Switching Columns

[MarcWJHetu]

Hello, this is my first time posting so hopefully I can get some good information.

I am trying to switch form a 250 mm Hypersil Gold column to a 100 mm Poroshell 120 (both packed with C18) and am using an Agilent 1100 series HPLC. I'm somewhat new to making large changes to methods and I'm wondering what some of the pitfalls are when going form a large column to a smaller one.

My questions include but are not limited to:

1) How important is it to change my gradient?

2) How much smaller should my injection volume be?

3) Should the concentration of my compound change?

4) How should the pressure be adjusted for a smaller column?

I usually use 2 different kinds of methods when using my HPLC, one for metal complexes and another for natural product extracts. The metal complex method has a runtime of 40 minutes + 20 minute wash period (10 before and 10 after each sample) with the final peak coming off the column at around 32-33 minutes. The natural product extracts have a 60 minute runtime with the same wash procedure, and the last compounds to come off happen around 54-55 minutes. The main reason I'm trying to switch columns is so that I can have faster/shorter runtimes so I can run more than 6 samples in a day.

Any information is greatly appreciated. If you have insight one for one method and not the other, I'm completely open to making the smaller column dedicated to the metal complexes.

[tom jupille]

There are two aspects to this kind of a switch:
- physical parameters (easy to adjust)
- chemical parameters (difficult to adjust, unless you are very lucky).

To take the easy first, assuming you keep the same flow rate, simply scale all your gradient times in proportion to the column volume (i.e., length x diameter^2 -- if the diameters are the same, then it's simply 40% in your case). In principle, the same thing goes for injection volume, but the solid-core packings are less tolerant of overload than are the fully-porous materials, so you may have to do a bit of experimentation (halve and double the injection volume to see what, if any, effect). Pressure should not be much of a problem; you will be using smaller particles (increased pressure) in a shorter column (decreased pressure); those should mostly cancel out so unless you were close to the upper limit on your system, you should be OK.

The hard part is the chemistry. All C18 columns are *not* equivalent. Even if you were switching to a different brand of fully-porous material, there is a good chance that the method would have to be re-developed to account for the chemistry differences. If you're lucky, the selectivity will be not too different and the changes will be minor. If you're unlucky, you may have to change the gradient profile, organic solvent, pH, buffer concentration, and/or temperature. The only way to know in a specific case is to try.

If your instrument can tolerate "UHPLC" columns (i.e., it has a high enough pressure capability and a small enough internal volume) an easier approach would be to use a smaller-particle version of the Hypersil Gold. In that case, the chemistry will (hopefully) be the same, and you can make straightforward adjustments to flow rate and gradient times. Thermo has an on-line calculator to help with the adjustments at:
http://www.hplctransfer.com/
or you can download an adjustment spreadsheet (free) from our web site:
http://www.lcresources.com/sandbox/UHPLC-HPLC.xls

[Gerhard Kratz]

I fully agree with Tom.
May I ask you WHY you want to switch to another column?

[MarcWJHetu]

I'm looking for ways to reduce the length of time it takes to run one of our metal complexes (I'm more concerned with improving that method). Currently our complexes come off the column between 20-35 minutes depending on the kind of complex, factoring in the wash periods, that's 1 hour per sample. Also, making a shorter analysis time will allow me to use less solvent per sample.

[bunnahabhain]

A Follow-up to Gerhard Kratz:
Why do you want to switch to another column? Hypersil Gold is available in all imagineable dimensions and particle sizes. So you just have to scale down the column using the abovementioned tools. Try to keep flow rate > 0.5 mL/min if you use a quarternary pump. Binary pumps should be OK down to 0.25 mL/min (my experience, YMMV).

Kind regards,
Jörg

[DJ]

What kind of resolution are you getting between your "critical pair(s)"? Have you tried optimizing your existing system at all for throughput?


Tell us more about your methods
What are "A" "B"?
Your gradient program?
Flow rate?
Column particle diameter?

[MarcWJHetu]

A Follow-up to Gerhard Kratz:
Why do you want to switch to another column? Hypersil Gold is available in all imagineable dimensions and particle sizes. So you just have to scale down the column using the abovementioned tools. Try to keep flow rate > 0.5 mL/min if you use a quarternary pump. Binary pumps should be OK down to 0.25 mL/min (my experience, YMMV).

Kind regards,
Jörg

The reason we don't just switch to another size of Hypersil Gold is because of money. The lab I work in is at a small university with a large undergraduate program, so money can be an issue at times. The reason for the switch to the Poroshell 120 specifically is because it was given to us as a sample by a very generous sales rep.

What kind of resolution are you getting between your "critical pair(s)"? Have you tried optimizing your existing system at all for throughput?


