ultracentrifugation(AMICON-ULTRA-Millipore) then I would say that the retention efficieny would be greater than 80%.
I personally think TFF(Tangential flow filtration) with a 30KDa cut off membrane would be a better bet in terms of retention efficieny or maybe in terms of maintaining the structural integrity of the protein of interest without too much degradation.
you cant purify proteins using concentrators or TFF casettes. the pore diameter which determines MW is based on a median value of the averages values of all the pores obtained using SEM. the best way is to use Size exclusion...if the proteins is 90% pure or use affinity purification if the protein has a tag
well, that depends how large the impurities are you want to get rid of?
you might be able to cut out most of the ~10kDA impurities with a 30k filter. For larger impurities you'd need a 50k-100k filter.
Remember this only works if all those proteins don't form multimeters. if your protein is a dimer, you can try a 50k filter to get rid of ~30k impurities.
You won't be able to purify to homogeneity with this method but might be able to cut out some impurities that would interfere with gel filtration or ion exchange.
So, yes, try. Try all of the methods available to you.
EDIT:
Also, could you try to verify that the ~20k bands are not your protein of interest that has degraded partially?
i don't have an answer to the above asked ques. but i am also facing a same problem. My protein of interest is also 35 kda and i am getting an impurity at 25 kda. this is after gel filtration chr. using superdex 10/300. and also please suggest hat how can one determine that the following impurity of 25 kda is d degraded product of our interest protein.