Professional Documents
Culture Documents
Starting Out Right, Parti-Selecting The Tools: Feature Index Home Main Index
Starting Out Right, Parti-Selecting The Tools: Feature Index Home Main Index
updated 09/18/2000
LC-GC - December 1999
Feature Index - Home - Main Index
John W. Dolan
LC Troubleshooting Editor
One of the key factors in minimizing separation problems is to start out with a good method.
By using a good method, it is much easier to maintain operation within specifications and to
correct problems when they arise. For the next few months, “LC Troubleshooting” will focus
on the components that make up a good separation and discuss how to put them into
action. The staff at my company has been teaching chromatographers these skills in short
courses for more than 25 years. Over time, a fairly simple approach to method development
emerged - we use this approach in our courses and contract laboratory - and I will use it as
the core of this series.
If you were about to cook a gourmet dinner, you wouldn’t just walk into the kitchen and start
cooking. First you’d consult a recipe, check your supplies, and go shopping to obtain the
ingredients. You’d make sure the pots and pans were clean and the knives were sharp.
Sure, you may need to cut some corners and make some substitutions, but you would use
the right ingredients and follow the recipe to get the best results. The same approach holds
true for developing an LC method. You need specific tools and reagents to obtain the best
results. If you start developing a method without proper planning, you may get an adequate
separation but the likelihood of a robust method that is tolerant of small changes is less than
if you had used a systematic approach.
COLUMN SELECTION
For example, most separations performed today use reversed-phase columns, and C8 and
C18 phases are the most popular of these. Chromatography, however, is not just a
popularity contest. The reason why these two phases are so popular is that they obtain the
desired separation in most cases. Although some standard sample types don’t point to
these columns as the first choice (see Table I), the C8 or C18 columns usually offer the best
stationary phase with which to start. There’s not enough difference between the C8 and
C18 phase for me to strongly recommend one over the other - I prefer C8 for most
applications, but I’ll leave that choice up to you.
The production of low-metal, high-purity silica is perhaps the most important practical
column development in the last 15 years. Detailed descriptions of the silica surface are
available elsewhere (2,3). These publications describe the traditional chromatographic silica
as a heterogeneous, acidic surface. After attaching the bonded phase to the surface, the
resulting columns could be characterized by tailing peaks for basic analytes and often
lower-than-desired column-to-column reproducibility.
Recent advances in silica production have resulted in a nearly metal-free silica with a very
homogeneous distribution of low -acidity surface silanol groups. The practical result of these
developments is columns that are more stable, more reproducible, and produce much better
peak shapes. This improved silica often is called Type B silica. Because of the benefits of
these phases, most manufacturers prominently distinguish these with names touting their
superiority such as Inertsil (GL Sciences Inc., Tokyo, Japan), BDS (base-deactivated silica,
Hypersil Inc., Needham Heights, Massachusetts), Symmetry (Waters Corp., Milford,
Massachusetts), and YMC-Basic (YMC, Inc., Wilmington, North Carolina). Most column
manufacturers make a Type B silica column, so check with your favorite column supplier for
the one you should use. Because of the superiority of these phases, I strongly recommend
that you set aside your older, Type A phases and start every new method development
project with a Type B phase.
COLUMN SIZE
Another important choice is the column dimensions and packing particle diameter (dp ). The
most popular particle size is 5 -µm dp, but 3.0 - and 3.5-µm particles also are widely used.
Any of these particles are suitable for starters, but most chromatographers still prefer the 5 -
µm particles because of their long usage history. Smaller particles produce columns with
higher plate numbers as well as higher back pressure. The 3.0-µm particles are packed with
smaller frits that are prone to fouling, so some manufacturers produce 3.5-µm particles that
enable the use of traditional 2.0-µm frits, which are less susceptible to fouling.
I prefer the 150 mm X 4.6 mm column size because it generates a sufficiently high plate
number ( N) with 5-µm particles to obtain adequate separations with most samples. An
alternative configuration is the 75 mm X 4.6 mm, 3.5-µm dp column, which provides
separations similar to those of the 150-mm long, 5-µm dp column but in half the time. An
added benefit of either of these column configurations is that the flow rate can be set at 1.5-
2.0 mL/min with reasonable back pressures (for example, less than 2000 psi). Higher flow
rates mean shorter run times, and as the saying goes, “time is money.” Columns as long as
250 mm generate a few more plates, but the benefit is marginal ( N grows only with the
square root of the length increase) and the penalties are longer run times and higher back
pressures.
Some workers suggest 30- or 50-mm column lengths for initial screening, but these
columns often do not generate a sufficiently high plate number for a reasonable separation.
Another option is narrow-bore (2-mm i.d.) or microbore (<1-mm i.d.) columns. These
columns use less solvent, but during the method development stage they place
unnecessary demands on the LC system, such as requiring smaller sample volumes and
minimizing extracolumn plumbing and detector cell sizes. My preference for using any of
these short or narrow diameter columns is to wait until the fine-tuning stage at the end of
method development.
In other words, I’ll stick with my 150 mm X 4.6 mm, 5-µm dp column as the workhorse for
method development. My alternate is the 75 mm X 4.6 mm, 3.5 -µm dp column. Typically I
recommend running these columns with a 1.5-mL/min flow rate.
Another choice that can have a bearing on the success of the separation is the organic
solvent. Reversed-phase separations provide three choices: methanol, acetonitrile, and
tetrahydrofuran. Each solvent has unique selectivity advantages, but chromatographers
seldom can predict which solvent will be the best choice on this basis. So you must choose
the starting solvent based on other considerations.
Most of the work in my laboratory requires analyzing pharmaceutical compounds. Many of
these samples have very poor UV absorbance, so analysts often find themselves working
with detector settings of 220 nm or less. Because of its high background absorbance,
tetrahydrofuran is not very useful below about 240 nm. Although low concentrations of
methanol can be used at low wavelengths, gradients with methanol usually drift off scale at
wavelengths shorter than 220 nm.
You want a solvent that is nonreactive with samples and the atmosphere. Tetrahydrofuran
can degrade and form peroxides, so workers must take special care if they use it. Some
workers have found that adding water to tetrahydrofuran greatly diminishes this problem (4).
Tetrahydrofuran also is much slower to equilibrate with the column and LC system than is
acetonitrile or methanol - it may take twice the solvent volume for initial equilibration with
tetrahydrofuran. And, of course, tetrahydrofuran’s unpleasant odor is another negative
factor.
A final beneficial property of acetonitrile and methanol is their relatively low back pressure at
flow rates of 1 -2 mL/min when compared with tetrahydrofuran.
