116 EQUIPMENT 100 80 60 40 20 0 (a) (b) %-response 02040 20 μL loop 20 μL μL dispensed 60 80 tube wall sample Figure 3.19 Effect of laminar flow on injection accuracy. (a) Comparison of detector response when loading different volumes into a 20 μL loop; (b) laminar flow profile of sample. Adapted from [12]. simplified description is further complicated by back-flushing and by mixing that takes place when changes in tube diameter or other flow disruptions are encountered. For manual injection, it is best to keep the injected volume constant and ≤50% or ≥200% of the loop volume for maximum accuracy. 3.6.2 Autosampler Designs Autosamplers have largely replaced manual injectors—primarily for convenience, but autosamplers also provide levels of precision that may not be possible with manual injection. Many of today’s autosamplers are capable of <0.5% RSD of peak area for injection volumes of ≥5 μL. Injection-volume accuracy may fall short of this level because of errors in calibration of the sample syringe or injection loop. Incomplete loop filling due to laminar flow (Section 3.6.1.2) can also introduce volumetric errors. Usually the precision of injection is more important than accuracy because of the compensating use of standards or calibrators. Autosamplers are best used in a constant-volume injection mode for each method, so the same exact (though not necessarily accurate) sample volume is injected for both calibration standards and samples. When this practice is followed, injection accuracy is less important, and the excellent precision will provide satisfactory analytical results. Carryover is indicated by the presence of a small peak(s) in a blank chro- matogram (no sample injection) that follows a separation where a sample was injected. Carryover results from part of a sample being retained in the system, 3.6 AUTOSAMPLERS 117 after a separation is complete; it is especially a problem when the injection of a low-concentration sample follows that of a high-concentration sample (Section 17.2.5.10). Many autosamplers have features that provide for a wash of the needle and sample-contacting passages between injections (to avoid carryover). These range from static wash-vials to active-flush capabilities When a wash solvent is used, the composition is chosen to readily dissolve the sample and to be compatible with the mobile phase. Samples generally are placed in individual sample vials or well-plates containing 96 or 384 sample wells. Vials most commonly are made of glass, sometimes specially treated to reduce adsorptive losses of sample. Sizes are 1 to 1.5-mL capacity for standard vials and 100 to 300 μL (or less) for microvials or inserts in standard vials. Vial closures use a cap and septum, usually made of silicon rubber and/or Teflon film. Sample plates generally are plastic with volumes of ≤1 mL per well. Plate closures are usually a press-on septum mat or iron-on metalized polymer film. Sample access most commonly is via movement of the sample needle in xyz axies to the sample container. Some autosamplers use a rotating tray to bring the sample vials to the needle, while others pick up an individual vial and move it to the needle. The cycle time is the amount of time it takes an autosampler to complete an injection from the time it is given the initial start signal. At a minimum, the cycle time includes the time it takes to pick up a sample and inject it onto the column. The addition of wash steps or other procedures can increase the cycle time. As long as the autosampler cycle time does not have a significant impact on sample throughput, it is not important. Thus a 1-min cycle time for a 30-minute gradient run is of little consequence, but the same cycle time would greatly reduce throughput for a 4-minute run. Cycle time that is ≤5% of the run time, generally is considered acceptable; as of this writing, few autosamplers have cycle times ≤15 sec with acceptable levels of carryover and injection precision. One way of reducing the negative impact on the autosampler cycle time is to use a ‘‘load-ahead’’ feature offered by some systems. In this implementation, the autosampler is programmed to perform its wash cycle(s) and pick up the sample while the previous sample is being eluted. As soon as the run is completed, the injection can be made, this reduces the effective cycle time to just a few seconds. Three common autosampler designs are in common use: • pull-to-fill • push-to-fill • needle-in-loop 3.6.2.1 Pull-to-Fill Autosamplers The pull-to-fill autosampler design is illustrated in Figure 3.20. A syringe is mounted on a mechanical drive and connected to the injection valve as shown. A sample loop corresponding to the desired injection volume is mounted on the valve. The needle is mounted on a piece of connecting tubing attached to the valve. The needle may be moved to the sample