UHPLC TIP: Reducing the Column Temperature to Offset Frictional Heating Effects (Causing Poor Resolution)

HPLC column temperature is a critical variable that we adjust and optimize during method development. We use it as a variable during the method development process to improve solubility, optimize peak shape and increase resolution. Once established, it must be carefully controlled during the method analysis to provide reliable and reproducible analysis results. Change the column temperature and you may also change the results obtained. This is a fundamental method development tool and must not be forgotten.

If you are developing a new UHPLC method OR perhaps scaling an HPLC method to utilize 2.5 micron or smaller support particles, then you may observe a loss of resolution or poor peak shape in the new method. There are many reasons why this may occur, and the most common ones relate to not optimizing all of the method parameters correctly when scaling the method (e.g. dwell volume too large, flow cell volume too large, injection volume too large, sample rate too slow, flow rate not optimized, mobile phase composition changes not in scale with the gradient...). But there is another reason...

Resolution may be reduced or lost when all of the initial scaling and instrument set-up parameters are optimized. What is the most likely reason for this? In many cases the use of substantially higher flow rates (relative to linear flow rates) and the use of smaller diameter particles results in much higher backpressures (you may recall that if you halve the particle size, the backpressure increases 4x). The resulting backpressure might be 2, 3 or even 4 times higher than observed in the original method. While these higher backpressures were well within the operating parameters of the HPLC system used, the results obtained were poor. The possible cause? The much higher backpressure increased the amount of frictional heating inside the column, raising the actual analysis method temperature and changing the separation conditions. 

Pushing mobile phase (liquid) through a chromatography column generates heat and pressure. The heat generated increases the actual temperature of the column and reduces the viscosity of the fluid. In conventional columns (i.e. 4.6 x 150 mm, 5u) at 1.00 ml/min, this heating effect is minimal, but at much greater column pressures, > 400 bars, the frictional effects may be substantial. These types of very high pressures may be seen with methods which utilize columns containing the smallest particles (1.9 to 2.5 micron). Enough to change the temperature in the column by several degrees (e.g. >5 degrees C) and result in different method conditions. So, what can you do about this? The most direct way to address the problem is to run the same method at a lower temperature (perhaps decrease by 5 C to start with). This will slightly raise the backpressure (lower temperature equals higher viscosity), but it should cool the column and restore the original temperature conditions used. Additionally, we suggest that you always start column equilibration using a flow ramp to gradually increase the flow over time and reduce the overall heating effect and resulting "shock" placed on the column. An initial delay at equilibration may help reduce these effects (gradually ramp up to the regular flow rate and hold). You may need to try several temperatures and this may be easiest to do if your HPLC has a column compartment with heating and COOLING capabilities. Optimizing the temperature and internal pressures may increase the column lifetime and result in better overall data reproducibility.

Determine the HPLC System Dwell Volume (Gradient Delay Volume)

Note: The total HPLC gradient system dwell volume is different than the HPLC column’s void volume. Two different terms for two very different measurements.

When we perform gradient HPLC analysis, the mobile phase composition is changed over a period of time. The mobile phase is mixed in real time by the pump(s), mixer and/or valves, then transported to the injector and finally, on to the head of the HPLC column. The total volume of liquid contained between where the mobile phase is mixed and the head of the column helps us determine when the newly mixed solution arrives at the column head (it is not instantaneous). This delay is often referred to as the gradient delay time (or delay volume) and its value will vary for different HPLC systems due mainly to differences in tubing dimensions used, pumping system type and the design of the flow path. 

For example: If the system dwell volume is found to be 1 ml and the flow rate used is 1.000 ml/min, then the gradient delay time is one minute. 

So how do we know what the system dwell volume or gradient delay volume is? Well, we measure it of course!

Measure the ‘System Dwell Volume’ (aka: Gradient Delay Volume)*:

(1) REMOVE any HPLC column(s) and install a Zero Dead Volume Union (*ZDV) or a restriction capillary of know volume in its place.

(2) Prepare Two Different Mobile phase solutions:

Bottle ‘A’: HPLC grade Methanol (MeOH).

Bottle ‘B’: HPLC grade Methanol with 0.1% acetone added (v/v).

(3) Set your UV/VIS detector to 265 nm (8 nm Bandwidth, Reference OFF).

(4) Program a suitable system flow rate and create a simple Gradient Method (linear change) which starts at 0.0 minutes with 100% ‘A’ (HPLC grade Methanol) and 0% B (HPLC grade Methanol with 0.1% acetone added) and runs to 0% ‘A’ and 100% ‘B’ for about 10.0 minutes (actual times used will depend on your selected flow rate).

(5) Flush and degas both solutions, ‘B’ first, then ‘A’ through the system until you get a nice clean, flat baseline. Make sure their is enough backpressure on the pump (>40 bars) to obtain a stable signal (use a restrictor or back-pressure regulator if needed).