When I consider the various desirable solvent properties - low viscosity, low UV
absorbance, low reactivity, and convenience - my first choice for an organic solvent is
acetonitrile. However, I would not argue with anyone who preferred methanol as a starting
solvent.
If the samples are neutral compounds, you may be able to use water as the aqueous
phase. However, ionic compounds generally are present in pharmaceutical analysis and
many other applications. Ionic compounds require analysts to maintain pH control to obtain
reproducible methods. You will get the best results if the compounds are fully protonated or
fully ionized. Thus, the pH of the mobile phase should be at least 1.5 units above or below
the p Ka if possible. For organic acids, working at pHs lower than pH 3 generally will be
satisfactory. However, bases will require buffers with pHs higher than pH 8, which exceeds
the recommended working pH of most silica-based columns. Silica columns with extended
pH ranges are available now, but few have extensive experience with these materials at
high pH, so I recommend organic buffers to reduce silica dissolution. These factors make
me wary of routine use of high-pH mobile phases with silica columns.
OTHER FACTORS
You should consider a few other factors before embarking on developing a new LC method.
Because retention changes by 1 -3% for a 1 °C change in temperature, it is important to
control the column temperature. Changes in temperature also can affect selectivity, which
should give you additional incentive to control the temperature. I generally operate the
column oven at a temperature just higher than room temperature (for example, at 35 °C) for
ease of control and to reduce the solvent viscosity for lower-pressure operation.
Sometimes you’ll need to use additional reagents such as ion-pairing reagents or
triethylamine for special sample types. I recommend, however, that unless you know that
you will need these additives, you should start with an additive-free mobile phase.
Remember the KISS principle - Keep It Simple, Stupid - and don’t complicate the mobile
phase unless it is necessary.
Finally, I strongly recommend that you select a column that will be available for years to
come. Chances are that your method will be used for several years, and the heart of
reproducible separation is obtaining a column of similar chemistry year after year. Use a
column vendor that you are confident can provide controlled column chemistry over the long
haul.
IN OTHER WORDS
The 150 mm X 4.6 mm, 5 -µm dp Type B silica column is a workhorse, and when operated
at low pH under controlled temperature it will be an excellent place to start method
development with most samples.
Now that I’ve described the tools to use, next month I’ll begin looking at the steps toward
obtaining the best separation in the minimum amount of time and effort.
REFERENCES
(2) L.R. Snyder, J.J. Kirkland, and J.L. Glajch, Practical HPLC Method Development (John
Wiley & Sons, New York, 2nd ed., 1997).
(3) U.D. Neue, HPLC Columns. Theory, Technology, and Practice (Wiley-VCH, New York,
1997).
John W. Dolan
John W. Dolan
LC Troubleshooting Editor
This month, I will cover the measurement of retention, peak shape, and resolution. These
quantitative measurements of separation quality are important from the first injection
through method validation and on into the application stage of an LC method.
Retention
The retention time (tR ) is measured as the time between sample injection and the apex of
the peak (Figure 1). Retention time probably is the most used chromatographic parameter.
Retention is said to be characteristic of a compound but not unique. It is characteristic
because if all conditions are held constant, an analyte will be eluted at the same time in
every run. For this reason, retention time often is used as a qualitative tool to identify a
compound. If a standard of the compound is injected, and the retention time is the same as
a peak in the sample, it is likely that the two compounds are the same. However, retention
is not unique because more than one compound can be eluted at the same retention time.
This potential for coelution is the basis of the separation challenge in LC.
Figure 1: Chromatogram illustrating retention time, column dead time, and
retention factor.
Retention Factor
Although retention time is a very useful measurement, the retention factor ( k) often is a
more useful parameter for method development. The retention factor provides analysts with
information about the quality of the separation. For example, chromatographic conditions
that generate values of 1 , k , 20, or better yet 2 ,, k , 10, are more likely to yield an
acceptable separation than those that generate k values outside these limits.
Chromatographers know from experience that if k is too small, the retention is short and the
peaks of interest tend to have problems with interferences at the injection front; if k is too
large, peaks become broad and hard to detect and the run time is excessive.
[1]
where t0 is the column dead time, usually noted by the first disturbance in the baseline or
the elution of a solvent peak (Figure 1). Although k can be calculated, usually it is easier to
estimate the value of k, and an estimate is good enough for method development purposes.
Note that the numerator of equation 1 is the corrected retention time - the time after t 0
required for the peak to be eluted. The denominator defines the units for k - units of t0. So to
estimate t0 , just start measuring at t0 and see how many t0 units the peak requires to be
eluted. Figure 1 illustrates this measurement below the time axis. The peak eluted at 6 min
comes out three t 0 units after t 0, so the k value is approximately 3. For method development
purposes, if k is estimated within 0.5-1 units, it will be satisfactory. It should be noted that k,
as defined in equation 1, is useful only for isocratic separation.
Let’s pause for a moment and compare tR and k. Retention is directly affected by changes
in the flow rate and column size, whereas k remains constant when either of these
parameters changes. Both t R and k change when the mobile-phase composition or the
column temperature changes. Thus, anything that affects t 0 and tR proportionally will not
affect k. The practical implication of this situation? After you obtain a separation that you like
on one column, you can change to a column of different dimensions, even if the flow rate is
changed, and the k value will remain unchanged, which means that the quality of the
separation, in terms of retention, is unchanged.
So when examining a chromatogram, you should make a quick estimate of the k values of
the first and last peaks to see if they fall within the 1 , k , 20 range. If everything is bunched
at the beginning (small values of k) or is strongly retained (large k values), a change in the
mobile-phase composition probably will be necessary to obtain a satisfactory separation.
Peak Shape
Factor (at 10%) Factor (at 5%) Tailing peaks indicate that more than one
retention mechanism is present in the
1.0 1.0 interaction of that peak with the stationary
phase, a situation that is undesirable. A
1.3 1.2 more practical concern is that when tailing
peaks are present, analysts must use
longer run times for baseline separations,
1.6 1.4 and because some of the area is under the
tail, the peak heights are smaller, which
can compromise detection limits. Usually
1.9 1.6
peak tailing can be minimized by using a
column based on Type B silica and
2.2 1.8 operated with a low-pH mobile phase, as
was discussed in last month’s “LC
2.5 2.0 Troubleshooting” (1). Sometimes mobile-
* Reprinted from reference 2 with phase additives such as triethylamine can
permission. be used to reduce peak tailing.
[2]
where L is the column length in centimeters and dp is the particle diameter in micrometers.
Thus, a 150-mm column containing 5-µm particles should have a plate number of
approximately 9000 under practical operating conditions. Using this guideline, you should
examine peaks in the chromatogram to see if they have reasonable plate numbers.
Chromatographers should consider corrective action if the plate number for a peak of
interest is more than about 20% below the guideline of equation 2.
