Basic principles of preparative HPLC Basically, preparative HPLC follows the same rules as analytical scale chromatography. However, there are important differences in the aims of the two techniques. In analytical HPLC chromatographers focus on peak shape, and resolution of all eluted analytes, whereas in preparative chromatography yield and purity of the final product, as well as cost-effectiveness of the method, are emphasized. throughput Typical relation between feed volume and resolution w w min demands on preparative separations 1 yield purity w w min = peak width = minimum peak width V 0 This chapter discusses the most important parameters, which have to be taken into account in preparative method development. Definition of the production rate includes information about the required purity of the isolated product. That is, the user has to define the minimum resolution of the product peak from neighbouring peaks. The parameters determining the resolution between two peaks are relative retention α (ratio of both k values), the plate number of the column and the k value (formulas see chapter Basic terms and definitions on page 173). The relative retention is determined by the chromatographic system, i.e. mobile and stationary phases. Proper choice of column and eluent will produce the required selectivity and resolution of the chromatographic system for the sample to be separated. In practice, one often has to use a non-optimal system, e.g. because of sample solubility or because of incompatibility of a chromatographically desirable eluent with the following work-up or application steps of the isolated product after chromatography. For a detailed discussion of the term resolution please refer to the paper by L.R. Snyder 1. For maximising throughput, preparative HPLC columns are often overloaded (for details concerning overload phenomena see below). Under these conditions the resolution depends on the sample mass and on the injection volume. With increasing feed volume V 0 the resolution remains almost constant, until from a certain value it decreases linearly with further increase of the feed volume (for details about volume overload see below). R R max 1 R = resolution R max = maximum resolution V 0 = feed volume V 0 61
Basic principles of preparative HPLC When speaking about the production rate of a preparative separation, the term loadability of the column should be considered, too. According to general understanding, this is the maximum sample size (with defined sample mass and volume) under which a column still provides optimum selectivity. The chemical literature contains several propositions for definition of the loadability 3 6. In practice, however, these theoretical approaches are seldom used, because the maximum injection size is determined empirically: one injects increasing amounts, until the peaks just touch. However, it should be noted, that for the same injection size the loadability for dilute samples is higher than for concentrated ones 7. The parameters which are important for the optimisation of the mass loadability of a column can be described by the following formula. Mass loadability of a column 2 M C 1 πr 2 d = lkda S C P 2 ----- l M = maximum sample mass C 1, C 2 = constants r = column radius l = column length K = partition coefficient d = packing density A S = adsorbent surface d P = particle diameter The significance of the individual parameters can be easily seen. However, two terms should be noted in detail: column length l and particle diameter d P. As can be seen from the term d P 2 /l, the mass loadability of the column decreases with increasing plate number (l/d P 2 is proportional to the plate number N). Experimentally, this relation can be easily shown with columns of different plate numbers. N The preceding diagram shows the dependence of the separation efficiency of two columns from the mass loading. The two columns were run under the same conditions, but with packings of different particle size (d P ) 8 If an increased loadability is required for a given separation efficiency, it is recommended to increase particle size and column length, the increase in column length being the square of the increase in particle diameter. This will produce a welcome side effect: the relative permeability of the column will also increase quadratically, and the pressure drop of the column will decrease quadratically as shown in the table below. Influence of particle size and column length on loadability and permeability for constant separation efficiency l [cm] d p [µm] rel. loadability rel. permeability 25 10 1.0 1.0 14 1.4 2.0 75 17 1.7 3.0 100 20 2.0 4.0 As can be seen from the following formula, the volume loadability linearly depends on the dead volume of the column used, and otherwise it only depends on the k values of the components to be separated and on the separation efficiency of the column. Volume loadability of a column 3) 2 V L = V 0 ( α 1)k A ------- ( 2+ k A + k B ) N V 0 = dead volume V L = maximum overload volume α = relative retention (k B /k A ) N = plate number k A,k B = capacity factors column 1 column 2 mass loading loadability column 1 loadability column 2 g (component i) ---------------------------------------- g (adsorbent) 62