Tell us more about your methods
What are "A" "B"?
Your gradient program?
Flow rate?
Column particle diameter?

When it comes to resolution, I have nice sharp peaks (Hypersil GOLD) depending on how pure the synthesis part of our lab can get the sample (for metal complexes).

For methods, this is our current set up.

A = 0.1% formic acid in H2O
B = 0.1% formic acid in MeOH


GRADIENT
The gradient for metal complexes is:

Time(min)
Hypersil GOLD-----A(%)--------B(%)
---0.00----------98----------2
---2.00----------95----------5
---5.00----------70----------30
---15.00---------70----------30
---20.00---------40----------60
---30.00---------5-----------95
---35.00---------98----------2
---40.00---------98----------2

The same gradient adjusted for the PoroShell is as follows (taken from the advice given by tom jupille - Thank you ).

Time (min)
PoroShell 120--------A(%)--------B(%)
---0.00-------------98---------2
---0.80-------------95---------5
---2.00-------------70---------30
---6.00-------------70---------30
---8.00-------------40---------60
---12.00------------5----------95
---14.00------------98---------2
---16.00------------98---------2

I have since tried this method on the PoroShell 120 but found my standards coming off too late. The standard I made is a 100µM solution (in DMSO) of some ligands commonly used in our lab, incase anyone is wondering. Since the PoroShell 120 gradient above let the standards come off too late in the run, I changed the gradient to that seen below

Time (min)
PoroShell 120--------A(%)--------B(%)
---0.00-------------98---------2
---0.80-------------90---------10
---2.00-------------70---------30
---6.00-------------40---------60
---8.00-------------20---------80
---12.00------------5----------95
---14.00------------98---------2
---16.00------------98---------2

I ran a photoejecting experiment (which is what I do with the HPLC and our metal complexes) using the method immediately above. The placement of the peaks before, during, and after ejection were relatively the same as using the Hypersil GOLD column. I now have a problem with peak tails (I believe that is the proper term, when the peak trails at the end). I only just got the result yesterday and I currently trying to correct for it but tips are welcome.

FLOW RATE
The flow rate for the Hypersil GOLD has always been set at 1.00 mL/min (pressure stayed at ~120 bar while on standby, and ~170 bar when running a sample) , and I have been setting the PoroShell 120 at 0.500 mL/min (pressure is ~200 bar on standby and hovers ~300 bar while running a sample). The subject of pressure begs a question I will make in a different post.

PARTICLE SIZE
The particle size for the columns used are as follows:
Hypersil GOLD = 5 µm
PoroShell 120 = 2.7 µm

I apologize for any mistakes or explanations that may not be clear. Let me know and I can try to elaborate and/or explain differently.

Thank you.

[DJ]

"Metal complex"
Can you provide more details?

When you say "When it comes to resolution, I have nice sharp peaks (Hypersil GOLD) depending on how pure the synthesis part of our lab can get the sample (for metal complexes)."

Do you mean there are impurities that may or may not co-elute? So in other words, you are not achieving baseline resolution of peaks corresponding to target vs synthesis impurities?

What is the goal of your analysis? Usually but not always, we use analytical HPLC to determine relative abundance of sample components. If your peak width(s) are changing "depending on purity" it sounds like you're not getting good resolution at all. Peak width in gradient HPLC is roughly constant for related solutes. If you're seeing wider peaks, it probably means something(s) underneath.

How complex is your sample?

What type of HPLC are you using? Most have stainless steel components. Just something to keep your guard up for- your ODS columns will sometimes collect transition metals that are "endogenous" if you will (leached from the pump) and also "exogenous" (present in your mobile phase). For this reason it is never a bad idea to passivate the pump with 6 M Nitric acid every 6 months or so, depending on, eg., use of corrosive mobile phases (chloride salts, especially in acidic solvents are notorious for corroding stainless.). If the HPLC is "communal" then you have to assume the worst-- I would not do another run until you passivate the pump.

The frits of the column have, by far, the largest surface area of SS in the entire system. When these become corroded, it can sometimes wreak havoc on a separation.

[DJ]

I have since tried this method on the PoroShell 120 but found my standards coming off too late. The standard I made is a 100µM solution (in DMSO) of some ligands commonly used in our lab, incase anyone is wondering. Since the PoroShell 120 gradient above let the standards come off too late in the run, I changed the gradient to that seen below

Generally speaking, it is advisable to try and load the sample in your starting mobile phase . DMSO can cause problems related to viscosity and "strength" as a solvent. I would dilute it with a few volumes of starting mobile phase . If your sample is not soluble in highly aqueous conditions (such as those used at the start of your gradient), you risk precipitation of your solute once the DMSO is "diluted" upon injection. A partially blocked frit can explain anomalous peak shapes and higher than normal back pressure.

