Liquid chromatography produces a separation, but not a chromatogram. This lesson introduces the fundamental requirements for LC detectors and provides a comprehensive overview of the most widely used detection techniques, including optical, gas-phase, and electrical detection. Through theoretical explanations and illustrative examples, students gain insights into the strengths and limitations of each detection method, their compatibility with gradient elution, and their suitability for quantitative and qualitative analysis. Special emphasis is placed on the concept of detector-induced dispersion and how it affects chromatographic performance. Detectors may also provide information on the identity of the analyte (qualitative analysis), but this is largely the domain of hyphenated systems, most notably liquid chromatography – mass spectrometry (LC-MS). These are described in Module 3.10 and not covered in this lesson.
Learning Goals #
After this lesson you should be able to
Compare the key characteristics (e.g. sensitivity, selectivity, compatibility, destructiveness) of commonly used LC detectors.
Evaluate the impact of detector design (e.g. cell volume, time constant) on band broadening and resolution.
Distinguish between universal, selective, and specific detection methods based on analyte and bulk properties.
Recognize the practical implications of detector choice in different types of LC methods (e.g. isocratic vs gradient, polymer vs small-molecule analysis).
READ SECTIONS 3.9.1-3.9.3
Optical Detection Techniques
1. Detection in Liquid Chromatography #
In almost all cases, on-line detection devices are used, as these are infinitely more practical than collecting and characterizing fractions from the LC effluent.
A distinction can be made between single-channel detectors that monitor one property and multi-channel detectors, which monitor several or many properties simultaneously. A key example of the formal is a basic UV detector that monitors absorption at a single wavelength (e.g. 254 nm when using a low-pressure mercury lamp). An example of the latter is a diode-array UV detector (or DAD), which allows recording complete UV spectra (with typically 2^n diodes, where <em>n = 7 to 10) at high frequencies (up to 240 Hz in commercial instruments). Common configurations are shown in Figure 1.
Figure 1. Common configurations of LC systems, emphasizing the role of the detector. A) Basic setup with a single-channel detector; B) Basic setup with a multi-channel detector; C) Setup with multiple detectors placed in series; D) Example of a setup with two detectors in parallel.
Figures 1A and 1B show basic setups, with a single-channel and a multi-channel detector, respectively. Figure 1C illustrates that several detectors may be positioned in series in the flow stream. A case in point are the triple-detector setups that are frequently used for polymer characterization, in combination with size-exclusion chromatography. Typically, such a setup would involve a refractive-index detector and a viscometry detector (both single-channel instruments) and a (multi-channel) static light-scattering detector. Figure 1D shows that detectors may also be positioned in parallel. A typical example may consist of a (single-channel or multi-channel) UV detector and a mass spectrometer. Accurate flow splitting is a requirement for accurate quantitative analysis using such systems.
1.1. Requirements #
At the onset of Module 3.9 the properties of an “ideal” detector for LC are discussed. These can be summarized as follows.
Table 1. Overview of some requirements for detectors for liquid chromatography.
| Property | Comment |
|---|---|
|
High sensitivity |
The sensitivity is the slope of a signal-vs.-concentration or a signal-vs.-mass-flow calibration curves |
|
Linear response |
Ideally, the signal intensity is proportional to the concentration or the mass flow of analyte |
|
Broad dynamic range |
Ideally (but not necessarily), the response is linear across a broad “linear working range” |
|
Low noise |
Random variations in the signal reduce the precision and impair the detection of trace amounts of analyte |
|
Low detection limits |
These result from a combination of high sensitivity and low noise |
|
Uniformity |
A universal detector responds to all analytes in a uniform manner |
|
Selectivity |
A selective detector shows a much higher response for certain (classes of) compounds, than for other (classes of) compounds |
|
Specificity |
A specific detector shows a (high) response for certain (classes of) compounds, and no response for other (classes of) compounds. |
|
Predictable response |
If the response to specific analytes can be predicted, calibration for individual analytes may not be necessary. |
|
No response to mobile phase |
A detector that does not respond to the components of the mobile phase is compatible with gradient elution. |
|
Good ruggedness |
Ideally, a detector should not respond to changes in temperature, flow rate or pressure (on the column or in the detector cell). |
|
Minimum dispersion |
The detector should not contribute significantly to the overall band broadening. |
|
Non-destructive |
Non-destructive detectors allow the collection of separated fractions after the detector, or detectors to be configures in series. |
|
Robust and reliable |
Including robustness with regard to pressure build-up (e.g. due to a blocked exit capillary). |
1.2. Extra-column band broadening #
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Ut elit tellus, luctus nec ullamcorper mattis, pulvinar dapibus leo.