Basic principles of preparative HPLC In addition to the loadability of the column used for a preparative separation another consideration can be important for the optimisation of the production rate 9. The production rate is directly proportional to the column diameter, the linear flow velocity of the mobile phase, the concentration of the component to be isolated (unless under mass overload conditions) and the term [1/N H 0 /l] 1/2, where H 0 is the plate height of the column under ideal conditions, l is the column length, and N is the plate number required for separation of the desired product with the purity required. The following limiting cases can be distinguished: l < H 0 N: Here a negative value is generated under the square root with the result of a physically meaningless production rate: the plate number of the column is smaller than required. l = H 0 N: The column plate number is just sufficient for the separation. The corresponding column length is a critical value where the production rate is zero. l > H 0 N: Further increase of the column length results in increasing production rates. Production rate [mg/h] 30 20 10 5 7 10 32,5 particle size [µm] 0 0 0,5 1 1,5 2 3 4 5 column length [m] Since in practice optimisation of the production rate of a desired substance often results in overload conditions of the column, we briefly wish to discuss the related phenomena. Typical cases, which can cause column overload conditions, are e.g. samples with low solubility in the mobile phase or injection of too large a sample volume, highly concentrated samples, samples which are dissolved in a solvent with much better solvation characteristics than the mobile phase. Here sample concentration at the column inlet can cause problems. When considering overload phenomena, one has to distinguish between concentration overload, volume overload and mass overload conditions. Concentration overload The concentration of the analytes in the injected sample solutions is increased, while the injection volume is kept constant. With increasing overload peaks are more and more distorted, with the peak shape approaching a triangle. Fronting as well as tailing can occur. Simultaneously, with increasing overload the peak maxima are shifted, in most cases towards the dead time. This type of overload is called concentration overload. A concentration overload is only possible, if the solubility of the analytes in the sample solution is large enough. Volume overload If the solubility of the analytes is low, overload can only be obtained if, for a given sample concentration, the injection volume is continuously increased. Increasing sample volume results in peak broadening, approaching a rectangular shape. However, peaks remain symmetrical and from a certain overload peak heights remain constant. This type of overload is called volume overload. A characteristic of volume overload is constant retention volumes of the peak fronts even under overload conditions 3. The rectangular peak shape obtained by volume overload is shown in the following chromatograms. The relation between column length and production rate can be seen in the diagram above, where typical values for production rates are plotted versus column length. The figure shows different curves for columns with different particle sizes and hence different separation efficiencies. It should be noted, that after a steep rise at relatively short column lengths the slope of these curves decreases drastically approaching a saturation value. This is reached when 1/N is much larger than H 0 /l. 63
Basic principles of preparative HPLC Volume overload is demonstrated for the separation of a 3- component mixture of benzene, naphthalene and anthracene and a 2-component mixture of naphthalene and anthracene. The masses of the solutes injected were kept constant in all cases. 3-component mixture 2-component mixture Transition from elution development towards frontal analysis caused by volume overload sample volume A 10 µl 0 µl A B C B C sample volume 1 µl sample volume 2 ml 3 ml sample volume 1 ml sample volume 2 ml A B C A B C sample volume 2 ml sample volume 4 ml sample volume 6 ml sample volume 3 ml sample volume 6 ml B+C A B C The constant retention volumes of the peak fronts combined with the rectangular peak shapes, however, allow a peak assignment with the aid of frontal analysis even for an otherwise insufficient resolution. When injecting such large volumes, certain precautions have to be taken when employing injection procedures with valve and sample loop. Improper injection can result in the concentration profile of the sample taking the form of a Poisson curve which will make frontal analysis difficult or impossible 3. For the decrease of separation efficiency under volume overload conditions one can make the following estimate: The plate heights will increase by about 20% when the sample volume is about 1.5% of the column void volume, and they will about double when the sample volume increases to about 3 3.5% 5. sample volume 16 ml A+B+C A A+B B+C C 64
