Sei sulla pagina 1di 42

Table of contents

Critical Choices
in HPLC

Video Introduction
Laura Bush

Column Selection
Tony Edge and
Dawn Watson

Gradient Methods
Dwight R. Stoll and
Scott Fletcher

Detectors
Scott Fletcher

CONTENTS

TOC

Welcome

Selecting Column Stationary


Phases and Dimensions

Gradient HPLC: Factors


to Consider

The Fundamentals of
HPLC Detectors

To take full advantage of the interactive featuers of this PDF,


be sure to view it in Adobe Acrobat Reader

INTRODUCTION

Welcome

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS
Tony Edge and Dawn Watson

SPONSORED
Comparison of ReversedPhase Selectivity of SolidCore HPLC Columns
Click to
view PDF

SPONSORED
Optimizing Chromatographic
Results with Mobile-Phase
Preheating
Click to
view PDF

Choosing a high performance liquid chromatography (HPLC) stationary phase


requires an understanding of the chemistry of both the stationary phase and the
molecules that are being separated. This article highlights some of the key criteria
to be aware of when selecting a column, including specifications about the columns
physical parameters, such as length and diameter, and also an understanding of
the chemistry, highlighting primary and secondary interactions with the stationary
phase and support substrate material. Additionally, this article discusses the
dependence of retention factor on the mobile-phase pH and how acids and bases
are affected. Ultimately, consideration of a columns physical characteristics,
combined with a thorough understanding of the stationary-phase chemistry, is
essential for achieving the best separation.
When considering the mode of chromatography that should be employed for a given separation, it is
necessary to understand some basic chemistry. In general, the stationary phase is designed to retain the
analyte, with the mobile phase providing additional retention by having limited solubility of the analyte.
In reversed-phase chromatography, the stationary phase is less polar than the mobile phase; therefore
less-polar molecules will be attracted to the stationary phase and the polar mobile phase will have limited
solubility, resulting in a greater retention of hydrophobic analytes on the stationary phase. The difference in
the retention of different analytes, based on this chemistry between the analyte, stationary phase, and mobile
phase, will determine the quality of the separation.
One physiochemical parameter that is very useful when considering the retention of an analyte on a reversedphase HPLC column is the solubility or the log partition coefficient (log P) of the analyte:

log Poct/wat = log

[solute]octanol
n-ionized
[solute]uwater

[1]

The log P value determines how soluble the compound is; larger positive
numbers indicate that the compound is more hydrophobic and less water soluble,
and negative numbers indicate that the compound is quite polar. In the case of
ionizable analytes the distribution coefficient (log D) provides a better estimate
of the analyte solubility as it takes into account all forms of the analyte molecule
(i.e., ionized and unionized, Equation 2). Log D is pH-dependent; hence, when it
is measured, the pH at which the measurement was carried out must be specified.

log Doct/wat = log

[solute]octanol
neutral
[solute]ionized
water
+ [solute]water

[2]

In general, the more carbon atoms present in a molecule, the greater the value
of log P, and in turn, the greater the retention under reversed-phase separation
conditions. The shape of the molecule can also affect analyte solubility, with
straight-chain molecules, in general, having larger log P values; hence greater
retention is seen for branched chain molecules. Furthermore, the greater the
saturation of the carbon-carbon bonds, the greater the log P value and hence,
a greater retention will be observed. In general, aliphatic compounds exhibit
greater retention than compounds with induced dipoles, which have greater
retention than compounds containing permanent dipoles, which have greater
retention than weak bases, weak acids, and strong acids. It should be noted at
this point that most molecules have many different functionalities, which can
make the exact interpretation quite tricky.
For a separation to occur the high performance liquid chromatography (HPLC)
column must be able to differentiate between similar molecules. As has already
been stated, this can be difficult to judge, because there may only be small
differences between molecules perhaps a difference of one carbon unit, or
perhaps two or three differences that could cancel each other out in terms of
the overall retention. It is necessary, therefore, to consider the analytes that will
be analyzed and how to maximize the differences in interactions between the
analytes and the stationary phase. The most predominant modes of interactions
when using a reversed-phased column are hydrophobic, dipoledipole, and
interactions.
There are other parameters to consider other than the chemistry between
the stationary phase and the analyte. For a separation to occur effectively,
the column has to have sufficient available surface area to load the sample. In
addition, the pH, temperature, and pressure can and do have an effect on the
selectivity of the separation mechanism and also on the robustness of the assay.

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

Column Specifications
Column manufacturers will, generally, provide information regarding the following
aspects of an HPLC column:
The nature of the solid support: This is the material to which the bonded
phase is attached, most commonly silica. Silica particles can be fully porous,
superficially porous, or nonporous. The silica particle type will have an effect on
the chromatography and can affect the efficiency of the separation mechanism.
Bonded phase: This is the chemistry of the moiety that is bonded to the silica
surface. Bonded phases are typically based on an alkyl or phenyl group, and it is
the interaction between the bonded phase and the analytes that primarily drives
the separation mechanism.
Particle size: Particle size is measured as the average diameter of the column
packing particles. Manufacturers will also report the distribution of the size
of the particles used to pack the column. In general, smaller particles and
tighter particle-size distributions will give sharper and, hence, more efficient
chromatography.
Particle shape (irregular and spherical): Irregularly shaped particles can be less
expensive, but they provide separations with poor efficiency because of the way
they pack into a column. It is much easier to pack a column with regularly shaped
particles than it is with irregularly shaped particles. Irregularly shaped particles
are also prone to shearing, which creates fines that can block columns, causing
both chromatographic and instrument-based problems, such as poor peak
shapes and increased back pressure.
Pore size: The majority of the stationary phase exists within the silica pore
structure; therefore, the analytes have to access the pores to interact with the
bulk of the bonded stationary phase. This means that the pore size needs to
be appropriate, because a big molecule will not fit into small pore. For small
molecules, the pore size should be about 150 or less. Larger molecules (>2000
Da) need bigger pores, of 300 . The larger the pores, the smaller the surface
area, which means that the analytes will have less bonded phase with which to
interact.
Surface area: Columns with high surface area may exhibit greater retention,
loading capacity, and resolution. However, low-surface-area columns have their
advantages. They equilibrate between runs more easily, which can be particularly
useful in gradient HPLC. Also, the reduced porosity results in better kinetics,
meaning that there is less dispersion in the column.

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

Si O Si
O
Si O H

C8 bonded phase

O
Si O Si
O
Si O H

C8 bonded phase

O
Si O Si

TMS group

O
Si O H

Lone acidic silanol

O
Si O Si
HO
Figure 1: Diagram showing various bonded phase groups, including the trimethylsilyl (TMS) group resulting from endcapping with trimethylchlorosilane.

Temperature limit: Manufacturers will often quote an upper temperature limit,


which is the highest temperature at which the column can be operated without
damaging the stationary phase.
pH range: This is the working pH range of the column. Pure silica has a working
pH range of 2.57.5; outside of this range, the silica will be hydrolyzed. At low
pH, hydrolysis of the silyl ether linkage between the bonded phase and silica
surface can also occur, resulting in a loss of both retention and loading capacity.
The bonded phase can act as a protective covering for the silica, but in general,
at high pH the silica surface will eventually hydrolyze. These problems can
both be exacerbated when operating at higher temperatures, especially as the
temperature limit of the column is reached.
Endcapping: The endcapping process covers surface silanol species, which
would otherwise cause unwanted secondary interactions and poor peak
shape, particularly when analyzing polar or ionizable species. To endcap a
column, the surface silanols are reacted with a small silylating reagent such as
trimethylchlorosilane, which produces an endcapped trimethylsilyl (TMS) species,
as shown in Figure 1.
Carbon load: Carbon load (%) describes the amount of ligand bonded to the
surface. It also describes the background carbon load that is present if using
unmodified silica. In general, the higher the carbon load, the lower the number
of surface silanols. It should be noted that that not all C18 columns will have the
same percent carbon and columns with different endcapping groups cannot be
compared, because endcap groups contain different numbers of carbon atoms.
Surface coverage: Surface coverage is a better measure of retention or the
hydrophobicity of a column. It is defined as the mass of stationary phase per
unit area, which is bonded to the support, and is expressed in units of mol/
m2. As can be seen in Figure 2, with high surface coverage there are fewer free
surface silanols with which analytes can interact to cause unwanted secondary
interactions. If there is lower surface coverage there will be more surface
silanol groups available to the analyte, which will ultimately result in different
interactions between the analyte and stationary phase. However, in some cases
such interactions could be advantageous if a change in selectivity is desired for a
separation.

Secondary Interactions
Silica is often referred to as type A or type B silica or type 1 and 2 silica. The
difference between the two types relates to the manufacturing process and
the resulting purity of the silica produced. Type 1 silica is manufactured by

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

polymerizing a metal silicate molecule, which results in high metal content in the
final silica that is produced. The metal atoms will tend to migrate to the surface,
where they are energetically favored. At the surface they affect the acidity
and, hence, the reactivity of the silica, increasing the strength of the secondary
interactions, which is very noticeable with basic compounds. Type 2 silica is
produced using an organosilicate monomer and therefore has less metal content;
this type of silica is less acidic and less reactive toward basic compounds. It is not
possible to say that one of these types of silica is better than another unless the
analytes are also discussed in the same context.

High surface coverage High ligand density

CH3
H3C Si CH3
O
O

Si

H3C Si CH3
OH

Si

O
O

Si

As well as type 1 and type 2 silicas there are also different forms of silanol groups
that exist at the surface. Different types of silanol species on the surface can
interact to different degrees. For example, acidic lone silanols will cause the
most peak tailing with basic analytes. A hydrated silanol will not induce much
interaction because it is lower in energy. Some examples of the different forms of
surface silica are shown in Figure 3.

CH3
H3C Si CH3
OH
O

Si

O
O

Si

H3C Si CH3
OH

Si

O
O

H3C Si CH3
OH

Si

Si

O
Si

H3C Si CH3
OH

Si

OH
O

Si

O
O

Si

Low surface coverage Low ligand density

Types of Solid Support


H3C Si CH3
OH
O

Si

O
O

Si

CH3

CH3

H3C Si CH3

H3C Si CH3

OH
O

Si

O
O

Si

OH
O

Si

CH3

O
O

Si

H3C Si CH3
OH

Si

OH
O

Si

OH
O

Si

O
O

Si

H3C Si CH3
OH

Si

OH
O

Si

Figure 2: Diagrams showing high surface coverage with high ligand


density (upper diagram) and low surface coverage with low ligand
density (lower diagram).

Bridged (vicinal)
Vicinal hydrated
Lone acidic

Surface metal ion

Metal activated

Geminal
Figure 3: Silica surface silanol groups.