vial, or the vial may be moved to a stationary needle. In the load position (Fig. 3.20a) the needle penetrates the septum on the vial and the syringe plunger is withdrawn to pull sample through the needle and connecting 118 EQUIPMENT 010 0 1 c p (a) 010 0 1 c p (b) Figure 3.20 Pull-to-fill autosampler design. (a) Transfer of sample from sample vial to injec- tor loop; (b) injection. p, flow from pump; c, flow to column. tubing until excess sample exits the sample loop. The valve rotor then is moved to the inject position (Fig. 3.20b), and the sample is pumped onto the column. Note that no needle seal is used in this type of autosampler. The pull-to-fill autosampler wastes sample because of the relatively large diameter of connecting tubing required to avoid blockage, and the need to flush excess sample through the loop (Section 3.6.1.1). It is therefore best used when the amount of sample is not limited, such as applications used to monitor production processes. The design is simple and reliable, and because it uses an overfilled, fixed-volume loop, it can have very good precision and accuracy. 3.6.2.2 Push-to-Fill Autosamplers The push-to-fill autosampler is an automated version of the manual injector. In the load mode (Fig. 3.21a) the needle draws sample from the sample vial into a connecting tube attached to a mechanically operated syringe. The needle then is withdrawn from the sample vial and pushed into the low-pressure needle-seal in the injection port (Fig. 3.21b); sample is then dispensed into the sample loop. Next the valve rotor is moved to the inject position (Fig. 3.21c), and the sample is pumped onto the column. The push-to-fill autosampler can be used in the filled-loop or partially filled loop injection mode (Sections 3.6.1.1 and 3.6.1.2). In the partially filled mode, the precision depends on the precision of the syringe controller. Because it is possible to inject nearly all of the sample, the push-to-fill autosampler does not waste much sample—perhaps 10 μL of sample is left in the needle and connecting passages. The push-to-fill autosampler design uses a low-pressure needle-seal that generally is trouble free. These autosamplers are very popular and are used for a wide range of applications. 3.6 AUTOSAMPLERS 119 010 0 1 010 0 1 010 0 1 c p c w w p (a)(b)(c) low-pressure needle-seal Figure 3.21 Push-to-fill autosampler design. (a) Transfer of sample from sample vial to injec- tion needle and syringe; (b) filling of sample loop; (c) injection. p, flow from pump; c, flow to column; w,towaste. 3.6.2.3 Needle-in-Loop Autosamplers The needle-in-loop autosampler uses a needle and loop that are one piece (needle-loop in Fig. 3.22a). In the load position, as shown in Figure 3.22a, the needle picks up the sample from the vial. The needle is then moved to a high-pressure needle-seal in the injection port, and the rotor is turned to inject the sample (Fig. 3.22b). Because the tip of the needle is in the flow stream, the needle-in-loop autosam- pler injects all of the sample that is withdrawn from the sample vial, so there is no wasted sample with this injection technique. This makes the needle-in-loop autosampler a favorite for methods in which the sample volume is very small. These autosamplers typically use a 100-μL sample needle-loop, which will accommodate most analytical requirements. If a larger injection volume is needed, the needle-loop must be replaced with a larger one; this type of needle-loop is much more expensive than a conventional sample loop. This autosampler also depends on a high-pressure seal between the sample needle and the injection valve, which is a weak point with some implementations of this design. These autosamplers are a popular design because of the low sample waste and generally minimal carryover. 3.6.3 Sample-Size Effects The amount of sample that is injected can influence the appearance of the chro- matogram, not only in peak height or area but also in retention time and peak shape. Peak broadening can be influenced by the volume of sample injected, V s , and the injection-solvent strength relative to the mobile phase, as discussed below. Sample-mass effects and overload are discussed in Section 2.6.2. 120 EQUIPMENT 010 0 1 c p (a) 010 0 1 c p (b) high pressure needle-seal needle-loop w w Figure 3.22 Needle-in-loop autosampler design. (a) Transfer of sample from sample vial to injector needle-loop; (b) injection. p, flow from pump; c, flow to column; w,towaste. 