Basic principles of preparative HPLC Mass overload If the injected sample mass (calculated from injection volume and concentration) exceeds a certain value, the local concentration of the sample in the column can be so large, that equilibration is no longer possible: this case is called mass overload. Mass overload is much more complex than volume overload and is based on three main effects 3 : Dispersion effects. The sample is distributed along the column by the mobile phase until it contacts sufficient adsorbent to permit equilibrium with the stationary phase. This phenomenon results in band spreading similar to volume overload. Deactivation of the adsorbent. In addition to dispersion effects a massive sample charge will occupy a significant portion of the stationary phase resulting in a changed effective polarity of the mobile phase compared to the stationary phase. This can cause reduced retention which will also affect components of the sample with lower concentration. Non-linear adsorption isotherms, caused by large sample concentrations, result in peak tailing. Thus, mass overload can be recognised by changed retention volumes of the peak fronts of all components of a mixture to be separated. Peak broadening and tailing will mainly affect the sample components which are present in excess relative to the column load capacity (see chromatograms on the left). Mass overload for the separation of benzene, naphthalene and anthracene B N A The figure shows mass overload for the separation of benzene, naphthalene and anthracene. The amount of benzene injected is increased from 180 µg via 8.1 mg to 16.9 mg, while the amounts of naphthalene and anthracene are kept constant. If the aim of a preparative separation is to obtain as much of the pure compound per time unit as possible, overload of the column will in most cases be necessary. In practice combinations of concentration and volume overload are most often used. With diluted samples volume overload will occur more often, while concentrated sample solutions will show a tendency towards concentration and mass overload. Often both effects are present and peaks approach the shape of a trapezoid. Concentration overload is to be preferred, because it allows separation of larger sample amounts. References: 1) L.R. Snyder, J. Chromatogr. Sci. 10 (1972) 200 and 369 2) A. Wehrli, U. Hermann and J. F. K. Huber, J. Chromatogr. 125 (1976) 59 3) R. P. W. Scott, P. Kucera, J. Chromatogr. 119 (1976) 467 4) L. R. Snyder, Anal. Chem. 39 (1967) 698 5) W. Beck, I. Halasz, Z. Anal. Chem. 291 (1978) 340 6) T. Roumeliotis, K. K. Unger, J. Chromatogr. 185 (1979) 445 7) J. J. De Stefano, H. C. Beachell, J. Chromatogr. Sci 10 (1972) 654 8) A. W. J. De Jong, H. Poppe, J. C. Kraak, J. Chromatogr. 209 (1981) 432 9) K. P. Hupe, H. H. Lauer, J. Chromatogr. 203 (1981) 41 10) V. R. Meyer, Praxis der Hochleistungs-Flüssigchromatographie, Otto Salle Verlag, Frankfurt, 7. Aufl., 1992 B B N A N A 65
Scale-up from analytical to preparative HPLC amount of adsorbent sample volume amount of analyte flow rate Scale-up factor } = f(id) However, overloading a column always means a considerable loss of separation efficiency. For this reason, overloading the chromatographic system must show sufficient resolution. Consequently, for optimising a preparative separation, the chromatographic selectivity has to be optimised under analytical conditions to reach a resolution as high as possible. The higher the selectivity for a given phase system, the more a column can be overloaded. For preparative separations under overload conditions stationary phases with larger mean particle sizes are usually sufficient. Prerequisite for a successful transfer from established analytical methods to the preparative scale are stationary phases with equal selectivity for both methods. However, heavy overload can also change the chromatographic selectivity relative to analytical conditions. For separation of larger amounts of a substance there are two general approaches: linear scale-up of the analytical system overloading the column In the case of linear scale-up the column length is kept constant and the column cross section is increased proportionally to the sample mass. Eluent flow and sample volume are adapted correspondingly. As in analytical HPLC, stationary phases with small particle sizes are used. The separation efficiency of the prep column is more or less the same compared with the analytical column. Peaks remain sharp and symmetrical. The procedure of linear scale-up is not very economical, since large columns and large amounts of eluents are required, and the substance yield is relatively low. For example, more than 5 kg stationary phase and l mobile phase are necessary to separate less than 1 g of a substance mixture in one run. For this reason linear scale-up alone is only used in exceptional cases, e.g. if for difficult separations the high separation efficiency is needed for the isolation of (in most cases small amounts of) substances with very high purity. A