O
O

Si

Advancements in solid support are helping ensure faster and more efficient
HPLC. They include the following supports:
Coreshell: Coreshell particles have a solid silica core and a porous outer layer.
In comparison to traditional fully porous silica supports they produce faster and
more efficient chromatography. They also have a narrow size distribution, which
can contribute to increased chromatographic efficiency.
Monolithic silica rods: Monolithic silica rods allow for high-speed separation with
good resolution and shorter analysis time. These supports contain macropores
that are greater than 50 nm in diameter and mesopores that are 250 nm in
diameter. This structure allows separations to be performed at very low back
pressures and at high mobile-phase linear velocities, or with samples that are
viscous. Monolithic silica rods are also good for direct injection of dirty samples
of plasma or food extracts. Because of the increased flow rate, analysis time is
also reduced.
Fully porous silica (traditional silica): Fully porous silica has a high surface
area and excellent mechanical strength. It can be used as a support material
for normal-phase chromatography, and with surface modification it can be
used for reversed-phase chromatography. As previously stated, one of the
major drawbacks of silica is its susceptibility to hydrolysis at pH extremes. One
way manufacturers have overcome this problem is to use organosilica hybrids.
An organo group grafted into the silica layers makes them more resistant to

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

dissolution at high pH, and this characteristic will extend the column life and
applicability in applications that require the use of high pH.
Porous graphitic carbon: This is a unique chemistry phase. Porous graphitic
carbon is composed of flat sheets of hexagonally arranged carbon atoms;
consequently it has no surface silanols and therefore, unwanted interactions will
not occur. Porous graphitic carbon phases have total pH stability, meaning that
they can be used over the full pH range. This wide applicability of pH makes
them ideal for the analysis of compounds where extreme pH levels are required
to drive the separation. This capability is very good for the separation of strong
acids and bases where the neutral form of the molecule may be required to
increase retention, which requires extremes of pH. This phase is very versatile and
can be used in reversed-phase LC, normal-phase LC, and hydrophilic interaction
chromatography (HILIC), and for LCmass spectrometry (MS) applications.

100

log k

10

0.1

Dependence of Retention Factor on pH


0

10

12

14

pH
Acetaminophen
Doxepin

Ibuprofen
Imipramine

Nortriptyline
p-Toluamide

Lidocaine

Figure 4: Plot showing the dependence of retention factor for various pharmaceutical compounds on pH. Mobile phase: 35% acetonitrile, 65% 20 mM buffer.

Nonpolar

Polar

N
C

OH
Si

Si

Si

Si

O O O

Alkyl

Dispersive

Phenyl

- interactions

Cyano

Electrostatic
/dipole

Silica

H-bonding

Figure 5: Structures of various stationary phases and the associated analyte


interactions.

The pH of the mobile phase is an important parameter for the retention of acidic
and basic compounds. As one changes the pH (Figure 4), it is possible to change
the ionization state of acidic and basic molecules; this renders them more or less
polar, which in turn affects their retention time. For basic compounds at a low pH,
the base can accept a proton to become positively charged. As the pH increases,
the protons in the surrounding environment are removed until eventually all the
basic protons within the analyte are abstracted, leaving a neutral species. When
the molecule is charged, there is little retention, but as pH increases, the neutral
form of the molecule becomes apparent, and retention is increased.
The opposite situation occurs for acids, which are proton donors. At low pH, the
neutral form of the molecule exists and, hence, the molecule will exhibit greater
retention. As the pH is increased above the analyte pKa, any acidic protons will be
removed from the analyte to produce a negatively charged species that exhibits
less retention in comparison to its neutral counterpart.
A good rule of thumb for determining the extent of analyte ionization is the 2
pH rule. For acids, at 2 pH units above the analyte pKa the analyte will exist in
the ionized (negative) form. Conversely, for basic moieties, adjusting the pH 2
pH units below the pKa will produce the ionized (positive) species. Therefore, for
ionizable molecules, retention can be altered and controlled by changing the pH
of the mobile phase.

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

We sometimes make the assumption that there is only one mode of interaction
in chromatography when actually there are multiple modes of interactions that
can occur simultaneously within a column. It is important to understand where
those different modes of interactions come from and that on some occasions a
separation scientist may want a particular interaction to drive a separation and
on other occasions that interaction may be undesirable. Thus, it is not possible
to say that a particular column is good or bad without describing the type of
compounds that are being separated.

CN phase
7

4,5

Phenyl phase

2
3

Initial Column Selection


and Analyte Functional Groups

So how do we go about selecting our column, given that there are no really
bad columns? To answer this we need to be able to fingerprint the retention
mechanisms of a column and better understand how they interact with the
molecules that we are trying to separate.

5
2
34

C8 phase
6

Time (min)

15

20

Figure 6: Chromatograms showing the shift in selectivity obtained


using cyano, phenyl, and C8 stationary phases.

C(2.8)

Figure 7: Column characterization plot. H = hydrophobicity, S = steric or shape


effect, A = hydrogen bond acidity, B = hydrogen bond basicity, C(2.8) = silanol
ionization at pH 2.8, C(7.0) = silanol ionization at pH 7.0.

10

A variety of modes of interaction potentially can exist between analytes and the
stationary phase:
Dispersive forces: These forces exist in all molecules and are the major retention
mechanism for alkyl phases. Retention is proportional to the hydrophobicity of
the molecule. This means that the more hydrophobic the molecule, the longer the
retention time.

H/10

C(7.0)/10

AnalyteStationary Phase Interactions

Charge-transfer (-) interactions: Charge-transfer interactions are prevalent in


both unsaturated and aromatic compounds, and greater retention is possible for
these compounds when a phase is used that exhibits these types of interactions.
Hydrogen bonding and dipoledipole interactions: As the polarity of the
analyte molecule is increased, different retention mechanisms need to be
investigated, such as hydrogen bonding and dipoledipole interactions. A polar
analyte interacts with the stationary phase through hydrogen bonding or a
dipoledipole interaction. Figure 5 illustrates the interactions based on phases
and modes.

Column Selection and Characterization


A change in selectivity can help change the retention mechanism and the elution
order of analytes. Figure 6 shows separations obtained using three phases:
cyano, phenyl, and C8. Differences can be seen in retention order, particularly for

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

H/10

the compounds that are eluted first. Some compounds are not eluted at the same
retention time from the various stationary phases and a degree of orthogonality
appears among these different phases.

H/10

C(7.0)/10

C(7.0)/10

We have talked about different modes of interactions, but how can we start to
quantify those modes? The Physical Quantitation Research Institute (PQRI) has
been trying to gain a better understanding of the different interactions that
molecules can have with the stationary phase. The radar plot shown in Figure
7 was generated for a Hypersil Beta Basic C18 column. This is the fingerprint or
characterization of this particular column. To get this information, it is necessary
to test individual columns under the same conditions, using identifiable test
probes throughout the testing regime.

Column Comparison
C(2.8)

Type A

C(2.8)

Type B

Both Type B

Figure 8: Column characterization plots for type A and type B columns (left)
and two type B columns. See Figure 7 for symbol identification.

Using the PQRI method of fingerprinting columns it is possible to compare and


contrast different column chemistries to assess which retention mechanisms
dominate and can be exploited to differentiate between differences in analyte
molecules. Figure 8 illustrates the difference between type A and type B silica
(both from the same manufacturer). The type A silica is made with sodium silicate
monomer, which has a high metal content; this metal content increases the acidity of
the surface silanols and thus may promote secondary interactions with basic analytes.
In comparison, the type B silica is manufactured from an organosilicate, which has
a very low metal concentration. As a consequence, the surface silanol activity at
pH 2.8 is markedly different. With the more acidic silanols, greater interaction of
positively charged analytes can occur, whereas, with the high-purity silica, these
types of interaction will be reduced.

Common Stationary-Phase Types


Some common stationary phases used in chromatography include the following:
C18 or octadecylsilane (ODS): This stationary phase is potentially the most
retentive alkyl phase and is used for 7080% of all applications.
Silica: Silica is used for normal-phase chromatography or HILIC. This stationary
phase is ideal for polar molecules.
Cyano: Cyano phases can be run in both normal-phase and reversed-phase
modes, but care must be taken when switching between these two modes to
ensure that both the column and HPLC system are suitably equilibrated with the
new mobile-phase composition.

11

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

Amino: Amino columns offer a variety of modes of interaction. In HILIC mode,


amino or amide phases are very good for sugar analysis, but they can also be run
in reversed-phase and normal-phase modes.
Phenyl and pentafluorophenyl (PFP): These stationary phases are predominantly
used for analyzing polar and moderately polar compounds.
Diol: Diol phases are commonly used in reversed-phase and normal-phase
separations, but are being used more frequently as HILIC phases.
Anion exchange: These stationary phases are good when trying to retain organic
acids.
Porous graphitic carbon: Porous graphitic carbon can be used for normal-phase
and reversed-phase separations, as well as in HILIC applications. These phases
are very good for separating extremely polar compounds.

Physical Properties of Columns


The physical properties of a column need to be considered when selecting a
column for a particular application. Some of these properties are
Particle size: A smaller particle size equates to better resolution; however, there
is a compromise: the smaller the particle size, the higher the back pressure in a
column. Efficiency is inversely proportional to particle size; therefore, if particle
size is decreased, efficiency will increase.
Length: Increasing the length of the column increases resolution; however, by
doubling the column length (which will double analysis time and increase the cost
of the column) a gain in resolution of only 1.4 times is achieved. It also should be
noted that increasing column length can alter analyte selectivity under gradient
elution conditions.
Internal diameter: Reducing the internal diameter of the column reduces the
flow rate that is required to reach the optimum linear velocity. If the absolute flow
rate is maintained, the back pressure will increase as column diameter decreases.
Maximize sensitivity: The sensitivity of an analytical separation can be improved by
adjusting various column and method parameters, including reducing the column
length and internal diameter, using smaller particle sizes (to increase the efficiency
of the separation), minimizing extracolumn volumes, and increasing the flow rate.
Sensitivity can also be increased by decreasing the background noise from other
matrix components by using appropriate sample preparation techniques.

12

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

Mass loading considerations: The amount of sample that can be loaded onto
a column is dependent on the column dimensions and stationary phase type.
Loading an excess of sample onto a column will result in poor peak shapes (broad
peaks, change in apex retention time, and fronting or tailing peaks) and will
ultimately decrease resolution.
Peak capacity: This parameter is important in modern HPLC and describes the
number of components that can be successfully separated with a given column
under gradient conditions. Peak capacity (P) is calculated using equation 3. The
peak capacity can be optimized by changing the gradient time as a function of
flow rate.

tg
P1+ w

[3]

where tg is the gradient time and w is average peak width.

Summary
It has been shown that numerous parameters pertaining to the stationary phase
and dimensions of an HPLC column should be considered to select the correct
column for a particular application.

This article is based on the LCGCCHROMacademy web seminar, Critical


Choices in HPLC Selecting Column Stationary Phase and Dimensions
presented on March 20, 2014, by Tony Edge and Dawn Watson.