(6) No injection should occur during this method.

(7) Start the method (RUN) and observe the 265 nm signal over time. At some point you should observe the signal begin to rise. When you see this signal change occur, the acetone has finally made it from the pump head to the detector’s flow cell. Make note of the time this occurs. 

Using the known flow rate and observed signal change time, you can now estimate the total system dwell volume. 

Example: If you observe the signal start to rise steeply at 2.00 minutes and your flow rate was 1.000 ml/min. Your system dwell volume would be 2.000 mls. 

A more accurate system dwell volume value can be obtained by next running the same method with an injection of acetone (e.g. 1 ul) and noting the time at which the injection peak is first seen. That will give you the time it takes the sample (and therefore the volume needed) to go from the injector to the flow cell. If you subtract this time off the system dwell time you recorded in the last test, you will have the actual measured time from the pump head (or proportioning valve) to the head of the column (vs the flow cell). Normally the volume contained in this tubing and flow cell are very small relative to the volume in the rest of the system, so we can ignore them. However, when using some of the very low volume columns (e.g. 2.1 x 50 mm), the volume contained in these areas can become significant so when appropriate, we need to be aware of them.

Failure to take into account changes in HPLC system dwell volumes can result in methods which no longer work or provide different results. This is because the gradient rate change you program in your method may not allow enough time for the new mobile phase composition to reach and flow all the way through the column in the time that you have programmed. A common mistake we see is when users forget to adjust the gradient profile when changing column dimensions or program changes using too fast a time.

BTW: One common trick we use to improve compatibility between systems which have different dwell volumes is to include an initial (time 0.0)  isocratic hold-time into the start of each method. If all systems used have system delay volumes under 3 mls, then add a 3 minute isocratic hold time at the start of each method (if 1.000 ml/min flow rates are used), before any gradient starts. While not the best way to deal with the issue, this type of “cheat” can make it possible to quickly adapt a method for use on several different system types.

*Note: This is a generic method to determine the system dwell volume or gradient delay volume. Detector signal buffering and flow cell volume also adds to the delay and in some cases, must also be accounted for too. There are many other methods which can be used for this determination as well. This proposed example serves to illustrate the concept only.

Common HPLC Calculations:

Capacity Factor / Retention Factor / Capacity Ratio:  k1 (K Prime)

k1 = T(R) - T(0) / T(0)
where T(R) equals the retention time of the peak in minutes and T(0) is
the retention time of an unretained peak. *For chromatography to take place, K Prime must be > 1.00 and for most modes of chromatography, should be greater than 1.5 or 2.0 for all samples !


Tailing Factor: USP: 't'

t = W(5.0)/tw/2

where tw equals the distance between peak front and T(R) at 5% of peak height units. W(5.0) equals peak width at 5% height, in minutes.


Theoretical Plates: USP and ASTM, 'N'

N = 5.54 x (T(R)/W(50))2          

Assumes width at peak half height (50)
* More info can be found at this link.


Resolution: USP, ASTM, 'R'

R = (T(R)(b)-T(R)(a)) x 2.35/(W(50)(b) + W(50)(a))/2

Assumes width at half height (50%) with peaks (a) and (b).

*Notes: Visually, "Baseline" resolution is R = 1.5. Your goal should be R = or > 2.0. ** R of 1.5 provides 99.8% separation which means you cannot accurately quantify a 0.1% impurity so develop the method to have a resolution value of at least 2.0.


Note: The appropriate formula(s) for use with your samples may depend on which of the many pharmaceutical guidelines and regulations apply in your country. Always consult the appropriate guidelines.

Chiral HPLC and SFC Column Screening Strategies for Method Development:

We are experts in chiral separations and have learned a great deal about how to efficiently and quickly identify the best conditions to resolve a chiral sample by HPLC and SFC. You should always have a clear strategy in mind when developing an automated HPLC or SFC chiral method development screening system to separate racemic samples. Here are a few key points to focus on.

(1)  Make sure the chiral sample is as “chemically” pure as possible before you start. By “chemically” pure I mean it should only contain the two complementary enantiomers and not other impurities, starting materials or chemicals which might interfere with the identification or separation of the racemate. Often, many of the intermediate chemicals used in the synthesis of a compound have similar absorbance or retention characteristics. When this happens, you can be fooled into thinking you have resolved your chiral component apart when in fact you have simply resolved a non-enantiomer apart from the racemate. Chiral columns are very poor at discriminating the racemate from other non-chiral species. As such, you may find that the impurities or other components make it difficult to determine the actual HPLC or SFC purity of your chiral sample. Try and remember this phrase: “The chiral purity of a sample is only as good as the chemical purity”. So start with as clean a sample as possible when developing a chiral method.

