Stream splitter

[gerrie]

Hi,
Does anyone have any information on the best type (s) of stream splitters to purchase, doing work in ESI + using Quattro Ultima Triplequad.
Thanks
G

[MG]

A simple mixing tee and some PEEK tubing will do the job. Adjust the length of the waste tubing and then measure the flow to waste using a small graduated cylinder, until you get the split ratio you need.

[gerrie]

MG, thanks for letting me know how to do this but going to try the smaller column trick since using a gradient and back pressure may affect split ratio throughout the run. Seems to make sense to me??

[MG]

OK, I assumed you just wanted to split flow to your MS. For pre-column splitting, you may need a commerical product, with which I'm not familiar. If you use the poor-man's technique I described, you may need to put a spare column or very small i.d. tubing on your waste line, depending on the split ratio you need. I don't know if split ratio would change throughout the gradient, but this could be measured.

[adam]

Actually what you describe as "the poor man's technique" seems to me to be the best approach in many situations (though not necasarily the one that started this thread). I think many people automatically go to a smaller ID column when they want to do mass spec work. But in my opinion the best way to do mass spec work is to stick with a standard ID column (like 4 or 4.6 mm) then split the flow post column so that only a certain amount goes to the mass spec. It is only when the quantity of sample is limited that a smaller ID column is the best way to go.

In order to accomplish this type of splitting without adding too much dead volume to the system, you would probably need to adjust the length and/or ID of the 2 pieces of tubing coming of the tee (as you have described)

[MG]

Adam, why do you think standard columns and split flow are the best way? There are several reasons I avoid it:

- More solvent is consumed.

- ESI needle is more likely to clog.

- A partial clog in or near the ESI source or waste line will change the split ratio, resulting in changes in response.

- Reduced sensitivity.

[tom jupille]

I'm at Pittsburgh Conference this week and I just came across an active splitter/flow controller from ASI (Analytical Scientific Instruments). They didn't have any printed literature on it (new product) but the test data they showed me looked good. Their phone number is (800) 379-5221. Web site is http://www.hplc-asi.com.

Just to clarify: I've known these guys for a long time (and worked with some of them), but I don't have any hands-on experience with this particular product.

[adam]

MG

I will try to elaborate. This will get pretty convoluted.

It is very common in books and papers to see the statement made that "smaller ID columns give better sensitivity". But this is actually not the case in most situations. What people forget is that the maximum injection volume is proportional to the square of the column diameter. So in situations where there is plenty of sample available, which is usually the case (it's only in certain specialized applications like proteomics work where the analyst has a limited volume of sample to work with) one would inject the maximum amount in all cases. So when we go to a smaller ID column it is true that there is less dilution of what we inject, but the part that people forget is that the amount we can inject is also reduced, by the square of the column diameter. And if you do the math, it winds up being a complete wash. Exactly the same sensitivity should be obtained, in theory.

Just to make life even more complicated, all of the above assumes a concentration sensitive detector. If we have a mass sensitive detector than reducing the diameter of the column does nothing for us - even if we do have a fixed limited volume of sample - for the simple reason that the sensitivity of such a detector is a function of the absolute mass injected, not concentration. But, of course, with a mass spec detector we need to cut the flow rate down for other reasons.

So my original point was that if some one wanted to do mass spec work, and we have the more common situation that there is plenty of sample to inject - then from a sensitivity standpoint - it is just as well to use a standard ID column and split off some fraction x of the post column flow (to waste), then to reduce the column ID so as to achieve the same flow rate reduction, and then reduce the injection volume by the square of whatever factor we reduce the column ID.

So far I'v argued that these methods are equivalent: from a sensitivity standpoint. So if they're equivalent, why do I advocate my approach. For the simple reason that if you go to a smaller ID column you get much (!!) more problems with respect to system dead volume. Most of us have experienced this even with 4.6 mm columns. You go to run the analysis, and there's a big fat peak staring back at you (or tailing). I find this type of problem particularly difficult to troubleshoot. It's is very difficult trying to figure out where the dead volume in the system may be. With smaller ID columns these problems are so much worse.

There's no question that my approach does consume more solvent, but it's a tradeoff I prefer to live with: especially in a QC environment.

Post note: There are other ways to save solvent. One is to shorten the analysis time. With isocratic analysis you can even drop the waste line directly right back into the mobile phase (!!) (a cyclic system). They make recylers for this purpose, but in theory it should work even without a recycler as long as the mobile phase as being mixed. It works under these conditions because the analytes that get into the mobile phase are not delivered to the column as a plug, but as a continuos stream. Hence no peak, just a baseline which elevates slightly with each run.

As to the ESI needle clogging, I disagree that this should be a problem, because either way you do it, the mass spec sees the same thing.

Anyway that's my take on it. I hope that made sense.
Adam

[MG]

Hey Adam, I agree with much of what you say, with respect to the sensitivity / flow rate / injection volume issue. Scale the injection volume, and there is no sensitivity advantage. I am actually glad to hear someone else say this.

I guess I am often sample limited. Not to the extent that the proteomics guys are, but I might want to inject a sample a few times to get statistics on it, or I might want to reinject a check standard many times out of the same vial rather than having multiple vials. I can inject 4.7 times less volume with a 2.1mm column versus with a 4.6mm, and get the same sensitivity.

On the dead volume issue, again I feel your pain. I've seen LC setups that ship with mass specs, that are plumbed with too much dead volume to use with 2mm columns. I run 0.005'' i.d. tubing everywhere between the pump and detector, when possible.

If someone came to me and told me that I should run 1mm or 0.32mm columns, I would give many of the same reasons you gave for not doing it, so I guess it is a matter of degree. But, if I had an older ion source that was very inefficient at desolvating the droplets, and a 2mm column required a splitter, then I might scale down to 1mm.

Which leads me back to the splitter issue and clogging, where I must disagree. I will assume for the sake of discussion that we are using ordinary splitters, not active splitters. The mass spec sees the same thing when there's no clog. When a clog starts to develop, the two situations are different. As a clog begins to form, there will be a pressure drop across the point of the clog. As the flow at that point becomes more and more restricted, the pressure drop continues to increase, up to some maximum. In the split-flow situation, the maximum pressure drop experienced, even with complete plugging, will simply be the backpressure from your waste line (perhaps a few hundred PSI at most). In the non-split situation, the pressure drop at the clog point will increase much more rapidly as it approaches total plugging, up to the maximum pressure specified in your method (perhaps 4000-5000 psi). In this situation, the clog is either going to be short-lived, or it's going to shut your system down.

Now imagine all the times you've ever had an ESI needle clog. Some of them could be "blown-out" with maybe 1000psi or less from your pump LC pump. I'll call these "minor" clogs. Some were indeed bad enough that even 5000psi could not blow them out (time replace the needle or some tubing). I'll call these "major" clogs.

In the non-split situation, only the major clogs will shut you down. The minor clogs will either persist without shutting you down, or they'll be "blown-out" during the course of your run and be transient events, only causing a problem if they happen at the wrong time. And, a persistent minor clog will not change the flow rate that your source sees for more than a second or two.

In the split-flow situation, even minor clogs will shut you down, or at the very least change the flow rate that your source is seeing. And if a partial clog causes your source to see 20uL/min instead of 200uL/min, you won't have any immediate warning of it.

For qualitative ID, I think splitters are fine, and sometimes will use one rather than scaling down an existing method. For quantitation, I am leery of using them.

[adam]

Fair enough. I will defer to your greater knowledge on the splitter issue. I haven't really had any hands on experience with that.

Cheers