Tony Edge, PhD, is a Scientific Advisor for Chromatography Consumables at


Thermo Fisher Scientific in Stockport, UK.
Dawn Watson, PhD, is a CHROMacademy Technical Expert with Crawford
Scientific in Strathaven, Lanarkshire, UK.

13

COLUMN SELECTION

SELECTING COLUMN
STATIONARY PHASES
AND DIMENSIONS

Samples are complex.


Separating them shouldnt be.
Every breakthrough starts with a challenge. We believe that challenge should be your science, not your
instrument. The Thermo Scientific Vanquish UHPLC delivers better separations, more results, and
easier interaction than ever before. In 2010 we embraced UHPLC as the standard for all of our liquid
chromatography solutions, and we have designed the Vanquish UHPLC as the instrument to solve your
chromatographic challenges and achieve that breakthrough.

Vanquish UHPLC System


Discover more at thermoscientific.com/Vanquish

Thermo Scientific
Accucore Vanquish Columns
1.5 m solid core particles for
unmatched resolution and throughput

2014 Thermo Fisher Scientific Inc. All rights reserved. All trademarks are
the property of Thermo Fisher Scientific and its subsidiaries.

Leading Separations for


Mass Spectrometry
Providing that extra level of confidence
with seamless integration

Thermo Scientific Dionex Chromeleon


Chromatography Data System
Integrated mass spectrometry support for
single point LC-MS control

Factors to Consider
By Dwight R. Stoll and Scott Fletcher

SPONSORED
Tune Your Mixing Volume for
Gradient Generation
Click to
view PDF

SPONSORED
Eliminating Delays Caused by
Column Wash and Reconditioning in Gradient Methods
Click to
view PDF

High-performance liquid chromatography (HPLC) separations using gradient


elution generally are more powerful than those performed using isocratic elution.
Gradient elution is more complex, however. This article provides the essential
information for understanding gradient elution and how to use it, including
how to account for dwell volume, determine the washout volume, calculate peak
elution and column reequilibration times, minimize drifting baselines, and how
to implement an isocratic hold. It also explains the various type of pumps used
in gradient separations and how to test the performance of your formed gradient.
It also explains the benefits of running a scouting gradient, which is the most
important step in developing any method to account for the wide polarity of
analytes. Lastly, tips are provided for effective method transfer of gradient methods.

Isocratic Versus Gradient Elution


First we are going to compare isocratic and gradient elution from the perspective of characteristics of
these separations. Figure 1 shows an example of an isocratic separation of a relatively simple mixture using
a mobile phase composed of 30% acetonitrile, the strong solvent. Some of the hallmarks of an isocratic
separation are that the early-eluted peaks are not resolved nearly as well as the peaks eluted midway
through the analysis. We see increasing peak widths with increasing retention time; one of the phenomena
that accompanies increasing width is decreasing peak height, which leads to poorer detection limits
and resolution for later-eluted analytes. We also have a relatively long analysis time because of the late
elution of the highly retained compounds and, especially with complex samples, we have the potential for
contamination of the column itself by the strong retention of highly retained components in the sample.
If we then look at a typical gradient elution chromatogram shown in Figure 2, the key difference compared
to the isocratic elution is that the solvent composition is changed during the run. In this case, we are starting

15

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider

initially at 20% acetonitrile in the mobile phase and then moving to 60% in a
linear gradient over 30 min. One of the key differences that results is that we have
improved resolution, for both the early- and late-eluted compounds. Also, when
we have analytes with very diverse chemistries, we have increased or improved
detection capabilities, because now the later-eluted compounds have much
narrower peak widths and therefore much higher peak heights.
We also have an increased ability to separate complex samples, mainly because
we can spread the peaks out better and because on average they have narrower
widths. This approach can translate to a shorter analysis time. Because the mobile
phase has the ability to elute strongly retained compounds at the end of the run,
column deterioration from the retention of those compounds is avoided.

1,2

One of the potential downsides of gradient elution is that the instrumentation


required tends to be more expensive. There is also a potential for precipitation
of buffer salts at the interface where the two solvents are mixed to produce
the gradient and for a change in mobile-phase composition over time.
Reequilibration of the column following the gradient separation inevitably
increases analysis time and differences between the pumping systems used in
different instruments can cause difficulty when transferring methods.

1
0

25

50

Gradient Elution Applications

75

Time (min)

Figure 1: An example of an isocratic


separation of a relatively simple
mixture of herbicides using a mobile
phase composed of 30% acetonitrile
in water where the solvent composition stays the same over the entire
run. Peaks: 1 = tebuthiuron, 2 =
prometon, 3 = prometryne, 4 = atrazine, 5 = bentazon, 6 = propazine, 7 =
propanil, 8 = metolachlor.

10

15

20

25

30

Time (min)

Figure 2: Example of a gradient


elution chromatogram of the same
sample mixture analyzed in Figure 1,
where a 2060% acetonitrile gradient
is used during the run.

Some of the common applications of gradient separations include rapid scouting


runs during method development to get a sense for how the compounds in
the sample are behaving. Gradient elution is also very effective for removal of
strongly retained compounds and interfering compounds in the sample. This is
the major reason why many chromatographers use gradient elution it is just
too risky to perform isocratic work on a sample that you dont know very well
because some of the analytes may remain in the column.
We also use gradient elution with low-concentration analytes, particularly when
those compounds are dissolved in a weak solvent, such as in the case of using
reversed-phase LC with a weak solvent like water. For example, it is possible
to inject extremely large volumes of sample into a reversed-phase column and
essentially preconcentrate, or focus, the analyte at the inlet of the column, which
can significantly improve detection limits.
It is also true that for large molecules, such as polymers of various kinds, including
peptides and small proteins, retention has a very strong dependence on the
composition of the mobile phase. In these cases, gradient elution is required;
otherwise it is very difficult to elute these compounds from the column, which can
lead to irreversible retention of those compounds. This relationship is exemplified

16

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider
in Figure 3, which is a plot of log of retention factor, k, versus the composition of
the mobile phase expressed as a ratio, .

100

As can be seen in the figure, for a rather small simple molecule like benzene,
the retention of that molecule is reduced as we increase the amount of organic
solvent in the mobile phase, but that change is rather slow compared to a
peptide like enkephalin, which has a much steeper slope. For a small protein like
lysozyme this dependence becomes very strong, and with a small change in the
concentration of organic solvent in the mobile phase, the compound is either very
highly retained or not retained at all. So this dependence of the retention of these
molecules on the mobile-phase composition is very important.

Leucine enkephalin
s = 11

Lysozyme
s = 40
Benzene
s = 2.7

As mentioned, one of the major benefits of gradient elution is the fact that
narrow peaks are obtained, where the peak width is nominally independent of the
retention time. So lets investigate this advantage in greater detail. A significant
factor is the focusing of the analyte band at the inlet of the column. Figure 4
includes plots of two analytes and shows how they are affected during a gradient
separation below the column diagram. The top one shows the distance that the
analytes travel in the column as a function of time and the bottom plot shows the
retention as a function of time.

10

0.14

0.18

0.22

0.26

0.30

Benefits of Gradient Elution

0.34

0.38

0.42

Figure 3: A plot of retention factor versus the composition of the mobile


phase, showing that larger molecules are more sensitive than small molecules
to changes in the percentage of the organic components.

These two plots provide different perspectives on how the analytes are behaving
inside the column. But the conclusion is that when the elution strength of the mobile
phase is low, the analytes come into the column and basically stick at the column inlet
they have very high retention and very low velocity. As the elution strength of the
mobile phase increases, the retention of those compounds goes down, as shown in
the lower graph in Figure 4, and at the same time, their velocity increases.
A secondary effect that contributes to the narrow peak width is that the mobilephase composition in the column close to the analyte band is weaker than the
solvent composition thats coming behind the band. Thus, the mobile phase that
follows the analyte through the column tends to have a slightly higher elution
strength, which tends to give the analyte molecules in the tail of the peak a higher
velocity, whereas the solutes on the leading edge of the peak have slightly higher
retention and lower velocity. These factors again compress the band somewhat
and also lead to narrow peak widths.

17

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider

Gradient Delivery Pumps


High-Pressure Binary Pumps

Both high- and low-pressure pumping systems are used for gradient separations.
The first type, a high-pressure binary pumping system, is shown in Figure 5. In
the lower left and right parts of this figure are two independent pump heads.
One of them is pulling in solvent such as water from a bottle going through a
degasser and the other one is pulling in a second solvent, such as acetonitrile
or methanol. The solvent, or mobile phase, is then pumped out of these two
pump heads and mixed in a low-volume mixing chamber, where it goes through a
secondary mixture chamber and a pulse-dampening device to minimize pressure
fluctuations during the flow through the column.

20

30

40

50

60

70

80

90

100

Organic
modifier (%)

Distance (cm)

0 10
Start

Low-Pressure Quaternary and Ternary Pumps

End

14 min 22 min

Start

20

10

End 0

10

20

Time (min)

Figure 4: The focusing effect of an analyte as it moves through a column. The


upper plot shows the distance that the analyte travels through the column as a
function of time, and the lower plot shows the retention as a function of time.

18

Its important to emphasize that the solvents are mixed under high-pressure
conditions. This pump design is typically characterized by a low internal mixing
volume, which is a very important factor with respect to gradient dwell volume,
which is the volume in the system from the point where the gradient is formed to
the top of the column. But on the other hand, they tend to be more complicated
designs and typically are more expensive to purchase.

In contrast, the second approach is to use a low-pressure gradient pumping


system. Figure 6 shows schematic diagrams of low-pressure quaternary and
ternary systems. Functionally there is no difference between them; the choice just
depends on how many solvent options you need for producing the gradients. A
ternary system can mix up to three solvents to produce the mobile phase and a
quaternary system can mix up to four solvents to produce the mobile phase. In
this case, the mixing of the fluids happens before the point where the pressure of
the fluid is elevated to actually push it through the column.
The proportioning valve is frequently a bank of solenoid valves that open and
close at specified intervals to allow packets of solvent to enter the mixing point.
Figure 6 shows that these packets of solvent enter a single piece of tubing going
from the mixing point to the pump head itself; as these packets of solvent travel
through the pumping system they are gradually mixed, up to the point where
they enter the analytical column. Similar to the high-pressure system there is
also a pulse dampening unit and a secondary mixing chamber but the important
point here is that the solvent mixing happens at low pressure before it reaches
the pump head itself. However, because there is a greater volume of solvent
between the mixing point and the analytical column there is a larger gradient
dwell volume.

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider

Second mixing chamber

Pulse damper

Low-pressure systems tend to cost less than high-pressure mixing systems.


Mixing at low pressure can lead to complications, however, such as extremes
in flow rate or gradient composition, and can cause other potential problems
related to outgassing of the solvents.