(2)  Column choices are very important when using a fully automated Chiral Screening System such as the LC Spiderling™ Column Selection System. Here are some popular questions we are asked on this topic.

How do you know how many and which types of columns to include in your chiral screening system?

Keep the goal of creating a "screening system" in mind. Don't lose sight of this basic strategy. You want to select the smallest number of different columns which have the widest specificity for your expected sample types. Often five (5) to nine (9) different types of chiral columns will do the job. Screening more columns than that is often a waste of time. Leave a “test” position in your screening system so you can evaluate new columns all of the time. The key is to identify the best ones first.

Normal or Reverse Phase Columns?

Most chiral screening is performed in the normal phase mode (NP provides easy solvent removal for scale-up and there are a wide range of quality columns available which can be used to generate useful data). You can mix reverse phase and normal phase columns in the same system as long as you incorporate a bypass and flush step between the different methods to wash out the old solvent and bring in the new solvent safely. For this brief discussion, we will assume you have a dedicated normal or reversed phase screening system. 

How many different mobile phase systems should I use?

Quick answer is as few as possible. The concept of using a screening system is to quickly identify the column (1^st) and mobile phase (2^nd) which baseline resolves the racemate. Select three or four different mobile phase types, with and without modifier systems, which span the range of polarity needed to increase your chance of retaining the sample on the column, but for no longer than 30 minutes. Keep it simple! The goal is to retain the sample, hold, then elute it.

Which types of chiral columns are best?

Well, if you ask the different column suppliers this question, then they will most likely answer that the columns “they sell” are the best. At last count, there are over two hundred different chiral columns advertised on the market today. Most are advertised to be the 'best', but in fact they all can not be… In reality, you should evaluate as many different kinds as possible with your own samples to determine which the “best” are. Do not be fooled by examples of the column separating out very simple compounds such as racemic trans-Stibene oxide as this is one of the easiest compounds to separate even if 90% of the chiral stationary phase is missing! Consider also if the column type can separate compounds without the use of fancy modifiers or complex mobile phase mixtures. Simple is better and usually more easily reproduced too. Some of the most commonly used types of chiral columns are the: cellulose/amylase, Protein, Pirkle type and cyclodextrin based columns. All these columns have different preferred mobile phase choices (Most protein columns are run in reverse phase, while all of the others mentioned have versions which can be run in either normal or reverse phase). You should consult with the appropriate manufacturer about how to best use these columns. Do not necessarily select a column because the column has been “reported in the literature to be used by the largest number of people”. Who cares how many people used a particular column. This is not a scientifically valid argument that the column is useful. We routinely read published papers which describe a "novel" chiral method run on a specific column which clearly shows worse results than could have been obtained with another commercial column using a simpler mobile phase. Many try and force a column to resolve a sample apart because it is the only chiral column they have. The purpose seems to be directed at publishing a paper and not at developing high quality chiral HPLC or SFC methods at all. Just because someone tried it before does not mean that their method or column choice was a good one. Remember, very few chromatographers have practical experience developing CHIRAL methods. It takes special training to be successful at it. Invest some time learning about and evaluating the different chiral column types with your own samples to find out which ones are most applicable. Just as with achiral analysis, make sure the sample is fully soluble in the selected mobile phase before analysis. Select a chiral column based on your own scientific evaluation and testing. Start with a "full sized" columns to maximize the amount of stationary phase the sample comes in contact with. 

 

A Note about AVOIDING "Short Length" Chiral columns: We do not use or recommend any short "scouting" columns. Using SHORT columns will often result in you miss-identifying a column type that actually works. You will not see retention when in fact you would have if you used a standard sized column. Please don't make this novice mistake with chiral columns. Use chiral columns that are LONG, not short for screening (the goal is to maximize the amount of support the sample comes in contact with). The use of short chiral columns in HPLC / SFC column screening is often a waste of your money and time. If you want to identify which chiral columns will resolve your samples, stay away from short columns.

Can you provide some free advice as to which columns are the “best”?

OK, we have tried them all (but not for your samples), but here are two of our favorites (in no order):

The Pirkle based Whelk-O 1 (and/or Whelk-O 2) is the only Pirkle based column we have ever used which can produce a significant quantity of racemic separations using simple mobile phase systems. No other Pirkle based chiral column has ever proven to be as good as this one in real world chiral pharmaceutical drug method development. Every other one tried has disappointed us, but this one has been responsible for a number of successful separations in normal and reversed phase modes.

The coated polysaccharide chiral stationary phase made by Daicel, known as "Chiralcel OD" (OD-H) is one of the best chiral columns on the market. Note that these are the non-covalently bound coated versions of the column. This support type has a broad range of selectivity not seen in any other types of columns available. At this time, the more stable covalently bound versions are also very good, but just do not measure up to the high success rate this one has. BTW: It is normal to see different results between the covalently bound and coated versions of the same column. They are completely different supports and if budget allows, you can have both types of columns available for screening.

