CHAPTER FOUR DETECTION 4.1 INTRODUCTION, 148 4.2 DETECTOR CHARACTERISTICS, 149 4.2.1 General Layout, 149 4.2.2 Detection Techniques, 151 4.2.3 Signal, Noise, Drift, and Assay Precision, 152 4.2.4 Detection Limits, 157 4.2.5 Linearity, 158 4.3 INTRODUCTION TO INDIVIDUAL DETECTORS, 160 4.4 UV-VISIBLE DETECTORS, 160 4.4.1 Fixed-Wavelength Detectors, 163 4.4.2 Variable-Wavelength Detectors, 164 4.4.3 Diode-Array Detectors, 165 4.4.4 General UV-Detector Characteristics, 166 4.5 FLUORESCENCE DETECTORS, 167 4.6 ELECTROCHEMICAL (AMPEROMETRIC) DETECTORS, 170 4.7 RADIOACTIVITY DETECTORS, 172 4.8 CONDUCTIVITY DETECTORS, 174 4.9 CHEMILUMINESCENT NITROGEN DETECTOR, 174 4.10 CHIRAL DETECTORS, 175 4.11 REFRACTIVE INDEX DETECTORS, 177 4.12 LIGHT-SCATTERING DETECTORS, 180 4.12.1 Evaporative Light-Scattering Detector (ELSD), 181 4.12.2 Condensation Nucleation Light-Scattering Detector (CNLSD), 182 4.12.3 Laser Light-Scattering Detectors (LLSD), 183 4.13 CORONA-DISCHARGE DETECTOR (CAD), 184 4.14 MASS SPECTRAL DETECTORS (MS), 185 4.14.1 Interfaces, 186 4.14.2 Quadrupoles and Ion Traps, 188 4.14.3 Other MS Detectors, 190 4.15 OTHER HYPHENATED DETECTORS, 191 Introduction to Modern Liquid Chromatography, Third Edition, by Lloyd R. Snyder, Joseph J. Kirkland, and John W. Dolan Copyright © 2010 John Wiley & Sons, Inc. 147 148 DETECTION 4.15.1 Infrared (FTIR), 191 4.15.2 Nuclear Magnetic Resonance (NMR), 192 4.16 SAMPLE DERIVATIZATION AND REACTION DETECTORS, 194 4.1 INTRODUCTION The detector for the first liquid chromatographic separations was the human eye, used by Tswett in his classic experiments [1, 2]. For many years quantitative and qualitative analyses were accomplished by the collection of fractions eluted from the column (e.g., Fig. 4.1a), followed by off-line analysis using wet chemical, gravimetric, optical, or other analysis techniques. The concentration of analyte in each fraction could be plotted against fraction number, as in Figure 4.1b, resulting in a crude chromatogram. Collection and analysis of fractions is time-consuming, usually degrades chromatographic resolution, and is generally inconvenient, so on-line detection has many benefits. Some of the first applications of on-line liquid chromatographic detection included the refractive index detector reported by Tsielius [3] and the conductivity detector by Martin and Randall [4]. However, it was the introduction of the UV detector in 1966 by Horv ´ ath and Lipsky [5] that gave HPLC its most widely used detector. The UV detector was further improved in 1968 by Kirkland [6]; a flow-through cell, plus use of a mercury lamp, resulted in a great improvement in sensitivity (1 × 10 −5 AU noise). Single-wavelength detectors (254 nm) based on this principle were by far the most popular through the 1970s—until they were subsequently displaced by the variable-wavelength and diode-array UV detectors. fraction number off-line analysis (a)(b) concentration Figure 4.1 (a) Open-column LC; (b) plot of concentration of analyte in each fraction from (a). 4.2 DETECTOR CHARACTERISTICS 149 The chromatographic detector is a transducer that converts a physical or chemical property of an eluted analyte into an electrical signal that can be related to analyte concentration. Gas chromatography (GC) was well into its second decade of popular use when HPLC began its rise to popularity. This gave GC a head start in detector development as well. Much effort was given in the 1960s and 1970s to adapt GC detectors for HPLC use. A number of such detectors were described, including flame ionization [7, 8], flame photometric [9], and electrolytic conductivity [10]. However, adaptation of GC detectors required elimination of the mobile phase through evaporation, and until the introduction of the electrospray interface for LC-MS, the liquid-to-gas conversion process was very unreliable. As a result GC-based detectors for HPLC were doomed by operational problems. It was only after the development of an efficient nebulizer that detectors relying on mobile-phase-elimination (LC-MS, evaporative light scattering, corona discharge, FTIR, etc.) became practical. Despite the lack of a universal, sensitive detector, such as the GC flame ionization detector, there are many HPLC detectors that have been successful for general or specific applications. This chapter describes the principles of operation of the most popular HPLC detectors, presents example applications, and—where appropriate—compares the advantages and disadvantages of specific detectors. 