Testing the Gradient Performance


When operating a pumping system designed for gradient elution, its important
to be aware of what tests can be used to characterize the performance of the
system and troubleshoot problems. These gradient performance tests can be
used to troubleshoot or evaluate the performance of specific components of the
pumping system and also to compare different pumping systems in terms of the
accuracy and precision of the gradient profile that is produced.
There are many different ways to test a systems gradient performance. Most
pumping systems have a built-in test that can be run using the instrument
software. One of the most common tests is shown in Figure 7, in which a
step gradient begins and ends at 0% of the B solvent. With a solvent mixture
composed of solvents A and B, a gradient is run from 0 to 100% B in steps of 10%
B, passing it through a system where the analytical column has been replaced
with a restriction capillary such as a long length of narrow tubing.

To autosampler

Low-volume
mixing chamber

Figure 5: Schematic of a high-pressure binary pump.

This test can be done in different ways with various solvents used as solvents A
and B. One common way to conduct this test is to use pure water for A, and then
for B, to use water spiked with some compound that absorbs UV light, such as
acetone or benzyl alcohol.
One good approach is to use a 50:50 mixture of methanol and water for these
tests. If you use pure water or a pure organic solvent, sometimes the test

19

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider
Proportioning valve

molecule will adsorb onto various instrument components. There are other
considerations, too. In the case of biological applications, for example, you
should use mobile phases that are similar to the mobile phases that actually are
going to be used in your application. And when your mobile phases consist of
highly aqueous solutions, benzyl alcohol may not be soluble enough; in such
cases, acetone, uracil, or thiourea would all be good alternatives.

Pulse damper

Ternary pumps

Quaternary pumps

Outlet valve

To autosampler

Figure 7 is also an indicator of how the mobile-phase mixture is detected at the


detector. The signal actually changes as a function of flow rate, and given that
we know that we are programming it in 10% steps, we can get some sense for
how the solvent mixing system is performing. This can then be used as a way of
troubleshooting or characterizing the performance of this system.

Inlet valve

Calculating Dwell Volume

Figure 6: Examples of low-pressure pumping systems: schematics of a ternary


pump (left) and a quaternary pump (right).

500

0.125

150

0.250
0.550

100

75

100

50

50

25

0
0

20

40

60
Time (min)

80

100

Figure 7: Plots of absorbance and %B versus time for a two-solvent


step-gradient test of pump performance (1).

20

%B

Absorbance (mAU)

Flow rate (mL/min):

Another important factor for characterizing the pumping system is calculating


the gradient dwell volume, because differences in the dwell volume can cause
significant changes in selectivity and resolution when transferring a method
from one instrument to another. The dwell volume is measured in a similar way
to gradient performance using an A and a B solvent, where the B solvent is
spiked with some compound that absorbs UV light. Then a gradient is run from 0
to 100% B, in a linear fashion (not using steps as in the determination of gradient
performance). The goal is to determine the length of the delay between telling
the instrument to start making the gradient and when the gradient, or the change
in solvent composition, arrives at the detector. This delay time is called the dwell
time. The delay volume, which is the volume of solvent that has to go through the
system before the solvent change actually reaches the detector, is equal to the
delay time multiplied by the flow rate.

Accounting for Dwell Volume


Figure 8 shows that differences in gradient dwell volume between instruments
can have an impact on resolution, particularly for closely eluted pairs of
compounds, as shown by the improvement in resolution of 1.6 to 1.2 between
systems A and B. One way to account for two systems that have very different
gradient delay or dwell volumes is to make the system with the lower dwell

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider
RS = 6.97

volume act like the system with the higher dwell volume by deliberately
programming into the pumping system control an isocratic hold at the beginning
of the run to effectively mimic the high gradient delay volume.

RS = 1.63

Washout Volume
0

10

15

20

System A: Dwell volume = 0.5 mL Gradient = 1% B/min

RS = 5.91

RS = 1.19

10

15

20

System B: Dwell volume = 5.0 mL Gradient = 1% B/min


Figure 8: Differences in gradient dwell volume between instruments can
have an impact on analysis time.

20

Absorbance (mAU)

-0

Flow rate: 1mL/min


A: Water
B: 0.1% acetone in water
Detection: 254 nm

-20
-40
-60
-80
-100
-120
-140
-160
0.0

0.2

0.4

0.6

Time (min)

21

0.9

1.0

Figure 9: Graphical
display of washout
time, which is the delay
in time from when
the pumping system is
programmed to change
the solvent composition
relative to when the
composition actually
changes. Adapted
with permission from
reference (2).

So far we have discussed the characteristics of the gradient profile that we can
test by carrying out the composition steps and looking at what happens at
the detector. We also talked about the dwell volume, which is the delay of the
gradient actually arriving at the column. Lets now turn our attention to what
happens at the end of the gradient.
Typically a scouting type of gradient proceeds from 10 to 90% B during the run.
At the end of the gradient, we make a step change from 90% B back down to
10% B to equilibrate the system and column for the next injection of sample and
the next gradient elution. Chromatographers should be aware that there is also
a delay in that process caused by the washout volume in the system. Although a
step change is made from 90% down to 10%, it doesnt happen immediately.
This is exemplified in Figure 9, which shows the delay when using two solvents, A
and B, where B is spiked, in this case water spiked with acetone. If a step change
from 100% B to 0% B is made at time 0, we see that there is a slight delay and
then an exponential flush of the B solvent out of the system.
This delay is measured using an approach similar to that used to measure the
dwell volume, and for the purpose of discussion, we characterize this washout
volume by looking at the time it takes for the B solvent to be 97% flushed out
of the system. This washout volume becomes important in determining or
estimating how much time we should allow for reequilibration of the analytical
column because we want to make sure that the analytical column is prepared
for the next run by flushing the final mobile phase composition out and refilling
it with whatever solvent composition we are using at the start of the gradient
elution run.

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider
Gradient slope
New system

Programmed
gradient

We can devise a way to systematically determine times that we should use for
these various factors when transferring a method from one system to another.
With respect to washout volume, we can look at the ratio of the washout volumes
on the two systems (see Figure 10). Equation 1 can be used to readjust our
expectations for how much time we need to allow for the last segment in the
gradient on the new system:
Gradient slope
Original system

Composition or response

80%
Gradient profile
Original system

Essential Gradient Parameters

Gradient profile
New system

Lets now turn our attention to optimizing essential gradient parameters and
in particular the benefits of running a scouting gradient. A scouting gradient
is probably the most important step in developing any method and makes it
possible to account for the wide polarity of analytes.

20%

Gradient dwell
New system

Wash out time


New system

Wash out time


Original system

Time (min)

Figure 10: Plots showing how the washout volume can impact the transfer of a
method from one system to another.

Final %B

%B

tg

Purging

Conditioning

Initial isocratic
hold

Reequilibration

Initial %B

Time

Figure 11: Essential gradient parameters to be considered in optimizing a method.

22

New segment time = original segment time X (original system washout volume/
new system washout volume) [1]

When we dont know how many compounds or the types of compounds we are
looking for, we need to understand the range of analyte polarities during the
method development process (the essential gradient parameters are shown in
Figure 11) so that we can encompass and retain as many of those analytes as
possible. And to give ourselves the best chance of capturing these analytes, we
use a scouting gradient for the most nonpolar analytes that starts at 5% B and
goes up to 100% B (that is,100% organic mobile phase); this gradient elutes the
most highly retained, nonpolar (hydrophobic) analytes and also provides the best
chance of retaining the more polar, hydrophilic, analytes. The information that
we gather from this initial scouting gradient is helpful in determining whether a
gradient is needed or whether the method should be run isocratically.
Isocratic runs will provide the best resolving power for analytes of similar
polarties and the best indication of whether the analytes are interacting with the
stationary phase as much as possible. So a scouting gradient run may indicate
that an isocratic run is recommended or it might suggest the use of a gradient
run because of the differing polarity of analytes. However, it will be extremely
difficult to pick an isocratic mobile-phase composition that will retain the highly
polar analytes and not retard the more hydrophobic analytes so much that the
peaks broaden or remain bound onto the stationary phase. If the scouting run
is advising the use of an isocratic mobile phase it can also tell us what mobilephase composition to use and, if a gradient approach is suggested, it will indicate
whether we can actually increase our initial and final organic compositions or
perhaps decrease them to save time.

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider

Calculating Peak Elution


The initial approach to use when calculating peak elution is to determine the
percentage difference between the first and the last peak retention times using
the following equation:
Peak elution range = ([tf ti]/tG ) X 100 [2]

Initial %B Eluent compostion


of first peak 10%B

1.0

2.0

1.0

10

3.0

4.0

2.0

5.0

3.0

Initial %B 5
Final %B 100
%B/min 1.9
Gradient time 50 min

15

6.0

7.0

4.0

Initial %B 20
Final %B 100
%B/min 1.9
Gradient time 40 min

8.0

5.0

Initial %B 40
Final %B 100
%B/min 2.0
Gradient time 30 min

Figure 12: Optimization based on changing the eluent composition of the


first peak in a chromatogram.

where tf and ti are the final and initial retention times, respectively, and tG is the
total time during which the eluent composition is changing. If that difference is
25% or greater, then we typically recommend using a gradient, whereas if it is less
than 25%, an isocratic run is usually optimal. If the analytes are eluted significantly
below the 25% threshold of the gradient, we want to know what isocratic portion
to run. To identify that portion, there are a couple of further calculations that can
be used to better understand the average retention time that is, the retention
time in the middle of the peak elution window. We also need to calculate the rate
of change of the organic component of the mobile phase (the speed at which
the mobile-phase composition is changing every minute). For example, in the
method described previously, if we change from 95% aqueous down to 0% over
20 min, the rate is about 4.75%/min. This rate can be calculated by dividing the
difference between the initial and final %B by the time of the gradient. We can
then use these two values to carry out further optimization studies of the gradient
parameters. For the sake of clarity, these equations will not be described but
instead we will provide a general overview of the optimization procedure.
Initially, we need to know the percentage of organic solvent in the isocratic
mobile phase. It can be determined by adding the initial %B to the amount
that the organic composition has increased by the time a peak is eluted, or by
the time the middle of that peak is eluted, if its an isocratic elution. If we then
multiply the average retention time by the rates of change of %B, the summation
of that plus the initial concentration tells us what mobile-phase composition the
pumps are pumping, which is a very useful parameter to know.
However, that composition is not what is passing through the column. We therefore
need to account for the delay or dwell volume. The way we do that is to convert
the dwell volume back to a time by dividing dwell volume by the flow rate, and
then multiplying that value by the rate of change in units of %B per minute. Then,
by subtracting the %B value obtained from the previous calculation from what the
pumps are pumping, we can determine what mobile-phase composition is passing
through the column at the time the analytes are detected. Because the analytes have
passed through the column and have been detected, we subtract 10%. Essentially,
we are calculating what mobile-phase composition is passing through the column
when the middle of that peak grouping is eluted, and then we take away 10%.