Determination of HPLC Column Void Volume / Dead Volume, Dead Time (T zero):

Column Hold-up Volume, Column Dead Time or 'Column Void Volume' (the preferred name) are all different terms we apply to find the internal volume of a packed column  (divided by the flow rate and usually expressed in minutes for the Column Void Time). You must know what this value is BEFORE starting to run an HPLC method or perform liquid chromatography. The value for column void volume changes for different column dimensions and different column support types (e.g. fully porous, superficially porous etc) .

Are you peaks or samples eluting at or near the column void volume? If so, for most modes of chromatography, this implies that no chromatography has taken place and no HPLC method has been developed (SEC/GPC separate based on hydrodynamic volume, so elution at or near the column volume means the sample(s) were excluded from the column). Individuals with little to no chromatography training or experience often make this mistake and create methods which show poor retention. Make sure your methods are designed to retain each sample for a long enough time period on the column (K prime). How do you know how long is long enough? Start by estimating the Column Void Volume (use our table or calculate it for an estimate) then, calculate the K prime value for your sample. The K prime for each peak should be at least 1.5 (>2.0 is the accepted standard for most regulatory authorities) for the method to be useful and selective. *A more accurate value of column void volume will be found by measuring the void volume of your column (please read on).

Knowing the Column Void Volume and the Flow Rate used allows you to calculate the Column Void Time (which is the most useful initial value). Determining  the column void time or T0 ("Tee Zero" as we call it), is necessary to find other important chromatography values such as: the Resolution, Separation Factor and Capacity Factor (K prime aka: "K1") in a chromatography separation. Ideally, it is measured by injecting a sample which is unretained by the column & mobile phase (it passes right through the column support with little to no interaction). It may also be easily estimated for most fully porous, spherical, bare or coated silica supports if you know a few physical specifications of the column and media used. You should first estimate it, then measure it (the two values should be close, +/- 15%). Note: A practical "tip". You can also estimate T0 by noting when the small injector valve switching peak ('blip') appears on the baseline. It results from the change from switching the injection valve from the "load" to "inject" positions. Use a low UV wavelength to observe this deflection on the baseline.

Here is short list of typical HPLC column dimensions and their associated estimated void volumes for fully porous silica supports. At a flow rate of 1.000 ml/min these values would also be the same as the void time in minutes.

COLUMN DIMENSIONS (I.D. x Length (mm))                 VOID VOLUME (ml)

                         2.1 x  50                                                                  0.12
                         2.1 x 100                                                                 0.24
                         2.1 x 150                                                                 0.37
                         2.1 x 250                                                                 0.61
                         2.1 x 300                                                                 0.73

                         4.6 x  50                                                                  0.58
                         4.6 x 100                                                                 1.16
                         4.6 x 150                                                                 1.75
                         4.6 x 250                                                                 2.90
                         4.6 x 300                                                                 3.49

                       10.0 x 100                                                                 5.50
                       10.0 x 150                                                                 8.25
                       10.0 x 250                                                               13.75
                       10.0 x 300                                                               16.49

•  Column Void Volume Equation for Std Sized, FULLY Porous Supports:

Column Volume (ul) = (d^2 *Pi * L * 0.7) / 4 ;

•  Column Void Volume Equation for SUPERFICIALLY Porous Supports (e.g. Fused-Core, Core-Shell etc):

Column Volume (ul) = (d^2 *Pi * L * 0.5) / 4 .

   Note: Column Diameter & Length are in mm. Volumes are estimates (always measure to find the actual value).

[Note: All you need is the column's length and ID to estimate it. For most fully porous supports, use a 'Pore Volume' value of 0.70 in the above equation. This is the most commonly measures pore volume found for non-encapped, fully porous spherical bare silica support (please check with the manufacturer for the actual value of your support). For superficially porous supports, use a value of 0.50. Estimating the value will often get you close to the measured value, but due to the unique chemistries used to prepare supports, it is only an approximation.

Always measure the actual void volume of your specific HPLC column with a compound which is unretained by your column. For RP applications which utilize at least 20% organic, Uracil or Thiourea are often used, but some inorganic salts (e.g. sodium nitrite and sodium nitrate) have also been shown to work as well. When determining the "Column Void Volume", you are really measuring the void volume of the column plus any extra-column volume from the injection volume plus all lines connecting the injection to the column and the column to the flow cell. Note: This is very different from the "System Dwell Volume" which includes the volume from the pump (or gradient valve) to the column head.

A more detailed version of this table with other common HPLC Column Sizes and Tubing Volumes for capillary lines are available at the following links (Link #1) or (Link #2).