Putting It Together
So now you have three tools to help you examine the quality of a chromatogram. First, you
will want to estimate k values for the first and last peaks in the chromatogram to see if
retention is in the region most likely to yield a good separation. Next, examine the individual
peaks to see if they are well shaped and if the peak widths are reasonable. This set of
measurements will determine if the peaks are well behaved, increasing the likelihood of a
successful separation. However, these observations of the chromatogram are secondary to
the real goal of chromatography - obtaining a reasonable separation. To determine if the
separation is acceptable, you must make one more measurement - resolution.
Resolution
[3]
where t1 and t 2 are the retention times and w1 and w2 are the baseline widths of the two
peaks. An alternative formula uses the half-height peak widths
[4]
where w0.5,1 and w0.5,2 are the half-height peak widths. For Gaussian shaped peaks, the
valley between two peaks hits the baseline at Rs ' 1.5. To have a safety margin that allows
for some deterioration of a method, most workers strive for separations with a minimum
resolution of 1.7 -2.0. When resolution gets much larger than 2, no particular separation
improvement is achieved for most applications and run times can be excessive.
Although equations 3 and 4 are good tools for measuring resolution, during method
development resolution is more usefully defined using
[5]
where a is k2 /k1 , which is the ratio of k values for two adjacent peaks. Equation 5 comprises
three major components. Part i relates to the column quality. Columns with larger plate
numbers result in better resolution. However, resolution improves only as the square root of
the plate number, so to double the resolution, chromatographers must increase N fourfold.
This change is impractical. For example, connecting four columns in series would mean a
fourfold increase in run time and pressure plus a significant financial investment for a
twofold increase in resolution. The plate number is easily calculated from first principles,
and it is easy to predict how a change in particle size, column length, or flow rate will affect
N. Because plate number changes are so easily predicted, it is best to leave these changes
to the end of the method development process. Start with a column that will generate a
reasonable number of plates for most separations. Last month I suggested that a 150-mm,
5-µm dp or 75-mm, 3.5 -µm dp column was a good starting point.
Part iii of equation 5 is the retention term (in iii k is the average k value of the two peaks
under consideration). As retention (or k) increases, resolution also improves. An interplay
between the retention and selectivity terms exists, because both contain k; that is, a change
in k ( iii) generally will change a (ii) as well.
Last month, I looked at the importance of selecting a column and the mobile-phase
conditions that were most likely to provide a successful separation. Although success under
those conditions is not guaranteed, the chances are improved for obtaining an acceptable
separation with a minimum investment in method development time.
This month, I presented the tools for measuring the quality of the chromatogram. These
tools allow you to look at a chromatogram, make a few calculations, and determine if the
separation is satisfactory or not. An unsatisfactory peak shape or column plate number
indicates problems with the basic chromatographic process, and they should be addressed
before further method development. Generally, chromatographers can adjust retention ( k
value) and resolution by changing either the mobile- or stationary-phase conditions. Now
that you have these tools at your disposal, you can move into the method development
process and use them to move quickly to an adequate separation. Next month I’ll examine
how to control retention and selectivity, using equation 5 to help guide us.
References
(2) L.R. Snyder, J.J. Kirkland, and J.L. Glajch, Practical HPLC Method Development (John
Wiley & Sons, New York, 2nd ed., 1997), p. 210.
John W. Dolan
John W. Dolan
LC Troubleshooting Editor
There’s a saying to the effect that if you don’t know where you’re going, you’ll never know
when you get there. Addressing a similar concern with LC separations, the second
installment (2) emphasized the need to be able to quantitatively measure the attributes of
the separation. The specific measurements of interest were retention time ( tR ), retention
factor ( k), peak asymmetry (As ) or U.S. Pharmacopeia tailing factor (Tf), column plate
number ( N), and resolution (Rs).
This month, I’ll show you how these tools can be used to guide analysts to a successful
separation. In particular, the fundamental resolution equation (equation 1) will help evaluate
the effect of various selectivity-controlling parameters during method development.
The goal of the chromatographic process is to separate one or more sample components
from other materials in a sample. Last month, I described the measurement of resolution. At
a resolution of approximately 1.5, the valley between two peaks reaches the baseline. Most
workers like to have separations with resolution values of more than 1.7 to provide a little
leeway for loss in separation without making the separation unusable. Similarly, when
resolution is much greater than roughly 2.0, little is gained for most applications - run times
increase without any additional quantitative advantage.
[1]
where
a = k 2/k1…… [2]
and k1 and k2 are the retention factors of the two peaks of interest. For the balance of this
installment of “LC Troubleshooting,” we’ll concentrate on terms ii and iii in equation 1.
Finally, equations 1 and 2 show that as retention changes, peak spacing (a) also changes.
This relationship is the source of the general improvement in resolution as retention
increases (Figure 1). (Note that k in iii of equation 1 is the average of k1 and k2 of equation
2.)
Before you worry too much about selectivity, it is important to get the retention time within a
reasonable range. With isocratic separations, one simple way to adjust retention is to start
with 90% or 100% solvent B (organic solvent) and reduce the concentration 10% at a time.
So run 100% solvent B, then 90%, 80%, and so on until retention times fall in the 1 , k , 20
range desired for good chromatographic behavior. Figure 2 illustrates three steps of this
series of runs. As the organic solvent concentration is reduced from 80% to 70% to 60%,
the retention and resolution increase, as expected from equation 1 and Figure 1. The
column used for the runs of Figure 2 was a 150 mm X 4.6 mm column operated at a 1.5-
mL/min flow rate, so t0 is approximately 1 min. A quick examination of the chromatogram for
the 60% B mobile phase allows a retention factor estimation of 2 , k , 7, which is very good.
The peaks are baseline resolved and the run time is short, so the separation doesn’t need
much optimization.
Figure 2: Simulated chromatograms generated using (a) 80% B, (b)
70% B, and (c) 60% B solvent concentrations.
Note that the peaks in Figure 2 move in a regular fashion as the organic content of the
mobile phase is changed. The Rule of Three will guide us in the general peak movements.
The general rule states that retention factors will increase roughly threefold for each 10%
reduction in the organic-solvent content of the mobile phase. The k values are 1.1, 2.7, and
6.9 for the last peak in the runs of Figure 3, so the rule for this particular sample is more like
the Rule of 2.5, but as a general guideline, the Rule of Three is quite handy. Finally, note
that it is threefold for each 10% change, so 20% solvent B change would cause a 3 X 3 or
roughly ninefold change in retention factor.
All the peaks in Figure 2 follow Figure 1’s predicted pattern. That is, retention and resolution
increase as the mobile-phase strength is reduced. This pattern of change is a generality,
but fortunately it does not hold for every compound. The exceptions to the rule are
important, because they allow chromatographers to fine-tune the mobile-phase composition
and gain selectivity in a separation.