3.6.3.1 Injection Volume The influence of the injection volume on the peak width was discussed in Section 2.6.1, and is summarized in Table 3.3 and Equation (2.27) as V p = 4 3 V s 2 + V p0 2 1/2 (2.27) where V p is the observed peak volume, V p0 is the volume of the peak due to broadening within the column, and V s is the injection volume when the mobile phase is used as the injection solvent for isocratic separations. Equation (2.27) can be used to determine how large an injection volume can be made for a given increase in peak volume; if a 5% loss in resolution is acceptable, a 5% increase in peak width can be tolerated. The allowed injection volumes listed in Table 3.3 are calculated for a 5% increase in peak width for various columns and retention factors. (Note that the allowed injection volumes in Table 3.3 are smaller than those calculated by Eq. 2.27, because of the typical increase of V s by ≈50% as the sample leaves the sample loop; see Section 2.6.1.) It is obvious that smaller injection volumes are required for smaller volume columns, columns that generate larger plate numbers, and/or early-eluted peaks—all of which result in narrower peaks. On the other hand, 4.6-mm i.d. columns generate fairly broad peaks, even when packed with sub-2-μm particles and used in short, 50-mm lengths. For example, a 50 × 4.6-mm, 1.8-μm column gives V p0 ≈30 μLfork = 0.5, with an allowable sample volume of V s ≈ 5 μL. Most autosamplers yield an imprecision of ≤0.5% RSD for injection volumes ≥5 μL. Some autosamplers are able to maintain a similar precision for sample vol- umes of 1 to 2 μL or smaller. For columns ≤2.1-mm i.d. and packed with ≤3- μm particles, the autosampler must be capable of precisely injecting very small sample 3.6 AUTOSAMPLERS 121 Table 3.3 Allowed Injection Volumes for 5% Band Broadening Column Characteristics Peak Volume (Injection Volume) a (μL) L (mm) d c (mm) d p (μm) N (h ≈ 3) k = 0.5 k = 2 k = 20 150 4.6 5.0 10,000 90 b (10) c 180 (20) 1,270 (130) 150 2.1 5.0 10,000 20 (2) 40 (4) 265 (25) 100 4.6 3.0 11,100 60 (6) 115 (15) 800 (90) 100 2.1 3.0 11,100 12 (2) 25 (15) 170 (20) 100 2.1 1.8 18,500 9 (1) 20 (2) 130 (15) 100 1.0 1.8 18,500 2 (<1) 5 (<1) 30 (3) 50 4.6 3.0 5,500 40 (4) 80 (10) 565 (60) 50 2.1 3.0 5,500 8 (1) 20 (2) 120 (80) 50 1.0 3.0 5,500 2 (<1) 5 (<1) 25 (3) 50 4.6 1.8 9,200 30 (4) 65 (7) 440 (50) 50 2.1 1.8 9,200 7 (1) 15 (2) 90 (10) 50 1.0 1.8 9,200 2 (<1) 3 (<1) 20 (2) a Equation (2.27), V s /V p0 = 0.16; note that the latter (theoretical) value of the injection volume V s has been divided by 1.5, to take into account the spreading of the sample plug as it leaves the loop. b Peak volume (μL) c Injection volume (μL) volumes. In all cases for isocratic separation, the peak width increases with the retention time, so longer retained peaks can tolerate larger injection volumes. Modification of a method so as to increase retention is one (seldom used) approach for minimizing extra-column peak broadening, because the resulting reduction in peak height and increase in run time is usually a poor trade-off. 3.6.3.2 Injection Solvent As mentioned in Section 2.6.1, when the injection solvent is not matched to the mobile phase, Equation (2.27) no longer holds. If the injection solvent is sufficiently weaker than the mobile phase, the sample will be concentrated at the head of the column. Basedonachangeink of about 2.5-fold for a change in the mobile phase of 10% B (Section 6.2.1), an injection solvent 10% B weaker than the mobile phase should sig- nificantly retard the sample as it enters the column; larger solvent-strength differences will be even more effective. When large-volume sample injections are desired, dilution of the sample with water may allow the injection of a larger sample weight (Section 2.6.1). The use of an injection solvent stronger than the mobile phase will adversely affect early-eluted peaks more than more strongly retained ones (Section 17.4.5.3). Also it is important to match the injection solvent for standards and samples. Injection in solvents stronger than the mobile phase tends to ‘‘wash’’ the sample down the column until it becomes fully diluted in the mobile phase. As with the use of dilute injection solvents, the observed effects are a function of both the injection volume and the difference in solvent strength between the injection solvent and the mobile phase. With 150 × 4.6-mm columns, injection volumes of ≤10 μL 122 EQUIPMENT in a strong solvent (e.g., 100% B) generally can be tolerated. When larger injection volumes and/or smaller volume columns are used, it is wise to compare retention and peak shape with a small-volume injection of the sample dissolved in mobile phase, in order to see if the results are acceptable. For mobile phases of <50% B, diluting a sample dissolved in 100% B ratio 1:1 with water or buffer will often allow a larger volume and weight of injected sample without adverse consequences. 