4.2 DETECTOR CHARACTERISTICS The second edition of this book [11] listed nine characteristics of an ideal HPLC detector: • have high sensitivity and predictable response • respond to all solutes, or else have predictable specificity • be unaffected by changes in temperature and carrier flow • respond independently of the mobile phase • not contribute to extra-column peak broadening • be reliable and convenient to use • have a response that increases linearly with the amount of solute • be nondestructive of the solute • provide qualitative information on the detected peak Of course, no detector possesses all of these characteristics, nor is likely to in the foreseeable future. However, those HPLC detectors that have been the most successful have many of the properties listed above. 4.2.1 General Layout The HPLC detector is positioned directly after the column (Fig. 4.2) so as to minimize post-column peak broadening (Section 3.9). The column effluent is directed into a detector flow cell, where detection takes place. For most detectors the mobile phase is maintained in a liquid state, detection occurs, and the mobile phase passes out of the detector cell to a waste container, fraction collector, or another detector. 150 DETECTION (a)(b)(c) (d )(e)(f) Figure 4.2 HPLC system diagram. (a) Mobile phase reservoir; (b) pump; (c) autosampler or injector; (d) column; (e) detector; (f) data system. The UV flow cell shown in Figure 4.3a has evolved into its present configuration, in which the mobile phase flows through in a Z-shaped path, and UV light enters the quartz window at one end of the cell. The differential absorbance of light is monitored at the other end of the cell by a photosensitive diode, is converted to an electrical signal, and eventually is presented by the data system as a chromatogram. Different detectors have different flow-cell designs, as described in the following detector-specific sections. Optical detectors, such as UV and especially refractive index detectors, often are sensitive to small temperature changes in the mobile phase, which cause changes in the refractive index, and thus the amount of light that is transmitted through the flow cell. To stabilize the temperature, the detector cell is mounted in a draft free, and sometimes temperature-controlled, location in the detector case. The detector commonly includes a capillary heat-exchanger to stabilize the incoming mobile-phase temperature. One popular configuration is to wrap the capillary tubing around the detector cell body and embed it in a thermally conductive sealant. Detectors that make their measurements in the liquid state can be susceptible to optical or electrical disturbances when bubbles are present. Thorough degassing of the mobile phase (Section 3.3) often is sufficient to prevent bubble formation within the detector. As an additional precaution a small back-pressure can be applied to the detector outlet tube to prevent bubble formation. We recommend that a spring-loaded check valve be purchased from an HPLC parts-supplier for this purpose. If such a device is used, be careful that it does not exceed the pressure limit of the cell. For example, use a back-pressure regulator that produces 50 to 75 psi of back-pressure with a detector cell with a 150 psi pressure limit. Some workers rely on the use of a piece of capillary tubing (e.g., 1 m × 0.005-in. i.d.) as a back-pressure regulator. But the pressure created by such devices is proportional to the flow rate, 4.2 DETECTOR CHARACTERISTICS 151 light out end cap quartz window gasket inlet outlet light in cell body light in light out cell body internal- reflectance coatin g (a) (b) Figure 4.3 UV-detector flow cells. (a) Typical construction of a Z-path flow cell; (b) light-pipe flow cell lined with internally reflective surface. so an increase in the pump flow rate can inadvertently create excessive back pressure and cause cell leakage or damage. 4.2.2 Detection Techniques There are four general techniques that are used for HPLC detection: • bulk property or differential measurement • sample specific • mobile-phase modification • hyphenated techniques 4.2.2.1 Bulk Property Detectors A bulk property detector can be considered a universal detector as it measures a property that is common to all compounds. The detector measures a change in this property as a differential measurement between the mobile phase containing the sample and that without the sample. The most familiar of the bulk property detectors is the refractive index detector (Section 4.11). Bulk property detectors have the advantage that they detect all compounds. The universal nature of bulk mobile phase 152 DETECTION property detectors can be a disadvantage, as well, because all components of the sample that are eluted from the column will generate a detector signal. This means that additional chromatographic selectivity may be needed to make up for the lack of detection selectivity. As a general rule, universal detectors are inherently less sensitive, since they rely on the difference between two large measurements (solvent vs. solvent + solute). 