Figure 3 illustrates how you can take advantage of fine-tuning the mobile -phase
composition to improve a separation. From the top to the bottom of Figure 3, the mobile
phase is made weaker, much as it was in Figure 2. And in the same manner, you can see
that the resolution of peaks 2 and 3 improves, and retention time increases with a lower
percentage of solvent B. However, notice the movement of the last two peaks in each run.
The resolution of this peak pair changes in the opposite way. These changes are common
in samples submitted for LC analysis; that is, although the general pattern of longer
retention and better resolution with weaker mobile phases holds for many compounds, it is
quite common for resolution to decrease under the same circumstances for other
compounds.
When you observe peak movements as in Figure 3, you may need to compromise to
achieve acceptable separation. In the present case, the resolution of peaks 2 and 3 is better
at 46% solvent B, but the last two peaks are better resolved at 48% solvent B. Midway
between these two conditions, 47% solvent B, compromises the separation of both peak
pairs but gives the best overall separation of this sample.
Figure 3 shows one other separation feature that you should look for in method
development - robustness. Robustness is the ability of a separation to withstand small
changes in conditions without compromising the results. The separation at 47% solvent B in
Figure 3 certainly is not robust - a 61% change in organic solvent concentration significantly
degrades the separation. In routine use, this method would require undue care in mobile -
phase preparation, so these results should alert chromatographers to look for alternate
conditions that would produce a more satisfactory separation.
Solvent Type
The adjustment of solvent strength alone can allow you to find satisfactory separation
conditions for many samples. For other samples, however, no concentration of acetonitrile
will generate a suitable separation. At this point, you should take a different approach for
changing a. One powerful way to change a for many samples is to change to another
organic solvent in the mobile phase.
For the sample of Figure 4, acetonitrile-water mobile phases were unsuccessful in achieving
the desired separation. My next choice in mobile -phase solvents is methanol. Figure 4a
shows the best separation with methanol-water (55% methanol). The resolution for most of
the peaks in the chromatogram is satisfactory, but peaks 2 and 3 are poorly separated.
Fine-tuning the concentration of solvent B, as was done in the previous example, was
unsuccessful, so I tried still another solvent.
I substituted tetrahydrofuran for methanol for the separation of Figure 4b, and this figure
shows that the resolution of the later peaks in the run is satisfactory. However, the
beginning of the run still has trouble. Peaks 1 and 2 and peaks 4 and 5 are poorly resolved.
Once again, fine-tuning the concentration of solvent B provided no better results.
At this point, you might want to give up - acetonitrile, methanol, and tetrahydrofuran all
failed to provide a satisfactory separation. However, the separations of Figure 4 illustrate
another practice in vigilance that should be the habit of successful chromatographers. Note
that the critical peak pairs (those most poorly resolved) change between the runs of Figure
4a and 4b. Just as was the case for the sample of Figure 3, an intermediate set of
conditions should give peak separations and retention times falling between those observed
for the best methanol-water and tetrahydro- furan-water conditions. Indeed, this situation is
what I observed. When the isocratic solvents of Figure 4a and 4b were blended, I obtained
the separation of Figure 4c. This condition gives the desired baseline resolution of all peaks
in the chromatogram.
Summary
Guidelines for Adjusting The organic solvent plays a major role in controlling
selectivity in reversed-phase LC separations. I’ve
the Concentration of
summarized several guidelines that you should
Solvent B remember from the examples above.
• Adjust the concentration of
solvent B for acceptable First, adjust the concentration of solvent B
retention range (acetonitrile as the first choice) until all the peaks fall
in the 1 , k , 20 region. In general, resolution will
• Fine-tune the concentration of increase with increasing retention, but watch the
solvent B for desired resolution movement of individual peaks to help determine
when fine-tuning of the concentration of solvent B is
•Change solvent type to change justified.
selectivity
•Blend solvents, if necessary If acetonitrile is ineffective, try methanol and repeat
the optimization experiments; if methanol fails, use
•Selectivity changes between tetrahydrofuran. Keep relative peak movements in
runs can help you evaluate the mind as you make these changes, because changes
usefulness of intermediate in selectivity between conditions often indicate that
conditions an intermediate condition or solvent blend will be
useful to pull apart hard-to-separate peak pairs.
Solvent strength and type are two powerful variables that affect chromatographic selectivity,
but they are by no means the only variables available. Next month’s “LC Troubleshooting”
column will highlight some other variables that can be used to control selectivity. I will
provide some guidelines about when the use of these variables is appropriate.
References
John W. Dolan
John W. Dolan
LC Troubleshooting Editor
Although varying solvent strength and type provides powerful tools for controlling the
separation, these variables may be insufficient to obtain the desired resolution for a
particular sample. For this situation, chromatographers must use other parameters to adjust
selectivity. This month’s “LC Troubleshooting” column will concentrate on three additional
parameters: pH, temperature, and column type.
Mobile -Phase pH
Figure 1 illustrates the power of using pH to control selectivity (4). The figure shows that
acids will be protonated, hydrophobic, and thus more strongly retained at low pH than at
high pH. Conversely, bases will be neutral and well retained at high pH, whereas low pH
results in ionization and poor retention in a reversed-phase system. Neutral compounds are
unaffected by pH, so the retention will vary little, if at all, when the mobile-phase pH is
changed.
Figure 1 also illustrates that a mixture of acids, bases, and neutrals will cause many
retention reversals when the pH is changed. This effect can be a powerful tool to control
selectivity and obtain a separation impossible through the adjustment of other parameters,
but it also indicates that chromatographers must take care to control the pH and avoid any
unintended changes in selectivity.
The plot of retention versus pH for compound 4, which is a base, in Figure 1 shows a shape
closely resembling a titration curve. Not surprisingly, the midpoint of this curve is at the p Ka
for this compound, which is approximately pH 7. Acids show the same curve shape in mirror
image, although the pH range in Figure 1 is insufficient to show the entire curve for any
compound except number 4.
Figure 1 shows that an adequate separation of all sample components can be obtained at
pH 5 or 7. However, at pH 7, the method is much more likely to be sensitive to small
changes in pH, particularly for compounds 4 and 5. For maximum method robustness, it is
best to work at a pH at which the compound or compounds of interest are either fully
ionized or ionization is suppressed. This condition means that chromatographers should
operate at more than ±1.5 pH units from the pKa value. As usually is the case, workers may
need to compromise for the best results, but operation at pH 5 will be less sensitive to pH
changes than that at pH 7 for this sample.