My knee-jerk impression is that your core-shell column back pressure is a little high.

Regarding elution times after the end of the linear gradient- I would consider increasing the temperature or else using a stronger organic. The reason MeCN is popular is because it has a lower viscosity compared to aqueous methanol and is also a stronger solvent (so less "B" is required for elution). You may consider giving it a whirl. Since the back pressure is much lower than MeOH /water, you can run much higher flow rates for the same back pressure if you wished to optimize throughput

[MarcWJHetu]

"Metal complex"
Can you provide more details?

When you say "When it comes to resolution, I have nice sharp peaks (Hypersil GOLD) depending on how pure the synthesis part of our lab can get the sample (for metal complexes)."

Do you mean there are impurities that may or may not co-elute? So in other words, you are not achieving baseline resolution of peaks corresponding to target vs synthesis impurities?

What is the goal of your analysis? Usually but not always, we use analytical HPLC to determine relative abundance of sample components. If your peak width(s) are changing "depending on purity" it sounds like you're not getting good resolution at all. Peak width in gradient HPLC is roughly constant for related solutes. If you're seeing wider peaks, it probably means something(s) underneath.

How complex is your sample?

What type of HPLC are you using? Most have stainless steel components. Just something to keep your guard up for- your ODS columns will sometimes collect transition metals that are "endogenous" if you will (leached from the pump) and also "exogenous" (present in your mobile phase). For this reason it is never a bad idea to passivate the pump with 6 M Nitric acid every 6 months or so, depending on, eg., use of corrosive mobile phases (chloride salts, especially in acidic solvents are notorious for corroding stainless.). If the HPLC is "communal" then you have to assume the worst-- I would not do another run until you passivate the pump.

The frits of the column have, by far, the largest surface area of SS in the entire system. When these become corroded, it can sometimes wreak havoc on a separation.

Question 1:
Yes, sorry, I have been a little more than vague regarding the kind of metal complexes I work with. The lab I am in makes polypyridine complexes. We use an assortment of different ligands.

Very basic polypyridine structure here.
http://en.wikipedia.org/wiki/Polypyridine_complex

Question 2:
Yes, sometimes there are impurities in the samples I run. I suppose I miss spoke earlier. When the sample is pure, I generally have a near symmetric peak (I have never been able to get textbook peaks). When the sample is impure, at times, those impurities can affect the desired peak I want to see (generally if the elute close together). I don't know the reason why.

Question 3:
The main reason for my HPLC testing is photoejection analysis. After the complexes are exposed to light irradiation, one of the ligands may or may not come off. If it does, we like to track which one does come off. Sometimes we make stained samples (add methyl groups to specific places hoping it will slightly destabilize the complex, enough to make it more light sensitive but still kinetically stable), and we use the HPLC to check and make sure the desired ligand is ejected and visible on the trace. So we have a dark sample, 1/2 ejected sample and full ejected sample. Dark sample should have only the full complex, 1/2 will have the complex, photoproduct, and the ejected ligand should be somewhere on the trace as well, the full sample will have photoproduct and ligand.

Question 4:
The complexity of the sample can vary but they are predominantly 3 bi-dentate ligands around a metal.

Question 5:
As far as I know (meaning that is all I see without taking the whole thing apart) the HPLC we use is fitted with stainless steel components. It would be a good idea to properly clean the pump the whole HPLC. We use some samples that use chloride salts but we haven't used a main solvent with any. I'll talk to my boss about getting some acid for cleaning. We do not have a communal HPLC, though it is refurbished.

[DJ]

Cationic solutes are notorious for tailing on ODS columns due to silanol interaction. Run a sample with some neutral solutes (acetophenone, toluene). If these peaks tail, the problem is more than simple silanol interaction. Regeneration is possible, and there is no shortage of protocols, however, I have wasted unthinkable amounts of time (and $) trying to raise the dead. In these cases, I feel it's best to treat yourself to a new column.

I know what it's like to be on a budget. I thought Id have to set up a lemonade stand to subsidize research funds as a grad student (or else prepare my own ether from vodka and drain cleaner). I became proficient in the art finding DIY recipes for otherwise expensive kits (US patents). I suggest you Look on ebay. For <$100 you can find a brand new column. Make certain the seller specifies "new, in original box, un-opened". At this low of a price, it becomes more economical to toss the old column directly into the junk drawer than it does regenerating it (unless in a pinch).

regarding your metal complex analysis- what detection are you using? A PDA detector is nice. Tandem (UV, fluorescence) in series is even better. But firstly, unless you have "standards" (or proxies) representing expected degredation products, how can you be certain you're able to resolve the in-tact complex from one that is without a ligand?