23

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider

If we are optimizing the parameters for a gradient analysis, we repeat the same
calculation twice, but rather than using the average peak retention time, we use
the retention time of the first peak to be eluted, and then we calculate when the
last peak is eluted. When we use the initial peak retention time, we obtain the
initial %B, and when we use the final retention time, we obtain the final %B.

Initial %B 10
Final %B 100
%B/min 1.5
Gradient time 60 min
0

10

Initial %B 10
Final %B 60
%B/min 1.43
Gradient time 35 min
0

10

Initial %B 10
Final %B 40
%B/min 1.5
Gradient time 20 min
0

10

Figure 13: Optimization based on changing the eluent composition of last


peak in a chromatogram. (Note that only the first 14 min of each separation
is shown.)

An example of this appears in Figure 12, which shows a series of chromatograms


with values for the initial %B ranging from 5% to 40%. These chromatograms are
showing just the first portion of that gradient. As the initial %B is increased, the
selectivity remains fairly constant, but the resolution is degrading and the peaks
are getting broader. If the gradient is overly compressed, the analytes dont have
sufficient time to interact with the stationary phase.
Figure 13 shows the same chromatograms, but in this case, the final %B has been
optimized. As the final %B is reduced from 100% through 60% down to 40% B,
the gradient time decreases from 60 min to 35 min to 20 min, respectively. The
peaks and peak spacing remain in proportion and constant, primarily because we
are keeping the rates of change the same. Thus, as we reduce the final %B, we
reduce the gradient time accordingly.
To scale a gradient, the average retention factor, k*, must be calculated. We
typically cant have a retention factor for a gradient, because we are always
changing the mobile-phase composition, so we use an average retention factor:
k* = tG F/SVm [3]
where F is the flow rate, S is the slope of a plot of log k vs. , is the fractional
change in the organic composition during the gradient, and Vm is the column
volume.

100% B
Steep

Shallow

We typically use the same range as with an isocratic separation, looking for a
retention factor somewhere between 2 and 10 with conventional HPLC systems.
However, for modern ultrahigh-pressure liquid chromatography (UHPLC) columns,
values of 0.55 are fairly typical.

100% B

tg = 20

tg = 5
100% B

0% B
100% B

tg = 40

tg = 10

S = 0.25MW0.5 [4]

0% B

0% B
10

10

20

30

40

Figure 14: Chromatograms showing the effect of gradient slope on resolution


and selectivity.

24

To estimate S, we use the following equation:

So we take the square root of the molecular weight of the analyte, which really
drives its S value, and then we multiply it by 0.25. As a rule of thumb, if you work
on anything less than a 1000 Da in size, an S value of 5 is a very good starting
point.

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider
Initial %B 10
Final %B 90
%B/min 1.333
Gradient time 60 min
Flow rate 0.5 mL/min
Column length 150 mm
Column i.d. 4.6 mm

Rs = 2.16

10

15

20

Initial %B 10
Final %B 90
%B/min 5.333
Gradient time 15 min
Flow rate 2.0 mL/min
Column length 150 mm
Column i.d. 4.6 mm

Rs = 1.99

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

Initial %B 10
Final %B 90
%B/min 13.333
Gradient time 6 min
Flow rate 5.0 mL/min
Column length 150 mm
Column i.d. 4.6 mm

Rs = 1.66

1.0

2.0

9.0

3.0

Figure 15: Chromatograms showing the effect of changing flow rate and
gradient time on selectivity and sensitivity.

Equation 3 can be rearranged to account for tG , which can be very useful if you
are actually trying to calculate what a gradient time should be. With a known flow
rate, an S value of 5, a of 0.95, and a column volume that has been calculated
using the standard column volume calculation, we can then use a k* value of 5
because we know what we are looking for. And for a standard 150 mm x 4.6 mm
i.d. column with a flow rate of 2 mL/min, we obtain a k* value of 5, which will
result in a tG of about 20 min.
Figure 14 emphasizes what can happen when the rate of change is too fast, or the
slope of the line is too steep. If the gradient time is too short, there is too much
compression of the analyte elution window. Alternatively, if we make the slope
too shallow, we are wasting time, as can be seen with the tG = 40 chromatogram
where there is a significant dead time in the separation.
When analyzing a multiple-component sample, you will find that analytes can be
affected to a different degree by changes in the gradient time. Its not always
the case that reducing the gradient time will improve resolution or increasing
the gradient time will improve resolution depending on the composition of
a sample, the optimal gradient time can be found somewhere in the middle,
which is contrary to the results obtained with isocratic separations. In gradient
separations, changing the gradient time can also change the selectivity, which in
turn changes the resolution. Arbitrarily changing the gradient time can affect the
separation of your samples both positively and negatively.

Column Reequilibration Times


Historically column reequilibration has been discussed in terms of column
volumes and multiple column volumes. A general rule of thumb for column
reequilibration is expressed as equation 5:
Required reequilibration time = 2(Vd + Vm)/F [5]

Figure 16: Plots showing differences in baseline absorbance when using


methanol and acetonitrile as the organic solvent in a gradient run.

25

Where Vd is the dwell volume of the system. This rule of thumb is an incredibly
useful guide for estimating the reequilibration time that is required post-gradient.
An important parameter to remember is that a run time is not purely the gradient
time; it is a summation of the gradient time plus reequilibration time. It should always
be determined empirically. Although equation 5 provides a good estimate for the
required reequilibration time, you should always ensure that your analytes are not
affected by insufficient equilibration. Irreproducible retention times can be caused by
giving the column insufficient reequilibration time before the next injection.

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider

Method Transfer
Now we are going to discuss method transfer and translation in terms of flow
rate, length, and column internal diameter. Previously we talked about gradient
time and column flow rates. Changes in the flow rate can affect resolution and
selectivity. If you want to maintain selectivity, k* should remain the same for
the analytes and therefore resolution is maintained as much as possible. If the
flow rate is doubled, for example, the same k* value (sometimes referred to as
B value) can be maintained by halving the gradient time. If you want to maintain
selectivity, the equation must be balanced by making a proportional change to
the gradient time, as we did for the flow rate, and vice versa.
Figure 15 shows that as we go from a 60-min gradient in the top run, to 15 min
in the middle run, and down to 6 min with the bottom run, the resolution will be
affected. This order of magnitude reduction in run time can be accounted for
and selectivity can be maintained by ramping up the flow rate by an order of
magnitude. Yes, the efficiency has been lost, but selectivity is good and actually
the resolution will be quite adequate in most cases.

190
170
150

130
110

Optimal
range

90

70
50

20

40

60

80

100

120

140

160

tg/t0
Figure 17: Plot of peak capacity against the ratio of gradient time (tG) and the
unretained peak time (t0) showing the optimal range. Adapted with permission
from reference (3).

Changes in Column Length


Column length doesnt play as important a part in gradient analysis as it does
in isocratic analysis, because by the time the analytes reach the end of a 1015
cm column, they are actually residing purely in the mobile phase. As the mobilephase strength increases during a run, the analyte interactions with the stationary
phase will decrease, and as result they are traveling through the column at the
same velocity as the mobile phase. So the column length isnt as important as it
is in isocratic separations, where the analytes are continually partitioning in and
out of the stationary phase as they move though the column. For that reason,
separation or selectivity in gradient separations is driven by an analytes affinity
for the mobile phase as the mobile-phase composition changes.

How to Minimize Drifting Baselines


When there is an increase in absorbance or a change in the refractive index of the
more strongly absorbing solvents, the baseline will rise or drop during a gradient
run. This change in baseline absorbance will have an impact on the ability to
integrate precisely for quantification purposes, and it is one of the reasons
acetonitrile is often a preferred solvent. The plot of absorbance against time in
a gradient run shown in Figure 16 demonstrates that methanol is fairly strongly
absorbing, whereas the absorbance is fairly stable with acetonitrile over the same
time period.

26

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider
(95)

Peak Capacity
1

(5)

10

15

20

25

Figure 18: Chromatogram obtained using a 595% B gradient. The critical


peak pairs 1 and 2 are unresolved.

(95)

Peak capacity is a term that has gained favor in recent years, predominantly
because of the power of modern UHPLC systems, which can resolve a greater
number of peaks in a gradient separation. Peak capacity is defined as the ratio
of the gradient time and the average peak width of the first and last eluted peak,
added to 1, which gives us the theoretical number of peaks that can be resolved.
It is our experience that the practical empirical number of peaks that can be
resolved is an order of magnitude lower than the theoretical number. However, it
is a good way of understanding the efficiency of a separation.
The gradient length for optimum peak capacity should be neither too short nor
too long. Figure 17 is a plot of peak capacity against the ratio of gradient time (tG)
and the unretained peak time (t0), often known as the holdup time. The optimal
range is the highlighted blue zone, where the peak capacity is highest. Very long
gradients provide little increase in peak capacity.

The Impact of Gradient Profiles


1

(5)

10

15

20

25

(88)

(51)

(33)

10

Figure 19: Adjusting the gradient shown in Figure 18 to optimize


separation of critical peak pair 1.

27

15

There is no question that the gradient profile can affect certain peaks, as
exemplified by the two critical peak pairs shown in Figure 18. There is almost
baseline resolution between the peak pairing 1 and only very poor resolution of
peak pair 2. The segmented gradient used for this separation allows control over
early and later portions of the gradient, but there are no really hard and fast rules
for when to implement the segment change.
So what happens when we slow the gradient down? Figure 19 shows the initial
gradient at the top and the gradient slowed down on the bottom. In this
example, the critical peak pair 2 is resolved by the slower gradient, but peak pair
1 is still fairly problematic. A much better approach is to incorporate an isocratic
hold and isocratic segments within the gradient.

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider

By using the method described earlier, we can calculate the mobile-phase


composition where those peaks are being eluted. Lets take a look at the critical
peak pair 1 in Figure 20. By subtracting approximately 10% and incorporating
an isocratic hold and turning off the separation for peak pair 2, we can improve
the separation. We calculated that the peak pair 1 could be best resolved at 52%
B, and in this case if we subtract 12%, those peaks are pulled apart very nicely.
We typically use an isocratic hold of two to three column volumes as an initial
approximation.

(95)

A good place to start is 10% less than where each critical peak pair is eluted and
hold for two to three column volumes. If that hold time is not long enough, hold
for slightly longer. If the mobile phase is too strong, try using a lower % B. This
approach is a little more complex than using a traditional linear gradient from
5% to 95% or 100% B, but it is not that complex; using the calculation described
earlier it is very easy and straightforward to implement.

(5)

Summary of Gradient Elution Method Development

(52)
(95)

(40)

(40)

(5)
0

10

20

30

Figure 20: Chromatograms showing the benefits of incorporating an isocratic


hold within the gradient elution of the sample from Figure 18.