As mentioned earlier, I recommend starting the initial run with most ionic samples at low pH;
for example, pH 2.5. Under these conditions, adjust the organic solvent content of the
mobile phase to obtain a good retention range for neutral and nonionized samples. To
screen for gross effects of pH, chromatographers can change the mobile-phase pH while
keeping the organic solvent at a constant concentration. For example, make the initial run at
pH 2.5, then change to pH 7.0 for another run to see if dramatic changes occur. However,
to fine-tune the mobile-phase pH, it is best to vary the mobile phase in steps not exceeding
0.5 pH units because retention and peak spacing can change dramatically with small
changes in pH. These dramatic changes can make peak tracking difficult, especially if peak
areas change, which can happen with pH changes.
Figure 2 shows an example of fine-tuning the mobile -phase pH for a group of benzoic acids
(5). In each case, only the pH is varied. Whenever examining a set of chromatograms such
as in this figure, it is useful to concentrate on the critical (least resolved) peak pairs. Peaks
2, 6, and 9 move dramatically in relation to the other peaks in the sample. A satisfactory
separation of all nine components can be obtained at pH 3.0. As is expected with a group of
organic acids, lower pH will reduce the degree of ionization and make the components more
hydrophobic and thus more strongly retained.
Figure 2: Simulated separation of nine benzoic acids at (a) pH 2.5, (b) pH 3.0, (c) pH
3.5, and (d) pH 4.0. Peaks: 1 = 2 -nitrobenzoic acid, 2 = phthalic acid, 3 = impurity, 4 5 2 -
fluorobenzoic acid, 5 = 3-cyanobenzoic acid, 6 = 2 -chlorobenzoic acid, 7 = 3 -nitrobenzoic
acid, 8 = 3-fluorobenzoic acid, 9 = 2,6-dimethylbenzoic acid. The conditions are listed in
Table I. The simulated chromatograms were generated from the data of reference 5.
pH Varies pH 3.2
Temperature 35 °C Varies
Column Temperature
Another variable that can provide useful changes in selectivity is column temperature. Most
workers consider temperature only in terms of retention time changes. As a general rule,
retention in reversed -phase LC can be expected to change 1 -3% for each 1 °C change in
temperature. Figure 3 illustrates this rule; in this figure retention changes roughly 25% (19
min to 25 min for the last peak) with a 15 °C change in temperature, or approximately
1.7%/°C. Table I lists the conditions for this separation. Temperature control is important for
reproducible retention times.
Figure 3: Simulated separation of nine benzoic acids at (a) 30 °C, (b) 35 °C, (c) 40 °C,
and (d) 45 °C. Peaks: Same as in Figure 2. The conditions are described in Table I. The
simulated chromatograms were generated from the data of reference 5.
It is less appreciated that changes in temperature can cause changes in selectivity in LC
separations. This condition is illustrated in Figure 3 for the same benzoic acid sample that
was shown in Figure 2. Although the general trend is shorter retention times with increased
column temperature, important selectivity changes take place, too. Notice the movement of
peak 5 as the temperature is increased, changing from coelution with peak 6 at 30 °C to
beginning to merge with peak 4 at 45 °C. Because peaks move in a regular fashion with
changes in mobile-phase conditions, peaks 4 and 5 would completely merge when the
temperature is increased to higher than 45 °C (overlap is complete at 53 °C). Similarly,
peaks 8 and 9 move relative to each other when the column temperature is changed. For
this sample, a temperature of 40 °C will provide baseline separation of all sample
components.
Why does temperature provide such dramatic changes in selectivity for this sample? In this
case, the changes to a large degree are due to the ionic character of the sample. The
temperature will influence the ionization characteristics, hydrophobic retention of ionized
versus nonionized compounds, silanol interactions with ionic species, mobile -phase pH, and
the p Ka s of sample components. With all of these influences, it would be surprising if a
change in temperature did not influence selectivity.
The practical result for this benzoic acid sample is that workers can obtain a satisfactory
separation under more than one set of conditions. The similarity between the pH 3.0 (35 °C)
run of Figure 2b and 40 °C (pH 3.2) run of Figure 3c is striking. Both conditions provide
baseline separation of all components in approximately 25 min. When ionic species are
present in the sample, column temperature and mobile-phase pH will play an important role
in peak spacing.
Column Selectivity
Up to this point in the discussion, I have talked about changes in selectivity accomplished
by making changes in the mobile-phase composition. This approach usually is the most
convenient and least expensive way to change selectivity. An alternative approach, usually
best left until mobile-phase alternatives have been exhausted, is to rely on changes in
column selectivity to control the separation. As with changes in mobile -phase selectivity,
chromatographers should make significant changes in column selectivity to take maximum
advantage of this parameter. Therefore, changing from one brand of C18 column to another
or from a C18 to a C8 column will not be very fruitful. These kinds of changes may result in
selectivity variations, but they tend to be minor when compared with changes in the
stationary-phase chemistry.
For reversed-phase separations, three column types - C18 or C8, cyano, and phenyl -
traditionally are selected. Figures 4 and 5 illustrate the changes in selectivity that occur
between a C8, cyano, and phenyl bonded phase for two different samples. The sample of
Figure 4 (6) is a group of substituted benzoic acids, some of which are in the sample of
Figures 2 and 3. The most prominent change is in the retention order of the 4-5-6 triplet, in
which compound 4 comes out first with the C8 column, but last with the other two. Also, the
peak spacing between peaks 1 and 2 changes markedly.
An herbicide mixture was used for the separations of Figure 5 (6). Peaks 1 and 2 run
together on the cyano column and the separation of 6 and 7 is reduced as well. No retention
order changes are observable for the herbicide mixture on these columns.
It is apparent that column chemistry changes may yield dramatic selectivity changes, as is
the case for the benzoic acids in Figure 4, or the changes may be much more subtle, as
with the herbicides in Figure 5. It is impossible to accurately predict changes in selectivity
between columns, and no software is available to assist in fine-tuning selectivity changes
such as the software used for mobile -phase modifications, so column changes tend to be
trial-and-error in nature. For this reason, I recommend trying the easier and often more
effective mobile-phase changes before exploring column selectivity.
In my discussions with other users, I have found that the phenyl column is used infrequently
today as a tool to enhance selectivity. In the last few years, the polar embedded phases
have come into favor. These phases tend to be C18 or C8 phases with an amine-containing
polar function attached near the base of the C18 chain, typically on the third carbon from
the base. The result is bonded phases that often have distinctly different selectivity
characteristics than traditional C18 or C8 phases. In my laboratory, we have found these
phases to be a good third member to go with the C18-C8 and cyano phases when exploring
column selectivity. Some brand names are Discovery Amide (Supelco, Bellefonte,
Pennsylvania), Symmetry Shield (Waters Corp., Milford, Massachusetts), and Zorbax
Bonus RP (Agilent Technologies, Wilmington, Delaware). Check with your favorite column
supplier for other columns.