EXERCISE 1
If we assume that the various contributions to the observed (subscript “\text{obs}”) band broadening from the column, from outside the column (“\text{ec}”), the injector (“\text{inj}”), detector (“\text{det}”), and connections (“\text{con}”), are independent, we may add the respective variances, i.e.
If we want to make sure that the observed resolution width does not decrease by more than 5% due to extra-column band broadening, must the extra-column dispersion (\sigma_{\text{ec}}) be kept to less than 5% of the column dispersion (\sigma_{\text{col}})?
Since R_{\text{S}}=\Delta t / (2\sigma_{\text{obs,1}}+2\sigma_{\text{obs,2}}), a 5% decrease in resolution corresponds with a 5% increase in the peak with (\sigma_{\text{obs}}), or in mathematical terms \sigma_{\text{obs}} \le 1.05 \sigma_{\text{col}} or \sigma_{\text{col}} \ge 0.95 \sigma_{\text{obs}}.
From
\sigma_{\text{col}} \ge 0.95 \sigma_{\text{obs}}we obtain
\sigma^2_{\text{col}} \ge 0.91 \sigma^2_{\text{obs}}and
\sigma^2_{\text{col}} \ge 0.91 (\sigma^2_{\text{col}} + \sigma^2_{\text{ec}})This can be solved for the extra column variance (\sigma^2_{\text{ec}}).
Solving for the extra column variance (\sigma^2_{\text{ec}}) yields
\sigma^2_{\text{ec}} \le \frac{0.09}{0.91} \sigma^2_{\text{col}} = 0.10 \sigma^2_{\text{col}}or
\sigma_{\text{ec}} \le 0.32 \sigma_{\text{col}}Thus, the extra-column dispersion can be about one third of the column dispersion, a much greater tolerance than the 5% suggested in the question.
The detector dispersion is determined by several factors, including the detector-cell volume (V_{\text{det}}), the detector time constant (\tau_{\text{det}}), and the connection of the detector to the column, i.e.
where F denotes the volumetric flow rate and \theta_{\text{det}} is a constant. Opinions vary on the appropriate value of \theta_{\text{det}}, but a value of 4 may be reasonable. The other two factors should not be neglected. The detector time constant can be a limiting factor, especially in ultra-high-pressure liquid chromatography. Minimizing the length and (especially) the diameter of the connection capillary and ensuring perfect cutting surfaces and “zero-dead-volume” connectors is always crucial in LC.
Unfortunately, “ideal detectors” for LC do not exist. For example, a universal detector that does not respond to variations in the mobile-phase composition is an illusion. In following sections we will discuss whether and how specific detectors approach some of the “ideal” properties.
2. Optical detectors #
Optical detectors are among the most common devices employed in LC. In a refractive-index detector (RID) the angle of diffraction of a light beam at an interface is measured (see Figure 2A. If the interface involves a reference cell that contains mobile phase without analyte, the net influence of the analyte is measured. RIDs measure a bulk property (refractive index) and are near-universal. Main drawbacks are the relatively high detection limits and the high sensitivity of the refractive index to variations in mobile-phase composition and temperature. RIDs are incompatible with gradient elution and, therefore, unattractive for many modes of LC. A notable exception is the characterization of non-UV-absorbing polymers by size-exclusion chromatography (SEC, see Module 4.2), which is a strictly isocratic method.