4.2.2.2 Sample-Specific Detectors Some characteristic of a sample is unique to that sample, or at least is not common to all analytes, and the sample-specific detector responds to that characterstic. The UV detector is the most commonly used sample-specific detector. It responds to compounds that absorb UV light at a specific wavelength. The distinction between bulk property detectors and sample-specific detectors is somewhat vague; for example, at low wavelength (<210 nm), organic compounds all absorb in the UV to some extent, so the UV detector becomes less selective and more universal at low detection wavelengths. Other common sample-specific detectors rely on the ability of an analyte to fluoresce (fluorescence, Section 4.5), conduct electricity (conductivity, Section 4.8), or react under specific conditions (electrochemical, Section 4.6). 4.2.2.3 Mobile-Phase Modification Detectors These detectors change the mobile phase after the column to produce a change in the properties of the analyte. Such changes include a specific liquid-phase chemical reac- tion with the analyte (reaction detectors, Section 4.16), a gas-phase reaction (e.g., corona discharge, Section 4.13; mass spectrometric detectors, Section 4.14), or cre- ation of analyte particles suspended in a gas phase (e.g., evaporative light-scattering detectors, Section 4.12.1). 4.2.2.4 Hyphenated Techniques Hyphenated techniques refer to the coupling of an independent analytical instrument to the HPLC system to provide detection, and often are abbreviated with a hyphen as LC-(plus the technique). The most common hyphenated technique is LC-MS (Section 4.14), where a mass spectrometer is coupled with an HPLC system. Other less widely used techniques are LC-IR or LC-FTIR (Section 4.15.1) and LC-NMR (Section 4.15.2). One can speculate at what point a detector is no longer considered a hyphenated technique. Certainly the first UV detectors were UV spectrophotometers fitted with flow-through cells, but now UV detectors are just another HPLC detector. Generally, when the detector becomes widely used and the price of the detector is in the same general range as the rest of the HPLC system, it can be considered a detector. Thus it is reasonable to predict that mass spectrometers—which presently cost 2- to 5-fold more than an HPLC system—will eventually be regarded as just another HPLC detector. 4.2.3 Signal, Noise, Drift, and Assay Precision The ability of a detector to provide precise and accurate quantitative data is a function of the signal size generated by the analyte, background noise, and—to a certain extent—baseline drift. Additional discussion of these topics is deferred to Section 11.2.4. 4.2 DETECTOR CHARACTERISTICS 153 Response drift 0 123456789101112 Time (min) 6 4 2 ×10 4 AU Time (min) (a) (b) 0102030405060708090 short-term noise Figure 4.4 Detector noise and drift. (a) Baseline, showing noise and drift; inset expanded to show short-term noise; (b) long-term noise on baseline. 4.2.3.1 Noise and Drift As defined in Figure 4.4, short-term noise is the baseline disturbance that occurs with acycletimeof1 min (inset, Fig. 4.4a). Short-term noise is measured manually as the peak-to-peak noise bracketing the extremes of the baseline (between dashed lines of Fig. 4.4a; also shown in Fig. 4.7) in appropriate units (e.g., absorbance units for a UV detector). Alternatively, short-term noise can be determined by the data-system (usually with a root-mean-square calculation), if this feature is available. A common contribution to short-term noise is the ‘‘buzz’’ on the baseline resulting from noise contributed by motors or other appliances operated on the same electrical circuit (as in Fig. 4.5). The frequency of this noise corresponds to the 60 Hz electrical frequency, so it is referred to as 60-cycle noise (or 50-cycle in many countries). Short-term noise often can be reduced or eliminated through the use of electronic noise filtration. Most detectors have user-selectable noise filtration built into their electronics. Alternatively, a resistance-capacitance (RC) filter can be built using inexpensive components purchased from Radio Shack plus directions from an introductory physics book. An example of the effectiveness of electronic noise filtration is shown in Figure 4.5, where the noise on the raw signal is reduced by ≈300-fold through the application of an RC filter with a 1-sec time constant [12]. If the time constant is too large, it will also reduce the signal. As a rule of thumb, if the time constant is less than ≈10% of the peak width, the peak signal will not be compromised. For example, a 10,000 plate peak at k = 2 generated by a 154 DETECTION 0 Time ( min ) Response (mAU) 0 5 20 25 30 105 10 15 raw signal with 1-sec time constant Figure 4.5 Effect of a resistance-capacitance noise filter on detector noise. Raw detector signal with ≈16 mAU short-term noise; 1-sec time constant reduces noise to <1mAU. 012345 Time (min) 0 −5 −10 −15 5 10 15 0 −5 −10 −15 5 10 15 Detector signal (μV) (a) (b) Figure 4.6 Effect of data rate on detector noise. Baseline plots of data generated with (a)data collection rate of 15 Hz and (b) 1 Hz (with data system input leads shorted). Adapted from [12]. 