better way is overloading the column, since for a given yield less stationary phase and less eluent are required. This procedure allows injection and separation of amounts in the lower milligram range on an analytical HPLC column. For preparative separations on the gram or kilogram scale overloading is combined with a linear scale-up of the chromatographic system. Procedure for scale-up In general, an analytical chromatogram is used as a starting point for a preparative separation. It is necessary to find conditions, which separate the sample mixture isocratically with good resolution, since gradients are not recommended for preparative separations, because they require large efforts in every respect. If necessary, proper sample preparation may be required. The better the resolution in the analytical chromatogram, the larger the load on the preparative column can be. After optimisation of the separation on the analytical column the maximum sample amount for injection of a concentrated solution is determined empirically. In the simplest case overload is increased, until the peaks in the chromatogram of the separation just do not yet overlap, allowing isolation of the peaks and, after removal of the mobile phase, obtaining the pure substances with 100% yield. Undiluted solutions are not favourable. It is not recommended to dissolve the sample in a stronger solvent than the mobile phase. Since for preparative work in most cases larger volumes are injected, the solvent can severely interfere with the equilibrium in the column, even making reproducibility of the chromatographic system impossible. Solubility of the sample in the mobile phase has to be good, because otherwise the column may get plugged. 66
Scale-up from analytical to preparative HPLC The scale-up factor After optimisation of the analytical separation the preparative column is to be considered. Transfer from an analytical to a preparative separation is easiest, if both are packed with the same stationary phase (linear scale-up). A key for successful transfer of results gained for the analytical column to the preparative scale is the proper scale-up factor. If the analytical or the preparative factor is not listed in the table below, it can be calculated form the following formula: Now every parameter relevant for the separation has to be multiplied with the scale-up factor. This includes flow rate, sample volume, sample mass, and the IDs of the capillaries. It is important during scale-up to increase the inner diameters of the capillaries proportionally, since otherwise the back-pressure of the system can become to large. For our programme of capillary tubing for HPLC please see the chapter Accessories on p. 169) 2 L M p M p d = p a ------------------ 2 L a d a M L d a p = sample mass = column length = column diameter analytical column preparative column Scale-up factors and parameters for typical column dimensions Column dimensions [mm] 4 x 2 8 x 2 10 x 2 16 x 2 21 x 2 40 x 2.8 x 2 80 x 2 Particle sizes [µm] NUCLEOSIL NUCLEODUR 3, 5, 7, 10 3, 5 Linear scale-up factor 1 4 6.25 16 27.6 100 161.3 400 Typical sample mass * [mg] 0.02 2 0.08 8 0.15 13 0.3 35 0.6 60 2 210 3 3 10 8 Amount of packing / column [g ±20%] 2 8 13 35 60 210 3 8 Typical flow rate [ml/min] 0.5 1.5 2 6 3 9 8 24 14 40 1 80 2 200 600 * for RP material; the maximum amounts given here always depend on the separation problem and on the sample composition. In some cases even half of the amounts given can cause dramatic overload, in other cases the maximum amounts can still give acceptable separations. 67
columns for preparative HPLC As with our analytical columns, our preparative columns feature capillary connections with UNF inner threads 10-32 ( 1 / 16 ), thus meeting today s standard in HPLC technology. They are manufactured from stainless steel and are packed with NUCLEOSIL or NUCLEODUR packings from our range of silicas for HPLC. All packed columns are individually tested and supplied with the corresponding test certificate. We offer three different types of preparative columns: Standard-Prep, VarioPrep and Ecoprep. are available in lengths of 125 to 0 mm and inner diameters of 10 mm, 21 mm and.8 mm. Standard-Prep guard columns are mm long and available with the same inner diameters. Thus for scale-up the user can choose from a range of about two orders of magnitude in column diameters from the analytical 4 mm ID columns up to the.8 mm ID prep columns. have been developed in order to allow compensation of the dead volume, which could result at the column inlet after some time of operation, without need for opening the column. This special column technology is available as columns with one adjustable end fitting at the column inlet and a fixed end fitting at the column end. On request, we can also supply columns with two adjustable end fittings, an option which may e.g. be useful for frequent use of backflushing techniques. are available with lengths of 125 and 2 mm and inner diameters of 10, 21, 32, 40 and 80 mm. VarioPrep guard columns are mm long and available with inner diameters of 10, 21 and 40 mm. are