Other Variables
The discussion thus far has concentrated on the variables that will give the most leverage in
changing the selectivity of a separation. Many other variables can be explored. For
example, ion-pair reagents often are used to enhance selectivity, especially when a mixture
of acids and bases is present in the sample, but ion pairing tends to be more problematic
than simple pH control. Chromatographers can exploit special chemical characteristics of
the sample - such as the presence of chelators, chiral compounds, or shape differences - to
gain selectivity. When a good effort has been made at obtaining a separation with the
traditional reversed-phase variables, workers may want to explore modes of
chromatography such as ion exchange, normal phase, or size exclusion. Reference 7 is a
good resource for method development for these and other samples.
Conclusions
References
(4) P.J. Twitchett and A.C. Moffat, J. Chromatogr. 111, 149 (1975).
(5) J.W. Dolan, D.C. Lommen, and L.R. Snyder, J. Chromatogr. 535, 55-74 (1990).
(6) J.J. DeStefano, J.A. Lewis, and L.R. Snyder, LCGC 10(2), 130-139 (1992).
(7) L.R. Snyder, J.J. Kirkland, and J.L. Glajch, Practical HPLC Method Development (John
Wiley & Sons, New York, 2nd ed., 1997).
John W. Dolan
In February’s column (3), I introduced the fundamental resolution equation as a guide for
the method development process:
[1]
where resolution ( Rs) is a function of the column plate number (N), the i portion of the
equation; the selectivity factor (a), the ii portion of the equation; and the retention factor ( k),
the iii portion of the equation. The February and March installments (3,4) discussed
adjusting solvent strength and mobile-phase chemistry to control k and α and obtain the
best possible separation for one or more pairs of peaks in the sample. After these
parameters are optimized, chromatographers can do nothing else using that column,
solvent system, or temperature to improve the separation. However, at this point in the
method development process, the column plate number has not been optimized.
There are several schools of thought about the best column size to use when starting
method development. Some workers contend that a short, 30-50 mm column packed with 3-
µm particles should be used because it allows very fast runs and, thus, the chance to
explore many different mobile-phase conditions quickly. At the other extreme, others would
choose a 250-mm, 3 -µm dp column, so that they could obtain the maximum resolution from
every run. I tend to sit in the middle of the pack - my first choice would be a 150-mm, 5-µm
dp or 75-mm, 3.5-µm dp column. I feel that although the short columns allow more
separations to be run in a shorter time, the plate numbers aren’t large enough to provide
reasonable separations of anything but simple samples. The long, small-particle columns
require low flow rates for reasonable back pressures, so the run times are excessive. The
intermediate column lengths generate sufficient separation power for most separations, yet
they can be operated at flow rates of 2 mL/min or higher without excessive back pressure.
No matter which column size you select for the initial work, you may improve the overall
separation with a change in the column conditions, including the mobile-phase flow rate,
column size, and packing particle size. Sometimes these changes can transform a marginal
separation into a satisfactory one. At other times, changes in column conditions can trade
excess resolution for improved sample throughput. For many samples, the separation is
good enough already, so it isn’t worth trying to adjust the column conditions. If you have
access to chromatography modeling software, you can model changes in column conditions
to help you decide if this avenue of method improvement is worthwhile.
As it is with mobile -phase parameters, analysts always have a choice of which column
parameter to vary first. I usually choose to vary flow rate first, although flow rate is not likely
to make a large change in resolution, because it is easy and it will not change selectivity
due to the absence of potential chemical changes. The flow rate’s importance as a
parameter influencing resolution is decreased as the column packing’s particle size is
reduced. Figure 1 illustrates this relationship by comparing 10 -, 5-, and 3-µm dp column
packings. Compare the results of a change in flow rate for each packing size. In these
comparisons, it is easy to focus on the resolution, which is expressed as the depth of the
valley between the two peaks. With the 10-µm dp material, a dramatic improvement in
resolution occurs as the flow rate is reduced from 4 mL/min (Rs 0.8) to 2 mL/min to 1
mL/min ( Rs 1.1). The change for the 5 -µm dp column is less obvious, and the resolution
with the 3 -µm dp column changes only slightly. Another way to look at this relationship is
that increasing the flow rate from 1 mL/min to 4 mL/min will reduce the run time by a factor
of four in all cases, but it costs less than 10% in resolution for the 3-µm dp column, whereas
resolution drops by 25% with the 10-µm dp column.
Figure 1: Separations showing the effect
of flow rate on resolution for columns
packed with (a) 3 -, (b) 5 -, and (c) 10-µm dp
particles. Flow rates (from left to right) are
1, 2, and 4 mL/min.
For practical purposes, flow rate is not a very important factor with the 3-5 µm dp columns
commonly used today. Most workers choose to adjust the flow rate for the shortest run time
while keeping the system pressure within reasonable limits. In cases in which resolution is
marginal, a drop in flow rate sometimes can be advantageous, but the effect will be small,
as is illustrated in the examples of Figure 1. However, changes in flow rate are easy to
make by resetting the LC pump, and, because no chemical change is made, the relative
peak spacing will remain unchanged.
Where do you start when considering column size changes? Most manufacturers have a
large variety of column configurations from which to choose. For example, I made a quick
check of the catalog on the Waters Corp. (Milford, Massachusetts) web site
(www.waters.com) for the XTerra columns packed with the company’s MS phase. The
listing includes 60 separate part numbers distributed between combinations of 2.5-, 3.5-,
and 5-µm dp packings in columns of 20-, 30-, 50-, 100-, 150-, and 250 -mm lengths and 2.1-,
3.0-, 3.9-, and 4.6-mm inner diameters. The assumption here is that chromatographers
could freely change between any of these columns and have identical selectivity, as long as
the mobile -phase composition and column temperature didn’t change. Let’s take a look at
the influence of column diameter and length on the separation; particle size is covered in
the next section.
Column inner diameter should have no effect on the separation, if workers take two
important factors into account. First, if the linear velocity of the mobile phase is kept
constant, the separation should be identical for columns of the same length and packing
content. Therefore, if the column diameter is changed, the flow rate must be adjusted
according to the square of the diameter change. For example, changing from a 4.6 -mm to a
2.1-mm i.d. column results in approximately a fivefold change in cross-sectional area, so the
flow rate must be reduced by fivefold to maintain the same linear velocity.
Column length, on the other hand, is a convenient tool to adjust the method either to
increase the resolution or to reduce the run time. Consider an increase in resolution. The
second two peaks in Figure 2a are marginally separated with a resolution of roughly 1.4.