Figure 2. Overview of several optical detectors. (A) Refractive index detector, (B) UV-vis diode-array detector (DAD), (C) Fluorescence detector.
In a light-absorption detector (Figure 2B) a beam of light is passed through a detector cell and the transmitted light (I) is measured. After conversion to absorption (A), a signal is obtained that is proportional to the concentration of analyte i (c_i).
where I_0 is the background intensity, \alpha(\lambda) the absorption coefficient of analyte i at wavelength \lambda, and l the path length of the detector cell. While the principle applies to any kind of light, only UV and visible light are commonly used for LC detection. This is due to favourable differences between absorbance of the mobile phase (high I_0) and that of the analyte solution (high \alpha(\lambda)) and the availability of suitable materials (usually quartz) to construct cell windows. When using a grating and a diode array (diode-array detector, DAD, illustrated in Figure 2B), entire spectra can be recorded at a high frequency (10 Hz or more).
Absorption coefficients are reliable physical constants, ensuring a long-term stability of the response, they vary dramatically between different chemical compounds. This can be favourable, since solvents such as water, acetonitrile and – to a lesser extent – methanol show little or no background absorbance. On the other hand, many potentially relevant analytes also show no absorbance in this region. Calibration is typically performed using individual-component standard, but (relative) response factors tend to be quite constant.
EXERCISE 2
An important consideration in UV detection is the UV cut-off, which specifies the wavelength below which a solvent shows too much background absorption. Place the following solvents at their appropriate place in the table. With the lowest cutoff at the top, and the highest at the bottom. To help you, the cutoff value was provided for some solvents.
This is correct! Water has the most ideal cutoff, well below 190 nm. Whereas acetone absorbs strongly until about 330 nm.
This is unfortunately not correct. Try again! Consider the molecular structures of these solvents. Water has a low cut-off, whereas acetone’s is very high.
In a fluorescence detector (FLD, Figure 2C) emitted light with a longer (emission) wavelength () than the incident beam (excitation wavelength, ) is detected at some arbitrary angle. A high angle (often 90o) is common, as it minimizes straylight. A limited number of analytes absorb UV (or visible light) to reach a higher excited states. After a rapid non-radiative internal conversion occurs to the first excited state, from where a radiative decay results in the emission of fluorescent light.
EXERCISE 3
Mark the possible cause(s) of increased noise levels (and, thus, increased detection limits) in case of UV-vis detection.
This is correct! Each of these problems can to a varying degree cause problems with noise levels and affect the detection limits. Even the pump!
This is not correct. Try again. Hint: MANY of these possible cases are in fact capable of increasing noise levels.
For those analytes that show native fluorescence, the process can be quite selective and highly sensitive and up to a certain level the response is linear. These properties are sufficiently attractive to warrant the effort of making various non-native-fluorescent classes of analytes responsive by performing chemical “derivatization” reactions (see Module 8.3).
READ SECTIONS 3.9.4
Evaporative light scattering detection
3. Gas-phase detectors #
In contrast to the optical detectors discussed above, a few contemporary LC detectors operate in the gas phase, after nebulization of the effluent and evaporation of the mobile-phase. If the evaporation is complete analyte properties are measured (if it isn’t a bulk property may be measured).
3.1. Evaporative light-scattering detection #
In an evaporative light-scattering detector (ELSD, Figure 3A), the remaining, non-volatile analyte particles scatter light from a continuous source (such as a xenon lamp) and part of the scattered light is detected. Almost all low-volatility analytes provide a response, but volatile compounds, such as common mobile-phase components do not. This renders gas-phase detectors compatible with gradient elution, although their response may depend on the mobile-phase composition.