28

The method development optimization process for a gradient separation can be


summarized in the following steps:
Run a blank gradient to ensure there are no problems with baseline drift.
Run a scouting gradient (5100% B) and estimate initial and final %B; or begin
with a 20-min gradient with k* = 5 when F = 2 mL/min, for a typical 4.6 x 150
mm column.
Optimize gradient steepness for the conditions found from the scouting
gradient.
Perform the separation and repeat to ensure correct column reequilibration.
Vary the gradient time to assess the effect on the analysis (vary by twofold or
more), and note any changes in the resolution of critical pairs.
Initial and final %B may need to be adjusted.
If further optimization is required, vary the solvent type, and then the column
chemistry.
Gradient steepness should be reoptimized following any changes in solvent
or column.
For ionizable analytes, variation in pH or temperature should be investigated
before changing column chemistry.
Complex gradients can be used if required to reduce analysis time or to
affect retention and selectivity.
After conditions have been optimized using the steps above, the analysis
time can be reduced by varying the flow rate, column length, or particle size.
Keep k* constant when changing the column flow rate or length to maintain
selectivity.

GRADIENT METHODS

GRADIENT HPLC:

GRADIENT METHODS

GRADIENT HPLC:

Factors to Consider

Final adjustment of the reequilibration time can be made to optimize overall


analysis time; optimize the separation empirically noting any changes in
retention behavior.
Ensure that dwell and washout volumes have been taken into consideration.
References
(1) S. Marten, A. Knfel, and P. Fldi, LCGC Europe 21(7), 371379 (2008).
(2) A. Schellinger, D. Stoll, P. Carr, J. Chromatogr. A. 1064 (2005) 143156.
(3) M. Gilar, A.E. Daly, M. Kele, U.D. Neue, and J.C. Gebler, J. Chromatogr. A 1061, 183192 (2004).

This article is based on the LCGCCHROMacademy web seminar, Gradient HPLC


10 Things You Absolutely Need to Know, presented on June 19, 2014, by
Dwight R. Stoll and Scott Fletcher .

Dwight R. Stoll, PhD, is an Assistant Professor in the Department of Chemistry at


Gustavus Adolphus College in St. Peter, Minnesota.

Scott Fletcher is a technical business development manager at Crawford Scientific,


in Strathaven, Lanarkshire, UK, and a senior tutor for LCGCs CHROMacademy.

29

Versatility unleashed
Experience our innovative and versatile LC solutions for more productivity with less effort. Our philosophy of UHPLC-bydesign creates a culture dedicated to developing unique technologies and workflows to simplify your daily lab work. Ultrahighefficient columns, sample-specific chemistries and detection techniques let you see more analytes than ever before.
Combining intuitive automation tools and our world-class Thermo Scientific Dionex Chromeleon software increases
your productivity beyond traditional concepts, enabling you to tackle the most challenging analytical demands with ease.

Giving you more!


thermoscientific.com/LC

NEW: Thermo Scientific Dionex


UltiMate 3000 Electrochemical
Detector

2014 Thermo Fisher Scientific Inc. All rights reserved.


All trademarks are the property of Thermo Fisher
Scientific Inc. and its subsidiaries.

NEW: Chromeleon 7.2 CDS


Software

NEW: Bio and Specialty Columns

NEW: Dionex UltiMate 3000


XRS and BioRS systems. All
UltiMate systems are fully
UHPLC-compatible.

HPLC Detectors
By Scott Fletcher

SPONSORED
Charged Aerosol Detection
(CAD) Bibliography
Click to
view PDF

A variety of detectors may be used with high performance liquid chromatography


(HPLC). This article explains the operating principles and the strengths and
weaknesses of various types of detectors, including UVvisible, diode array,
refractive index, and fluorescence detectors, as well as novel detection approaches
such as evaporative light scattering detection, charged aerosol detection, and
electrochemical detection.

The Ideal Detector

SPONSORED
Electrochemical Detection
(ECD) Bibliography
Click to
view PDF

Lets start by considering the properties of the ideal detector for high performance liquid chromatography
(HPLC). Ideally, we would like to detect the presence of everything in a sample, independent of anything
else thats going on in the background of either the mobile or stationary phase. For example, we might
have a situation where we would like to detect as many of the analytes in our separation as we possibly can.
Alternatively, in a slightly different scenario, we might need more-selective detection, when we want to
measure only the solutes of interest, and make invisible the presence of matrix components that we are not
interested in measuring.
Obviously, we would like the detector to be stable, and for its performance not to vary with changes
in temperature or mobile phase. In a perfect world, we would also like to be able to detect very low
concentrations of analytes. We also want our detector to have certain physical properties that will not
negatively affect the separation procedure. For example, we dont want the detector cell to increase the
volume, because this will cause dispersion of our chromatographic peaks and thus will not only make it more
difficult to maintain the quality of the separation, but also to ensure sensitivity and detection capability.
On the other hand, we also would like to be able to detect the narrow peaks that are associated with
increasingly high performance forms of chromatography, such as ultrahigh-pressure LC (UHPLC), where the

31

DETECTORS

T H E F U N DA M E N TA L S O F

HPLC Detectors

peak volumes may be extremely small. If the detector response time is too slow, it
may miss very sharp peaks that arise between the detector observation periods.
And finally, we would like the detector to be robust and easy to optimize.

Detector Figures of Merit


Its important to understand the terminology and the figures of merit used
in detector technology. One important concept is selectivity. If we use a
nonselective detector, such as a refractive-index (RI) detector, the property of
the analytes we wish to monitor must be as universal as possible, so that we can
detect the presence of whatever is eluted from the column, irrespective of its
structure or physical properties. Nonselective detectors are not very common,
however, because its very difficult to monitor one property covering all analyte
molecules one may encounter.
Selective detectors, on the other hand, respond to a specific property of
the analyte. Lets take a UV detector as an example. A UV detector requires
interaction between the UV radiation and the molecules of interest. If there is no
UV activity and the UV light just passes straight through the sample, then as far as
the detector is concerned, nothing is present.
The sensitivity of a detector defines how easily it can detect very small signals
above the background noise. At low analyte levels, the signal will be very erratic
and unstable, and will be difficult to measure with a high degree of precision
or accuracy. This is important, because when you first optimize a detector you
typically set it up so the noise level is minimal. In addition, sensitivity affects the
detection of your analytes. Its universally recognized that you cannot confidently
assign a signal unless its at least three times the average noise value. In fact, to
be rigorous with analytical quantitation, its also generally accepted that the limit
of quantitation should be an order of magnitude greater than the noise.
Lets now focus on the linear range of the detector. In a perfect world we would
like our detector to be linear forever in all directions. In other words, it would
have the capability of detecting one molecule of our substance above the noise
and then continue to be able to detect increasing quantities of that molecule,
and never run out of linearity even if we have an infinite number of molecules
reaching the detector. This scenario is not very realistic, and in the real world,
the detector gets to a point where it cant respond proportionately to any more
analyte signal. We need to know when that occurs; otherwise the detector wont
be counting the molecules correctly. This can potentially be very problematic, not
just in measuring the concentration of a molecule, but also in assigning the size of
a contaminant peak such as an impurity, because we are making an assumption
that the contribution of the analyte is proportional to the area of the peak.

32

DETECTORS

T H E F U N D A M E N TA L S O F

HPLC Detectors
Detection method

Additionally, if we try to detect above the linear range of the detector, then we
overestimate the quantity of any impurities, because we are not counting the area
of the main peak proportionately compared to the increase in the height of the
impurities.

Selectivity

Sensitivity

Refractive index

Low

15 g

Conductivity

Low

1050 ng

Medium

0.51.0 ng

Electrochemical

High

50500 pg

Fluorescence

High

10100 pg

Evaporative light scattering

Low

0.11.0 ng

Charged aerosol

Low

0.11.0 ng

UVvis

Table I: Selectivity and sensitivity for various HPLC detection methods

Inlet capillary
Mobile-phase
flow from
column

Chromatogram
Flow cell
window

Collimated light
from UVvis source

Detector
diode

Mobile-phase flow to
waste, second detector,
or fraction collector

Figure 1: Schematic of a typical UVvis flow cell.

33

Outlet capillary

When detector signal is plotted against concentration, the slope is typically used
to determine the sensitivity of the method and the intercept indicates the degree
of error within the method, which is a direct result of the background response.
However, this is an area of much debate when we start talking about what
constitutes the limit of detection and the minimal detectible amount against the
signal-to-noise ratio.
Table I shows the typical selectivity and sensitivity of seven commonly employed
detectors. As can be seen, the most selective detection methods typically are
the most sensitive. When we require that a detector be more selective, we are
effectively demanding an increase in the specificity of detection parameters,
and its very unlikely that all of these criteria would be met by anything in the
general background noise. In fluorescence, for example, you just dont set the
wavelength at which your compound absorbs; you also effectively couple that
with the emission wavelength. And the chances are extremely unlikely that any
given interfering molecule will have the same set of coupled conditions as the
analyte. Similarly, with electrochemical detectors, you can set the parameters
of the detector to observe only the electrochemical effect of the molecule of
interest, which will often be in a range that other background contaminants are
not responsive to.
But for a nonspecific, nonselective detector, such as an RI detector, noise,
temperature, and environmental changes may affect its performance, so it is quite
difficult to measure very small changes in concentration. Additionally, with some
detectors, particularly with low-selectivity detectors such as RI, its very difficult to
eliminate all the background effects that affect detection capability.

UVvis Detection
Lets now turn our attention to UVvisible, or UVvis, detection by first explaining
what happens in the flow cell. Figure 1 is a diagram of a generic UVvis flow cell
showing the liquid flow from the chromatograph arriving at the cell and passing
through the collimated light of the UVvisible source, which is in line with the
detector. We can use this principle to measure the difference between what is
going into the cell at the front end and what is passing through the cell and being
detected at the back end. This difference in the transmission of light can be
converted into an absorbance signal, which is shown here as the chromatogram.
This peak will be proportional to the concentration, so the more analyte

DETECTORS

T H E F U N D A M E N TA L S O F

HPLC Detectors

molecules of a substance that pass through the cell, the more light is absorbed,
and therefore the less that comes out at the back end, which results in a larger
peak appearing in the chromatogram.