Let’s assume that our method goal is resolution of at least 1.7. By increasing the column
length from 150 mm to 250 mm, resolution is improved to 1.8 (Figure 2b). (Table I lists
conditions for these separations.) According to equation 1, resolution increases with the
square root of the plate number, and the present case demonstrates that relationship. The
plate number increases in proportion to the column length, so (250/150) 0.5 1.3, and
resolution improves by (1.8/1.4 ) 1.3. However, the increase in resolution is not without
cost. Note that the pressure increases in proportion to column length, as does the run time.
If you want to keep the system pressure near 2000 psi, the flow rate must be reduced by
one-third, which would further increase the run time by one-third to approximately 10 min.
Because of the square-root relationship between N and Rs (equation 1), you can make only
minor gains in resolution by changing column length - perhaps 30-40% gain is possible in
favorable conditions.
Figure 2: Simulated separations generated using (a) 150- and (b)
250-mm long columns. Pressures were (a) 1800 and (b) 3000 psi.
Other conditions are summarized in Table I. The computer
simulations are based on data from reference 5.
The runs of Figure 3 illustrate another case in which adjustment of the column length will be
useful. In this case, the initial separation (Figure 3a) used a 150-mm long column operated
at 1 mL/min. The desired separation for the second two peaks was a resolution of at least
1.7; the actual resolution is 2.1. Although many workers would leave the separation as is, if
you need a resolution of only 1.7, this separation is wasting time. Reducing the column
length from 150 mm to 100 mm (Figure 3b) reduces the resolution to the 1.7 target, but it
also shortens the run time and results in lower back pressure. Doubling the flow rate has no
practical effect on the resolution, yet the run time is halved while maintaining a reasonable
system pressure (Figure 3c). Thus, in this example, shortening the column by one-third
allows the run time to be reduced by two-thirds.
Figure 3: Influence of packing particle size, column length, and
flow rate on resolution. Conditions: (a) 150-mm long, 3-µm dp
column operated at a 1 -mL/min flow rate; (b) 100-mm long, 3-µm dp
column operated at a 1 -mL/min flow rate; (c) 100-mm long, 3-µm dp
column operated at a 2 -mL/min flow rate. Other conditions are
listed in Table I. Same sample as in Figure 2.
To take advantage of adjusting column length to throw away excess resolution, you must
start with a separation that has excess resolution. This is one argument for doing the extra
work to get the maximum possible resolution with a separation through adjustment of the
mobile-phase chemistry - any excess resolution can be traded for shorter run times by
adjusting the column conditions.
My observations from my laboratory and those I visit lead me to believe that the 5 -µm
particles still are the workhorse packing for routine use. Many times analysts can improve
resolution or shorten run times by using 3 -µm packing material. When available, I prefer the
3.5-µm particles over the 3.0-µm ones because they allow use of column hardware that is
less prone to blockage and provide nearly the same chromatographic performance as 3.0 -
µm particles. Smaller particles, such as 1.5-µm dp, are available, but they are best used for
specialty applications such as separation of proteins or fast runs when a short, low-plate
number column can be traded for speed.
You can fine-tune a separation by making adjustments in the flow rate, column dimensions,
and packing particle size. My preference is to start with the flow rate, because it is easy and
leaves the selectivity unchanged. Next, I’d change the column length, because it is unlikely
that any significant column chemistry changes will occur between columns from the same
manufacturer packed with the same size and description of packing material. Finally, I’d
look at changing the packing particle size. Although particle size is a powerful way to
change the plate number, it carries the most risk of selectivity changes because the
different particle sizes were made in a slightly different synthesis process. Fortunately,
today’s column manufacturers work very hard to give chromatographers a variety of column
configurations that have as little variability in column chemistry as possible.
Next month ’s “LC Troubleshooting” will wrap up the “Starting Out Right” series by
considering the use of gradient scouting runs to speed the method development process.
References
(5) L.R. Snyder, J.W. Dolan, and M.P. Rigney, LCGC Mag. 4(9), 921-929 (1986)
John W. Dolan
John W. Dolan
LC Troubleshooting Editor
When I presented the strategy of using step-wise mobile -phase strength changes to identify
effective isocratic conditions (3), I acknowledged a major weakness of this technique.
Wasted runs are almost guaranteed. The first run usually is made with 100% organic
solvent (solvent B) or perhaps 90% solvent B. Under these conditions, nearly every
compound will come off the column very quickly. In my previous discussion, I reduced
mobile-phase strength in 10% increments until a separation began to appear (3). I used the
Rule of Three as a guideline that a 10% change in the percentage of organic solvent
generates an approximately threefold change in retention. Even with experience and luck,
chromatographers generally need four to six runs before they can obtain a separation in
which the peaks are eluted in the desired 1 , k , 20 region. Next, they undertake
adjustments to improve selectivity.
One advantage of gradient elution is that runs can be designed to end at a high percentage
of organic solvent, often 100% solvent B. At 100% solvent B in isocratic separations, most
compounds are eluted with k , 1. In a similar manner, when the gradient reaches 100%
organic solvent, compounds that haven’t been eluted yet generally will be eluted in less
than 5 min. So now for a scouting run ending at 100% organic solvent, in most cases we ’re
guaranteed that all sample peaks will be eluted within 5-10 min after the gradient is
complete. For example, a 20 -min gradient followed by a 10-min hold should be sufficient for
the elution of all components from nearly any sample.
From a practical standpoint, you can compare the use of a scouting gradient with the
sample used for Figures 1-3. If you use the isocratic step procedure - starting with 100%
acetonitrile, 90% acetonitrile, and so forth - and allow 5 min after the last peak is eluted
before changing to the next solvent plus a 10-min solvent changeover, it will take just less
than 3 h before you find conditions under which the first peak is eluted with k . 1. Contrast
this situation with a scouting run of 20, 30, or even 60 min, in which all the peaks would be
eluted within the defined gradient time.
Another advantage of the gradient as a scouting run is that it automatically flushes the
column between each run. You don’t have to wait for an undetermined amount of time after
each run and hope that all the sample and matrix components have been eluted before
setting up for the next run.