3.6.4 Other Valve Applications Automated injection valves are most widely used in autosamplers, but the same valves also are used for other applications. These include: • column switching • fraction collection • waste diversion 3.6.4.1 Column Switching High-pressure switching valves are available in many configurations other than the simple six-port valve illustrated in Figure 3.17. These may be purchased as motorized valves with switching controlled through the external-events outputs of the HPLC system controller. Two general configurations are popular: two-position valves, such as shown in Figure 3.17, and multi-position valves, which allow a single input tube to be connected to one of many output tubes (see later the discussion of Fig. 3.25). Three applications are discussed here, and another in Section 2.7.6; many additional applications are available on the valve manufacturer’s websites (e.g., [6, 13]). One popular application of the two-position valve, sample enrichment,is shown in Figure 3.23. The objective is to concentrate a dilute sample and then inject the concentrated fraction onto the analytical column. An example of this is the concentration of a nonpolar analyte from a water sample (e.g., an environmental monitoring application). In the enrichment phase the valve is set as shown in Figure 3.23a, where the first pump pushes a dilute sample through an enrichment column, while the previous sample is separated on the analytical column, using a second pump. In this example the aqueous sample might be passed through a C 18 column in a weak mobile phase to trap the nonpolar materials. Once the entire sample is concentrated on the enrichment column, the valve is switched (Fig. 3.23b) and the sample is back-flushed onto the analytical column. Because of the reversed direction of flow and the sudden increase in mobile-phase strength, the sample is released from the enrichment column onto the analytical column in a narrow band for analysis. Other applications of column switching as in Figure 3.23 are discussed in Section 16.9. Figure 3.24 shows a valving configuration that allows regeneration of one column while a second column is eluting the sample to the detector. This can increase throughput and be advantageous for gradient applications, while at the same time increasing the utilization rate of an expensive mass spectrometric detector, such as for the analysis of drugs in plasma samples. In the configuration shown in Figure 3.24a, the sample is injected and analyzed on column 2 in the normal manner using gradient elution. Meanwhile column 1 is regenerated by the mobile phase delivered by a second pumping system. As soon as the sample is 3.6 AUTOSAMPLERS 123 Figure 3.23 Column switching for sample enrichment. (a) Valve set for loading sample enrichment column; (b) back-flushing enrichment column to analytical column. eluted from column 2 to the detector, the valves are switched to the configuration of Figure 3.24b, and a new sample is injected onto column 1 while column 2 is regenerated. This application increases throughput by the elimination of the time normally spent waiting for the column to be re-equilibrated to the starting conditions after a gradient run. However, it should be noted that other means exist to minimize the time required for column equilibration after gradient elution (Section 9.3.7). In Figure 3.25, a multi-position valve is used to select from one of three columns. With this setup, three separate columns can be evaluated automatically, for use in method development. One application of this technique involves a setup similar to Figure 3.25, but with as many as 32 different chiral columns installed on two 32-port valves (see Section 14.6.1). In a Gatling-gun approach, a sample is sequentially injected on each column in an unattended series of runs. The chromatograms are later inspected to determine which column provided the best separation. 3.6.4.2 Fraction Collectors Preparative chromatography (Section 15.2.4) requires fraction collection for either (1) the retrieval of individual peaks from a chromatogram or (2) the collection of fractions from an overlapped peak (as part of the purification of some compound). A fraction collector, which is commonly used for this purpose, resembles an autosampler that is used in a reversed mode. A single sample stream from the column is distributed into multiple vials through use of a mechanism that moves the outlet tube to the desired vial. In the simplest implementation a fraction collector is operated on a timed-collection basis. At some selected time after injection, collection starts and the sample is collected for a fixed time in each collection tube. For 124 EQUIPMENT Figure 3.24 Column switching for column regeneration. (a) Sample is injected on column 2 and directed to detector (MS) while column 1 is regenerated by pump 1; (b) sample is separated on column 1 while column 2 is regenerated. V1, V2, six-port switching valves. Figure 3.25 Use of multi-port switching valves for selection of one of several columns. As shown valve, V1, directs flow from pump, to column-1, through valve V2, to detector. Con- necting passage (- - -) moves to desired column under software control. example, if the peak of interest was eluted at 7 minutes, the fraction collector might be programmed to start collecting 20-second fractions starting at 6 minutes and ending at 8 minutes. This way several ‘‘cuts’’ across the peak would be collected. Another popular implementation of the fraction collector is to use an electronic circuit to monitor the detector output so that fractions are collected only when a peak is eluted. A delay coil of tubing can be mounted between the detector and the fraction collector to allow for peak detection just prior to the capture of a fraction. 