UVvis Detectors: Quantitation


To apply UV-vis detection to quantitative analysis we first think about the fact
that absorbance is proportional to the analyte concentration. The Beer-Lambert
law basically tells us that absorbance is proportional to the concentration of
the sample and the pathlength of the sample cell. However, the pathlength is
typically fixed in the detector. Thus, we are effectively suggesting the BeerLambert law in fact says that absorbance is proportional to the concentration of
the sample passing through the cell.
However, if there are any interferences present in the sample or if the
concentration becomes sufficiently high, some of the light will be scattered
rather than being absorbed, and as a result, the law starts to break down.
With UV detectors installed on modern HPLC and UHPLC systems, the peak
absorbance should be in the order of 1.5 absorbance units (AU) or lower. Once
the absorbance exceeds that range, the Beer-Lambert law may not apply and
you may start to see nonlinear effects. So a general rule of thumb is to keep the
absorbance below 1.5 AU, by either reducing the concentration or the amount of
injected sample.
The molar absorption coefficient is a measurement of how strongly a molecular
species absorbs light at a given wavelength. This is a very useful property because
it allows us to translate this light absorption back to the concentration of a sample,
once we have calibrated the measurement using a reference material. If we dont
know the concentration, we can calculate it using a standard and then compare it
with an unknown concentration, based on its being the same molecule under the
same conditions. However, in the real world, we often dont know the value of the
molar absorption coefficient and we have to make the assumption that there will be
an equal response from each component of a sample, based on the likelihood that
for similar structural features, molar absorption coefficients are also similar.
We tend to use peak area for quantitation as opposed to peak height, because
in the real world, peaks dont always behave perfectly and peak area is a much
more robust measurement than peak height. For that reason, peak area is a much
better measurement to use, because it is more tolerant of changes in the actual
chromatographic separation.

Chromophores
UV chromophores give the molecule its UV activity. This activity is typically
electronic in nature, so the more mobile the electrons in the conjugated

34

DETECTORS

T H E F U N D A M E N TA L S O F

HPLC Detectors

system are, the easier it is to see good UV activity. Additionally, more highly
conjugated molecules will tend to absorb higher wavelengths, which translate
to lower energies of UV radiation. A general rule of thumb is that some solvents,
particularly acetonitrile, are transparent to UV light at 190 nm. With methanol
and some other common solvents, it is difficult to detect them below 220 nm. So
broadly speaking, to avoid seeing any significant effect from the background, we
should work above the 210220 nm range, particularly when running gradients
where a changing composition in the background of the solvent could lead to a
sizeable baseline drift.

Variable-Wavelength UVvis Detectors

Achromatic
lens

Vis lamp

UV lamp

Detector Optical
flow cell slit

254 nm

Holmium
filter

E1 / E2

240 nm

Grating

254 +
380 nm

Diode
array

240 +
320 nm

240

320 nm

In variable-wavelength UVvis detectors, the wavelength of interest is selected


by moving a monochromator. We start with a polychromatic light source, which is
a mixture of all wavelengths, and effectively filter out the wavelength that we are
interested in, using a diffraction grating. The grating allows only the wavelength
of interest to pass through the flow cell, which will give us information based
specifically on the absorption of that particular wavelength of light. This capability
is very useful when analyzing a suite of samples that dont have the same
molecular template and that would otherwise not be detected if other sample
components were present.

UVvis Detection: Advantages and Disadvantages


240

320 nm

320 nm

Figure 2: Schematic of a diode-array detector and chromatograms showing


how it can be used for detection at single or multiple wavelengths.

Lets sum up the advantages and disadvantages of UVvis detectors. They


are very sensitive and can be used for quantitation of unknown molecules.
In addition, they are ideally suited for gradient elution, and respond to many
analytes, providing they absorb at that wavelength. Their disadvantages are
that no structural information is generated, absorption is dependent on solution
conditions, and response factors have to be calculated, particularly when it
comes to impurity quantification. However, UVvis detectors are suitable for small
organic molecules such as aromatic hydrocarbons and for analyte molecules with
double bonds, because in such cases you are likely to see plenty of UV activity.

Diode-Array Detection
Lets now take a look at diode-array detection (DAD). With these detectors, you
are looking at all wavelengths that are passing through the flow cell, instead of
just one wavelength as occurs with a UVvis detector. There is no wavelength
separation before the detection process. The detector determines which
wavelengths are missing from the original input light source (in other words,
which wavelengths were absorbed by the sample), after absorption has taken
place. So with diode-array detectors, you dont just get an absorption signal from
your solute at a specific wavelength; you actually get real-time spectra from the
molecule. These principles are presented schematically in Figure 2, which shows

35

DETECTORS

T H E F U N D A M E N TA L S O F

HPLC Detectors
Absorbance (mAU)

This advantage of looking at multiple wavelengths is probably the biggest reason


why there has been such an increase in the use of diode arrays, particularly if
your analyte molecule has a unique spectrum, because it becomes a way of
identifying individual molecules. Of course if your molecules have very similar
spectra, the benefits are not so obvious. However, even if this is the case, there
is no downside to using a diode-array detector, because it can just be used
as a variable-wavelength detector, albeit with some sensitivity compromises.
Additionally, the cost of diode-array detectors has gone down and they have
become much more affordable.

Anisic acid
Optimum: Slit: 8 mm (16)
Signal: 255/30 Ref. 340/100

Analytical
wavelength

50

that DAD can be used for detection at single or multiple wavelengths, where
spectra can be dynamically obtained and stored for peak purity analysis, library
searching, or extraction of signals.

40
Reference bandwidth
100 nm

30
20
10
0

Reference wavelength
(290 nm + 50 nm)
340 nm

30 nm

Bandwidth at
50% peak height

220

240

260

280

300

320

340

360

380

400

Wavelength (nm)
Figure 3: Spectrum of an analyte molecule (anisic acid) showing how a diode-array detector can be used monitor both the analytical wavelength and a reference
wavelength at the same time.

Mirror

Xenon flash lamp

Emission
monochromator

Lens

Lens
Photomultiplier

Excitation
monochromator

Flow cell

Figure 4: Schematic of a typical fluorescence detector.

36

Its worth spending some time to understand how the response rate is optimized
for a diode-array detector. Basically, the faster you make the response time,
the faster the ability to respond to whatever species is coming through the cell,
and the more likelihood of increasing peak sensitivity. However, as the response
time goes down, the noise also goes up, so the overall sensitivity that results
from using a higher response factor may not be any better than using a lower
response factor, and may even be worse in some cases. Thus, to get the best
signal-to-noise ratio, these parameters have to be optimized based on the
chromatographic separation conditions and the detection capability required.
Generally speaking, on modern UPHLC instruments where you are using very
efficient chromatography and getting peaks that are 23 s in width, you rarely
get any better response frequency than 40 measurements per second, which
means you dont have to use anything faster than a response coefficient of 40 Hz.
Modern detectors go up to 240 Hz, but as soon you go higher than 40 Hz, you
can start to run into problems with noise.

Photodiode

Another important capability of diode-array detectors is that we can use a


reference wavelength to get a better understanding of what is going on in the
cell without the sample being present. For example, if you want to compensate
for background shifts caused by the mobile phase or other sample components,
another wavelength or range of wavelengths can be selected to investigate
those effects in the reference cell, enabling you to compensate for changes in
the sample. Generally speaking, a reference wavelength or wavelength range is
chosen that does not interfere with the absorbance of the analyte molecule, as
shown in Figure 3.
The biggest advantage with diode-array detectors is that simultaneous,
multiwavelength detection can be carried out very quickly. By careful setup
of a DAD system, you can detect and display all wavelengths at once, even if

DETECTORS

T H E F U N D A M E N TA L S O F

Internal
conversion

40
35
30
25
20
15
10

Excitation
spectrum

5
0
300

350

400

450

500

550

600

Wavelength (nm)

Ground state So

Figure 5: Example excitation and emission spectra (left) and a diagram of


electronic transitions (right) for an analyte.

Purge valve 2

Purge valve 1

Waste

Figure 6: Schematic of a typical refractive-index detector.

Fluorescence

250 nm

Excitation

S1

250

you dont want to look at all spectral information. For this purpose the most
important settings on a DAD are the detection wavelength and the bandwidth.
For example, you can choose a detection wavelength such as 250 nm and set the
bandwidth to 7080 nm. In this way, you will actually be detecting everything
that absorbs light at wavelengths ranging from 210 to 290 nm. This can be
problematic with quantitation in a mixture, but it gives you the best chance of
detecting any unknown components in the sample.

S2

Emission
spectrum

200 nm

Norm.

HPLC Detectors

However, caution should be exercised when using diode-array detectors for


the estimation of peak purity. Its true that diode-array detectors can detect
the presence of one component that is coeluted with another one. However,
that detection relies on there being a significant difference in the spectra. If the
coeluted peaks have structural features that are very similar to those of the main
molecule or to another solute in your mixture, its highly likely that you wont
see a significant difference in the spectra, and therefore the peak will look pure
when actually there is an impurity present. But you can search the spectra against
library reference spectra, and in this way DAD can be used as a semiqualitative
tool to confirm the identity of some components that have very characteristic
UV spectra. Additional limitations of diode-array detectors are that sensitivity is
usually lower than that of a single-wavelength detector and these detectors are
also susceptible to lamp fluctuations.

Fluorescence Detection
A schematic of a fluorescence detector is shown in Figure 4. The radiation
source is typically a xenon arc flash lamp, which flashes every 3 s, producing a
continuous spectrum of light from 200 nm to 900 nm. Radiation from the lamp
is focused by the first lens, then reflected by the mirror onto the excitation
monochromator grating, which disperses and reflects the emitted radiation. The
light is then split in the flow cell to allow light to reach both the reference diode
and photomultiplier tube. Before the light reaches the emission monochromator,
a cutoff filter removes light below a certain wavelength to reduce noise from
first-order scatter and second-order stray light. The emission monochromator
determines the wavelength range of light reaching the photomultiplier tube
where the incident photons hit the photocathode and generate electrons, thus
multiplying the signal.
The most important parameters to optimize in a fluorescence detector are the
excitation and emission wavelengths. The excitation wavelength can be taken
from the excitation spectrum obtained on a spectrofluorimeter. The optimum
emission wavelength is dependent on the particular instrument and compound.
Fluorescence detectors can be extremely sensitive, but they detect only

37

DETECTORS

T H E F U N D A M E N TA L S O F

HPLC Detectors

Eluent
only

Lamp

Photomultiplier

Eluent +
sample

Photomultiplier

Eluent
only

Lamp

molecules that fluoresce. Unfortunately, not many molecules fluoresce, so these


detectors have limited applicability. The types of molecules that fluoresce can be
broken down into organic and inorganic molecules, and some that intrinsically
fluoresce, such as the fluorophores. The most common one is fluorescein, which
is typically used as a fluoro tag. Because of its sensitivity as a fluorescence tag, it
is fairly common to actually bind it to analytes to detect and measure compounds
that dont naturally fluoresce. In addition to fluorescein, other common fluoro
tags include fluorescent dyes such as acridine, and also fluorescent proteins.
There are also inorganic fluorophores, such as lanthanide-based probes, and also
CdSe-based quantum dots.

Figure 7: Diagrams showing the basis of refractive-index detection.

As mentioned above, the sensitivity of any detector is not only related to the
intensity of the peak height but also the intensity of the signal noise. Very often
the noise drives down sensitivity and ultimately impacts the detection limit.
Figure 5 exemplifies this for a fluorescence detector. Here is a great example
using a second-order filter. We have a specific excitation wavelength. It can be
seen from the electronic transitions that photons travel from the ground state
to the excited state, and then relax back down to the ground state. This occurs
at approximately 450 nm, where we actually measure the signal. So it is actually
the emission spectrum and not the excitation response that gives us the secondorder separation of the peak from the interference and the background signal. In
this example, it can be seen that the excitation wavelength is within the UV range,
while the emission spectrum is much broader, less defined and usually far more
practical to measure.