With isocratic separations, I demonstrated that conditions for good chromatography were
those that generated 1 , k , 20, or better yet 2 , k , 10 (3). These conditions were found by
step-wise exploration of isocratic conditions. For gradient separations, the solvent strength
changes within the run, which means that k values change within the run, so the isocratic
expression of retention factors doesn’t apply with gradients. Instead, I can use the average
retention factor ( k*) to express retention factors in gradient elution. It turns out that k* values
in gradient elution are roughly the same for all peaks in the run. A target for good
chromatography in gradient elution sets is 1 , k* , 10. I like k* to be approximately 6 for a
scouting run. Fortunately, gradient elution theory is developed well enough that it is possible
to predict separation conditions that can generate any desired k* values; for example, see
chapter 8 of reference 6. For k* approximately equal to 6, theory tells us that for a full-range
(5-100%) gradient:
t = 30V /F ……[1]
G M
where t is the gradient time in minutes, V is the column volume in milliliters, and F is the
G M
flow rate in milliliters per minute. The recommended initial conditions I’ve been using so far
in this series are a 150 mm X 4.6 mm column with a column volume of approximately 1.5
mL at a flow rate of 1.5 mL/min. These conditions mean that a 30 -min gradient should, with
high probability, yield a good separation on the first injection. Obviously, adjustments will be
necessary to get the best separation, but as a starting point, these conditions are likely to
generate a useful chromatogram with the first injection.
Is Isocratic Possible?
The conditions shown in Table I are recommended starting conditions, according to the
guidelines of equation 1. These conditions are the gradient version of the isocratic starting
conditions recommended in Table II of reference 1. Under these conditions, you should
allow at least 10 min to equilibrate a column under the initial solvent conditions (5% B)
between each run and place a 5 -min hold at 100% B at the end of the gradient. If phosphate
buffer will be used, it is wise to verify the solubility of the buffer in acetonitrile on your LC
system before making the first gradient run. Some systems have problems mixing
phosphate and acetonitrile when the acetonitrile composition exceeds 80%. Problems can
be minimized by dropping the phosphate concentration to 10 mM or choosing a more
soluble buffer such as trifluoroacetic acid.
Figure 1 shows an example of the results of a scouting run using the conditions of Table I
for a 14-component polynuclear aromatic hydrocarbon sample. By examining the
chromatogram of Figure 1, you can see the complexity of the sample and get an idea of
how difficult it will be to obtain a satisfactory separation. Although all the peaks are not
baseline separated, all 14 peaks are present in the sample, so I would feel confident that
fine-tuning the conditions will allow me to obtain a separation with baseline resolution of all
peaks.
Figure 1: Simulated chromatogram for a separation of 14
polynuclear aromatic hydrocarbon compounds. See text for details.
For the example of Figure 1, the retention times of the initial and final peaks are
approximately 18.5 and 28.5 min, respectively, so Dtg is 10 min. This time period can then
be used to calculate Dt /t ' 10/30 ' 0.33 . This amount falls between the clear guidelines
g G
for isocratic and gradient methods, so you could explore both methods, if desired.
Continuing with the example of Figure 1, first assume that you desire an isocratic
separation. You can use the information from the scouting run to find conditions for an
isocratic separation with the aid of Figure 4. First, determine the average retention time for
the scouting run - (t 1 t )/2 ' (28.5 + 18.5)/2 ' 23.5 min . Then, select the column at the
f i
top of Figure 4 corresponding most closely to the column length (assuming an inner
diameter of 4.6 mm) and the system dwell volume (V ). The figure shows typical dwell
D
volumes for low-dwell volume systems, such as Agilent Technologies’ HP 1100 LC system
(Wilmington, Delaware) and Waters’ model 2690 LC system (Milford, Massachusetts), high-
pressure mixing systems (2.3 mL), and low-pressure mixing systems (4.5 mL).
The current example was modeled on a 150 mm X 4.6 mm column with a high-pressure
mixing system, so choose data under the 150/2.3 heading. Find the retention time below the
heading that corresponds closely to the average retention time for the scouting run and read
the recommended isocratic mobile phase to the left. The average retention of 23.5 min falls
between 21 and 25 min, so I’d choose 65% acetonitrile for the first isocratic run. This mobile
phase should provide an isocratic k value of approximately 4 for the peaks in the middle of
the run. Because Figure 4 is based on several generalizations, it is likely that the first
isocratic run will be a bit long or a bit short, but it should give you a reasonable starting point
for further isocratic optimization.
Figure 2 shows a simulated run under the recommended conditions of Figure 4. Only 12 of
the 14 sample components are clearly visible in the separation. The peak at 20 min
obviously is a doublet, but the location of the second overlapped peak pair (peaks 1 and 2)
is not apparent. The k range for the separation is 1 , k , 22, which is a bit broad but usable.
Earlier “LC Troubleshooting” columns in this series showed that resolution generally
increased with increased overall retention. Improving this separation would require even
longer run times. With these results, I would be pessimistic about obtaining an isocratic
method that yields baseline separation of all the peaks in the sample.
With discouraging results for the prospects of developing a good isocratic separation, you
could explore a gradient method next. By examining the separation of Figure 1 again, you
see that all 14 components are nearly baseline resolved, but the first half of the
chromatogram is completely wasted. A simple way to think about peak movement in
gradients is that a given analyte sits at the head of the column until the arrival of a solvent
strong enough to push the analyte through the column. With this in mind, it seems that
almost half of the gradient is unused, so perhaps starting at 50% B would work. Similarly,
you should be able to stop the gradient as soon as the last peak is eluted. Rather than
guess at the starting conditions, you can use Figure 5 to guide the selection of new initial
and final gradient conditions.
Use Figure 5 in the same manner as Figure 4 to find the data set corresponding to the
column and LC system - a 150-mm long column and a 2.3 -mL dwell volume for this
example. Look down in the table until you find the retention time of the initial peak and read
the recommended initial percentage of solvent B at the left. In this case, 18.5 min falls
between 17 and 20 min, so I would select 45% acetonitrile for the new starting percentage
of solvent B. Similarly, you can find the retention time of the final peak (28.5 min) and read
to the right to find the recommended final percentage of solvent B (90% acetonitrile). This
figure tells us that we can get rid of the wasted time in the run of Figure 1 by using a
gradient of 45 -90% acetonitrile instead of the 5 -100% range selected for the scouting
gradient.
In gradient elution, peak spacing and k* values are controlled by the steepness of the
gradient in a manner similar to changing solvent strength (percentage B solvent) in isocratic
separation. For this reason, it is good to make the next run with the refined gradient range
under conditions that give the same gradient steepness (percentage B solvent per minute)
as in the initial run. This new gradient rate is calculated as a simple proportion based on the
scouting run:
……[2]
So the next gradient run should be 45-90% acetonitrile in 14 min, with all other conditions
the same as the scouting run. Figure 3 shows the results of this new run. The separation is
roughly the same as that in Figure 1, but the run time is approximately 12 min shorter. The
chromatogram still has a 6 -min region at the beginning of the run in which no peaks are
eluted, so it may be possible to increase the starting percentage of solvent B a little more,
perhaps to 50%. Even so, peaks occupy approximately two-thirds of the separation time
instead of the one-third used in Figure 1. You could continue from here by trying different
gradient times to optimize the gradient separation.