3.6.4.3 Waste Diversion The possible applications of switching valves in HPLC are practically limitless. Two additional uses of switching valves are discussed here: the diversion of a waste stream to protect the detector, and the recycle of mobile phase for isocratic applications. One method of sample cleanup for the analysis of drugs in plasma is plasma precipitation (Table 16.11). Although quick and inexpensive, plasma precipitation 3.7 COLUMN OVENS 125 often leaves a significant nonvolatile protein burden in the sample, which can accumulate in the interface of an MS detector. One way to avoid this is to use a switching valve to divert to waste the portion of the column effluent that contains most of the protein—usually the early portion of the chromatogram near the column dead-volume. When actuated by the system controller as part of the method program, automated waste diversion will minimize interface contamination due to nonvolatile materials in the injected sample. Although the cost of mobile-phase solvents is not a large fraction of the total expense of sample analysis, it can be significant when the costs of disposal are considered. To reduce this expense, as well as for environmental reasons, some users attempt to reduce the volume of solvents used. One approach is to reuse the mobile phase. The simplest procedure is to direct the waste stream back to the reservoir (mounted on a stir-plate). As the waste stream is mixed with the remaining mobile phase, impurities are diluted and pumped back into the column at a steady state, so no interfering peaks will appear. Over time, however, the contaminant concentration in the reservoir will increase. A simple way to minimize this is to recycle only the portion of mobile phase that does not contain sample peaks. There are several commercial units (e.g., Axxiom’s SolventTrak) that include a switching valve and a sensor that monitors the detector output. When a peak is detected, the valve switches the detector effluent to waste; when no peaks are present, the valve directs the effluent to the mobile-phase reservoir. Thus only the ‘‘clean’’ mobile phase is recycled. A quantitative evaluation of mobile-phase recycling can be found in [14]. An alternative way to reduce solvent consumption is to decrease the column internal diameter. Solvent consumption is proportional to the cross-sectional area of the column, so replacement of a 4.6-mm i.d. column by a 2.1-mm i.d. column will reduce solvent usage by (4.6) 2 /(2.1) 2 ≈ 5-fold (the flow rate should be simul- taneously reduced by the same amount to maintain a constant linear velocity). The greater importance of extra-column peak broadening for small-diameter columns should be kept in mind, however. 3.7 COLUMN OVENS It has long been known that column temperature plays an important role in HPLC retention and selectivity. For additional information, see Section 6.3.3 and [15–17] (and associated references). 3.7.1 Temperature-Control Requirements A rule of thumb for reversed-phase isocratic separation is that a 1 ◦ Cincrease in column temperature will decrease values of k by about 2%. Temperature can also affect chromatographic selectivity (Sections 6.3.3, 7.3.2.2, 8.33, [15]), so close control of column temperature can be important—especially for separations with marginal resolution (e.g., R s ≤ 2). If the mobile phase entering the column is not preheated to the temperature of the column, distorted peaks can result. To avoid peak distortion, the temperature of the mobile phase as it enters the column should be within ±6 ◦ C of the column temperature. . cycle time to just a few seconds. Three common autosampler designs are in common use: • pull -to- fill • push -to- fill • needle-in-loop 3.6.2.1 Pull -to- Fill Autosamplers The pull -to- fill autosampler. valve rotor then is moved to the inject position (Fig. 3.20b), and the sample is pumped onto the column. Note that no needle seal is used in this type of autosampler. The pull -to- fill autosampler. Autosamplers The push -to- fill autosampler is an automated version of the manual injector. In the load mode (Fig. 3.21a) the needle draws sample from the sample vial into a connecting tube attached to a mechanically