Figure 3. Schematic overview of the (A) evaporative light-scattering detector (ELSD) and, (B) charged-aerosol detector (CAD).
Non-volatile salts or buffers should be used in the mobile phase. The response of an ELSD is distinctly non-linear, with the signal increasing exponentially with the analyte concentration. As a result, the top of a peak is amplified, whereas the edges are diminished. This makes peaks appear sharper than they really are, and renders plate-height measurements overly optimistic. Because of the low detection limits offered for non-UV-active analytes and the compatibility with gradient elution, ELSDs have gained popularity over the last decades. However, calibration is not trivial.
3.2. Charged-aerosol detection (CAD) #
In a charged-aerosol detector (CAD, Figure 3B) evaporation of the mobile phase is followed by adding excess nitrogen ions, produced using a corona discharge, to the particle stream. After removing excess gas ions, the charge on the aerosol is measured. CADs may provide even lower detection limits than ELSD, but the response is also non-linear, with curvature in the opposite direction. The result is an apparent flattening and broadening of the peaks.
READ SECTIONS 3.9.5
Electrical detection
4. Electrical detection #
We may distinguish between conductivity detectors, which monitor a bulk property, and electrochemical detectors, which monitor analyte specific redox properties.
The most common electrochemical detector (ECD) for LC is amperometric, i.e. a voltage is applied between a working electrode and a reference electrode to enforce an oxidation or reduction reaction. The measured electrical current is proportional to the concentration of analyte present. Many, mostly polar, compounds can be measured with an ECD and for compounds that show little or no UV activity electrochemical detection offers a valid alternative. ECDs lend themselves to miniaturization and are often the method of choice on chip-based separation systems. However, while many elegant research systems have been described, such chips have not made their way to routine practice. Fouling of the electrodes is one of the explanations for the limited popularity of ECDs in LC practice.
In conductivity detection a bulk property of the effluent is measured. This principle has found successful implementation in instrumentation for the separation of small ions, know as ion chromatography, where the conductivity of the mobile phase is suppressed after the ion-exchange column, using, for example, a membrane suppressor. This allows the concenration of analytes to be measured against a low background conductivity.
The capacitively coupled contactless conductivity detector (C4D) has been the subject of increased interest. It is a simple device, that can be used as a detector around a fused-silica capillary or, for example, a piece of PTFE tubing. As such, it has more applications than common detectors. In capillary separation systems it can be used anywhere along the column. In regular LC systems, it may, for example, be used to monitor the true composition of a gradient before it enters the column.
EXERCISE 4
Each of the following detectors: refractive-index detection (RID), diode-array UV/vis detector (DAD), fluorescence detector (FLD), evaporative light-scattering detector (ELSD), charged-aerosol detector (CAD), electrochemical detector (ECD) and capacitively coupled contactless conductivity detector (C4D), belongs with one of the statements in the table below. Put each detector on its appropriate place by filling in its acronym (e.g. “DAD”) in one of the fields.
Correct!
Try again!
Correct!
Try again!
Correct!
Try again!
Correct!
Try again!
Correct!
Try again!
Correct!
Try again!
Correct!
Try again!
Concluding remarks #
The figure below shows a categorization of main detection principles in terms of universal, selective and specific detectors. Detectors that measure a bulk property are in principle universal, although the response of different analytes may vary greatly. Analyte-property detectors, such as ELSD and MS, can also be quite universal (as indicated by the blue shading in the top bar), but are compatible with gradients.
Figure 4. Overview of LC detectors.
A number of niche detectors are briefly mentioned in Section 3.9.7 of the book. Among these, the sulphur-chemiluminescence detector SCD) and the nitrogen-chemiluminescence detector NCD) are specific for particular elements. Section 3.9.6 in the book provides a qualitative overview of the main detection methods around the following figure. As can be seen from the figure, no LC detector is perfect (i.e. fully blue).
Figure 5. Overview of strengths and weaknesses of some of the detectors discussed in this lesson.