150 × 4.6-mm, 5-μm column operated at 1.5 mL/min would have a peak width of ≈7 sec, which would translate into a suitable time constant of ≤0.7 sec. The data-system sampling rate (Section 11.2.1.1) also can affect short-term noise. This is illustrated in Figure 4.6, where data sampling rates of 15 Hz (points per 4.2 DETECTOR CHARACTERISTICS 155 sec) and 1 Hz are compared [12]. The noise should be reduced by the square root of the data rate change; therefore the change in the sampling rate from 15 Hz (Fig. 4.6a) to 1 Hz (b) should reduce the noise ≈4-fold, as seen in Figure 4.6. The noise in Figure 4.6 is reduced from ≈12 to ≈3 μV, although it looks like much more because the high-frequency noise of Figure 4.6a is mostly eliminated in Figure 4.6b.Aruleof thumb for setting the data sampling rate is to select a rate that collects ≥ 20 points across a peak. For the N = 10,000, k = 2 example above (peak width ≈7 sec), a data rate <7/20 sec or ≥ 20/7 sec −1 ≈ 3 Hz would be suitable. Remember that it is better to err on the side of oversampling when the data are initially recorded; the data rate can be reduced by post–run processing, but never increased (Section 11.2.1.1). Although the detector time-constant (noise filtration) and the data-system sampling rate (or data collection rate for processed data) both can be used to reduce baseline noise, they accomplish this in a different manner. It is therefore wise to evaluate both techniques (individually or in combination) to determine whether one or a combination is more effective at noise reduction. Long-term noise is the fluctuation in the baseline that occurs with a periodicity in the same range as chromatographic peaks, as in Figure 4.4b. The longer period of this type of noise means that electronic noise filtration will not be effective. Long-term noise often is the result of elution of strongly retained materials from previous runs. As retention is increased, peak width is increased, eventually to the point where the peak deteriorates into baseline wander. A column flush with strong solvent (e.g., 100% ACN or MeOH) will often reduce long-term noise. Baseline drift is the gradual shift in the baseline over the course of one or more runs, as illustrated by the heavy dashed line in Figure 4.4a. Baseline drift is common with gradient elution because of the difference in detector response (the concentration range over which the detector output is proportional to analyte concentration e.g., UV absorbance) between the starting and ending mobile phase (Section 17.4.5.1). Baseline drift with UV detection can often be compensated by adding a nonretained, UV-absorbing compound to the less-absorbant mobile-phase component (p. 178 of [13]). A change in the temperature of the mobile phase also can cause baseline drift, especially for refractive-index detectors and other detectors that are sensitive to refractive-index changes. Careful temperature control of the column, connecting lines, and detector, as well as isolation of the entire HPLC system from drafts can minimize drift. As long as the data system is able to adequately integrate peaks, a moderately drifting baseline will not have any detrimental effect on data quality. Short-term and long-term noise are superimposed on baseline drift, as seen in Figure 4.4a. A good approach to minimize noise and drift problems is to (1) use appropriate detector time-constants and data-system sampling rates, (2) regularly flush the column with strong solvent, and (3) minimize temperature fluctuations within and near the HPLC system. Remember that some baseline drift is common in gradient elution, because of the difference in detector response to the A- and B-solvents, and may not be changed by the practices above. 4.2.3.2 Signal-to-Noise Ratio (S/N) This ratio is more important than either the signal or noise alone as an indicator of detector performance for a particular peak. Noise is measured as described in Section 4.3.2.1 for short-term noise, as shown in Figure 4.7. The signal is measured . MS Detectors, 190 4.15 OTHER HYPHENATED DETECTORS, 191 Introduction to Modern Liquid Chromatography, Third Edition, by Lloyd R. Snyder, Joseph J. Kirkland, and John W. Dolan Copyright © 201 0 John. 158 4.3 INTRODUCTION TO INDIVIDUAL DETECTORS, 160 4.4 UV-VISIBLE DETECTORS, 160 4.4.1 Fixed-Wavelength Detectors, 163 4.4.2 Variable-Wavelength Detectors, 164 4.4.3 Diode-Array Detectors, 165 4.4.4. detectors the mobile phase is maintained in a liquid state, detection occurs, and the mobile phase passes out of the detector cell to a waste container, fraction collector, or another detector. 150