Column effluent

Nebulizer

Nebulizer gas
(air or nitrogen)

Nebulizer
chamber

Drift tube
(heated-zone
evaporation stage)

Analyte

Photomultiplier tube
or photodiode

Light source

Light-scattering
cell

Single output
Amplifier

Figure 8: Schematic of an evaporative light scattering detector.

38

The main advantage of fluorescence detectors is that not only do you achieve
good selectivity (because only a small handful of molecules fluoresce), but
you also get high sensitivity, which means that only small sample volumes are
required. But of course, the selectivity of these detectors can actually be a
disadvantage, because of the fact that not many compounds naturally fluoresce.
In addition, this type of detector can be affected by temperature because of
the energy required and the additional collisions that take place, and because
were looking at excitation and relaxation. And both the excitation and emission
wavelengths have to be optimized; you cannot just label the excitation and
emission wavelengths to be used, as is typically done with a UV detector. Also,
these settings tend be very detector-specific; with fluorescence detection,
both the excitation and emission wavelengths have to be set on every different
instrument.

Refractive-Index Detection
Figure 6 shows a schematic that explains how an RI detector works. We see that
there are two cells. On the right hand side, we can see the light path passing

DETECTORS

T H E F U N D A M E N TA L S O F

HPLC Detectors

through two cells. We have a reference and a sample cell. Before the analysis,
both cells are flushed with the mobile phase. When the injection is made, the
valve is rotated and column effluent then passes through the sample cell, with
the reference cell being filled with just the mobile phase. This technique relies
on comparing the degree of bending or refracting the light between the mobile
phase and the mobile phase containing the sample. So when only pure mobile
phase is coming from the column, that light is perfectly balanced and there is no
signal. As soon as anything different is eluted from the column and into the flow
cell, the degree by which the light is bent changes; the change in refractive index
can be caused by a sample compound or just by a change in the mobile phase.
This process is shown in Figure 7.

Electrometer
Charge is drawn off
and measured by
a sensitive electrometer

Nebulizer and
impactor
HPLC
column
eluent

Signal out
Signal is directly
proportional to
quantity of analyte
in sample

Drying
tube

Ion trap
Negatively charged
ion trap removes
high-mobility particles

Gas
inlet

Collector
Analyte particles
transfer their charge
Large
droplets
to waste

Positive charged
transferred to
analyte particles
by charged opposing
secondary gas steam

Secondary gas stream


positively charged
by a high-voltage
platinum corona wire

Figure 9: Diagram of a charged aerosol detection system.

The main advantage of a refractive-index detector is that it detects everything,


so it is considered a universal detector. Therefore, it is particularly good for
the detection of nonionic compounds, analytes that do not have a UVvisible
chromophore, and molecules that do not fluoresce. However, it is the least
sensitive of all detectors. Another major drawback is that RI detection cannot
be used for gradient LC separations because the changes in the mobile-phase
composition make it impossible for the detector to compare the column effluent
to a reference. Another limitation of RI detectors is that they take a long time to
equilibrate. So if you are analyzing a polar compound by hydrophilic interaction
liquid chromatography (HILIC) mode using an RI detector, it has to be allowed
to equilibrate for the better part of a week between runs. Even then, it might
only work in the evenings and on weekends, because these detectors are so
temperature sensitive that with people coming in and out of the laboratory
and air conditioning going on and off, the detector signal is very unstable.
Thermocouples are used to compensate for these temperature changes, but they
are only partially effective.

Evaporative Light Scattering Detection


Evaporative light scattering detection (ELSD) and charged aerosol detection
(CAD) are very similar in nature. With these approaches, the column effluent
travels out of the column and then is nebulized, using an inert gas, to produce
an aerosol, similar to the initial process of electrospray ionization (ESI) mass
spectrometry. The mobile phase is evaporated into droplets to produce
nonvolatile particles of the analytes. As the light hits these particles, the light
is scattered to various degrees; the amount of scattering is determined by the
particle size, so the larger the particle size, the greater the scattering of light. This
principle is depicted in Figure 8.
ESLD is an excellent approach for analyzing many nonvolatile species, so it is fairly
universal in its applicability. It has very broad applicability, almost as broad as that

39

DETECTORS

T H E F U N D A M E N TA L S O F

HPLC Detectors

of the refractive-index detector. In addition, it can be used for analytes that dont
have any chromophoric properties, and unlike an RI detector, it can be used for
gradient separations. Its biggest drawback, however, is the fact that you cant use
it for volatile samples because they will be lost via evaporation in amongst the
mobile phase.
Additionally, the mobile phase must be volatile for this technique to work,
although this is not a huge drawback. Another challenge with these detectors is
that the signal does not respond linearly to the concentration.

Charged Aerosol Detection

Working
electrode

Reference
electrode

A similar type of detection to ESLD is charged aerosol detection (CAD), which


uses a nebulized inert gas to produce an aerosol to evaporate off the mobile
phase. An impactor is used to remove large particles, but rather than looking at
light scattering, as occurs in ESLD, we are looking at charge transfer processes.
A stream of charged gas (N2) is used to collide with the analytes, and the charge
is transferred to the analytes. The particles pick up charge according to their
surface area and as they enter the collector and electrometer, the signal is
measured. This process is shown schematically in Figure 9. The benefits of this
approach are that it covers a broad range of analytes and compounds with good
selectivity, and it provides reasonably high sensitivity with good dynamic range,
meaning that it can quantitatively respond to small components in the presence
of much larger ones in the same run. In addition, like ELSD, its also compatible
with gradient elution. However, it has similar limitations with volatile analytes.

Electrochemical Detection
Counter
electrode
Figure 10: Schematic of an electrochemical detector.

The last type of detection method we are going to look at is electrochemical


detection (ECD), which is shown in Figure 10. There are many variations of this
detection approach. However, they all have one thing in common: They measure
the property of an electrical current using three electrodes: a working electrode,
a counter electrode, and a reference electrode.
There are a number of different electrochemical detectors available on the
market. The most common, and the one that has the widest range in terms
of applicability, is the conductivity detector, which measures the magnitude
of the current within an applied electric field. It can be used with any organic
or inorganic compounds that are ionic in nature, including cations, anions,
zwitterions, strong acids, and strong bases.
Another type of ECD is the DC amperometric detection, which looks at an
oxidation or reduction reaction taking place on the surface of an electrode.

40

DETECTORS

T H E F U N D A M E N TA L S O F

DETECTORS

T H E F U N D A M E N TA L S O F

HPLC Detectors
Typical samples that are applicable to this type of detection include phenol,
hydroxybenzene, catechol, dihydroxybenzene and similar types of aromatic
functional groups. Other sample matrices that lend themselves to amperometric
detection are catecholamine, dopamine, and epinephrine.

Methoxy
Hydroxy

Nucleobase
Amine

Phenol
Catechol
Nucleoside
Quinol
Quinone

Glycoside
Thiol

A variation on the DC amperometric detection approach is integrated and pulsed


amperometric detection. However, it works slightly differently with regard to the
electronics. It also detects the current, but measures the current by integration
during a repeated potential versus time waveform. It is applied via a standard
or background current in a square-post wave, so its the frequency of the
pulsing that is typically measured. This approach is well suited to the analysis of
carbohydrates and related molecules, where good sensitivity and linearity can
be achieved. Figure 11 gives examples of the types of molecules and functional
groups that are well-suited to electrochemical detection.
Summing up the relative pros and cons of ECD, it is highly selective, with good
sensitivity and a linear range of approximately five orders of magnitude with
a very fast response time. However, the analytes have to be electrochemically
active. Electrode fouling is also fairly common, so some sample types are
not really suited for ECD because of this limitation. But applications like
catecholamine, natural products, and neurotransmitters lend themselves nicely to
electrochemical detection.

Carbohydrate

Figure 11: Structures of molecules and functional groups well-suited for


electrochemical detection.

This article is based on the LCGCCHROMacademy web seminar, HPLC


Detectors What, Where, When, and How, presented on January 23, 2014.

Scott Fletcher is a technical business development manager at Crawford Scientific,


in Strathaven, Lanarkshire, UK, and a senior tutor for LCGCs CHROMacademy.

41

The Only Universal LC Detector


Your Lab Will Ever Need
See What Other Detectors Are Missing

Unbiased Universal Detection

Charged aerosol detection is a revolutionary technology that will change the way you view

Charged aerosol detection has the flexibility to be used for a broad range of analytes in

every sample. This technique delivers consistent analyte response independent of chemical

many different matrices, opening new opportunities for broad discovery and enhanced

characteristics over a wide dynamic range while providing sensitivity at sub-nanogram

routine analysis.

levels.

Drugs, impurities, and contaminants


Biomolecules

Improve Inter-Analyte Response

Foods and beverages

An analytes response to charged aerosol detection does not depend on optical properties,

Natural products, supplements, and botanicals

light scattering or the ability to ionize. Chromophores, radiolabels, ionizable moieties, or

Specialty chemicals

chemical derivatization are not essential for detection. Charged aerosol detection is a

Surfactants and polymers

mass-sensitive technique that measures any non-volatile and many semi-volatile analytes.
Variance in inter-analyte relative response is minimal, whether analyzing small molecules

Easy Integration With Any LC System

or proteins. And, this technique is gradient compatible.

The new Thermo Scientific Dionex Corona Veo detector is designed to integrate into
any HPLC/UHPLC system. When combined with a UV, diode array, or mass spectrometer, it

25

Citric acid

provides an orthogonal and complementary detection solution, making it the ideal detector

Theophylline
Phenylalanine

for any laboratory.

Diclofenac

Progesterone

Naproxen
pA

Reliable Results Without Intricate Optimization

Propranolol

Charged aerosol

The Corona Veo charged aerosol detector delivers sensitive, universal response through
a simple yet flexible design, perfectly matched for applications with capillary, microbore,

-2
600

and analytical scale columns.

Naproxen

Propranolol
mAU

UV

Diclofenac
Phenylalanine

Progesterone

Citric acid

-50
0

Minutes

10

12

14

16

18

Six pharmaceutical agents with an excipient (citric acid) were fully resolved using gradient reversedphase HPLC and their responses measured first by UV detection and then by charged aerosol detection.
As can be seen, UV detection significantly underestimates the levels of most analytes.

Download an application guide or watch a video and see how


charged aerosol detection works: thermoscientific.com/Veo

@2014 Thermo Fisher Scientific Inc. All rights reserved. All trademarks are the property of Thermo Fisher Scientific and
its subsidiaries. Specifications, terms and pricing are subject to change. Not all products are available in all countries.
Please consult your local sales representative for details.

Potrebbero piacerti anche