available with lengths of 30, 70, 125 and 2 mm. The column head is an enlarged version of the analytical EC columns (s. p. 137). The inner diameters are intermediate sizes of 8 and 16 mm. These columns are especially suited for SMB systems. Shorter columns (individually tested) and guard columns are available on request. On request, our preparative columns can also be custom-packed with other types of NUCLEOSIL packings as well as with NUCLEODUR silicas. In addition, they can be made in lengths other than those indicated in the tabulated summary on pages 69 to 72. Preparative columns different types of column end fittings 1 Standard-Prep column with 10 mm ID 2 Standard-Prep column with 21 mm ID 3 VarioPrep column with 21 mm ID 4 EcoPrep column with 16 mm ID 1 2 3 4 EcoPrep columns 68
Preparative columns packed with NUCLEOSIL and NUCLEODUR Ordering information (choice of columns) * Length 30 mm 70 mm 125 mm 2 mm Guard columns mm Base deactivated RP phases NUCLEODUR C 18 Gravity, 5 µm Particle size 5 µm, pore size 110 Å; octadecyl phase, endcapped, 18% C; eluent in column acetonitrile / water 10 mm ID 762104 762105 21 mm ID 762102 762103 NUCLEOSIL 100-5 C 18 HD Particle size 5 µm, pore size 100 Å; octadecyl phase, endcapped, monomeric coating, 20% C; eluent in column acetonitrile / water 10 mm ID 715387 21 mm ID 715386 10 mm ID 715849 21 mm ID 715697 715851 8 mm ID 715290.80 715291.80 715292.80 715293.80 NUCLEOSIL 100-7 C 18 HD Particle size 7 µm, pore size 100 Å; octadecyl phase, endcapped, monomeric coating, 20% C; eluent in column acetonitrile / water 16 mm ID 715311.160 715312.160 715313.160 715314.160 NUCLEOSIL 100-5 PROTECT I Particle size 5 µm, pore size 100 Å; special RP phase, endcapped, monomeric coating, 11% C; eluent in column acetonitrile / water 10 mm ID 715307 8 mm ID 715304.80 16 mm ID 715304.160 * On request, all preparative columns are available with any NUCLEOSIL or NUCLEODUR packing. Each column is individually tested and supplied with test chromatogram and test conditions 69
Preparative columns packed with NUCLEOSIL and NUCLEODUR Ordering information (choice of columns) * Length 30 mm 70 mm 125 mm 2 mm Guard columns mm Standard octadecyl phases NUCLEODUR 100-5 C 18 ec Particle size 5 µm, pore size 110 Å; octadecyl phase, endcapped, 17.5% C; eluent in column acetonitrile / water 10 mm ID 762001 762005 21 mm ID 762002 762003 NUCLEOSIL 100-5 C 18 Particle size 5 µm, pore size 100 Å; octadecyl phase, endcapped, 15% C; eluent in column acetonitrile / water 10 mm ID 715830 21 mm ID 715836 NUCLEODUR 100-7 C 18 ec Particle size 7 µm, pore size 110 Å; octadecyl phase, endcapped, 17.5% C; eluent in column acetonitrile / water 10 mm DI 762046 762047 762049 21 mm DI 762042 762043 762051 10 mm DI 762044 762045 762048 21 mm DI 762040 762041 7620 NUCLEOSIL 100-7 C 18 Particle size 7 µm, pore size 100 Å; octadecyl phase, endcapped, 15% C; eluent in column acetonitrile / water 10 mm ID 7102 715203 21 mm ID 7120 715205.8 mm (2 ) ID 71 10 mm ID 715809 715802 715827 21 mm ID 715659 715652 715666 40 mm ID 715691 8 mm ID 715330.80 715331.80 715332.80 16 mm ID 715330.160 715331.160 715332.160 NUCLEOSIL 100-10 C 18 Particle size 10 µm, pore size 100 Å; octadecyl phase, endcapped, 15% C; eluent in column acetonitrile / water 16 mm ID 715274.160 715273.160 NUCLEOSIL 120-7 C 18 Particle size 7 µm, pore size 120 Å; octadecyl phase, endcapped, 11% C; eluent in column acetonitrile / water 10 mm ID 715113 * On request, all preparative columns are available with any NUCLEOSIL or NUCLEODUR packing. 70
Preparative columns packed with NUCLEOSIL Ordering information (choice of columns) * Length 30 mm 70 mm 125 mm 2 mm Guard columns mm NUCLEOSIL 300-7 C 18 Particle size 7 µm, pore size 300 Å; octadecyl phase, endcapped, 6.5% C; eluent in column acetonitrile / water 10 mm ID 7125 21 mm ID 7131 10 mm ID 715806 21 mm ID 715656 Standard octyl phases NUCLEOSIL 100-7 C 8 Particle size 7 µm, pore size 100 Å; octyl phase, not endcapped, 8.5% C; eluent in column acetonitrile / water 10 mm ID 7101 21 mm ID 7119 10 mm ID 715801 21 mm ID 715651 8 mm ID 715630.80 16 mm ID 715630.160 NUCLEOSIL 300-7 C 8 Particle size 7 µm, pore size 300 Å; octyl phase, not endcapped, ~3% C; eluent in column acetonitrile / water 10 mm ID 7124 21 mm ID 7176 10 mm ID 715805 21 mm ID 715655 * On request, all preparative columns are available with any NUCLEOSIL or NUCLEODUR packing. 71
Preparative columns packed with NUCLEOSIL Ordering information (choice of columns) * Length 30 mm 70 mm 125 mm 2 mm Guard columns mm Standard butyl phases NUCLEOSIL 120-7 C 4 Particle size 7 µm, pore size 120 Å; butyl phase, endcapped, eluent in column acetonitrile / water 10 mm ID 715112 NUCLEOSIL 300-7 C 4 Particle size 7 µm, pore size 300 Å; butyl phase, endcapped, ~2% C; eluent in column acetonitrile / water 10 mm ID 7123 21 mm ID 7132 10 mm ID 715804 715971 21 mm ID 715654 Standard phenyl phases NUCLEOSIL 100-7 C 6 H 5 Particle size 7 µm, pore size 100 Å; phenyl phase, not endcapped, eluent in column acetonitrile / water 10 mm ID 7103 10 mm ID 715803 21 mm ID 715653 Unmodified silica NUCLEOSIL -7 Particle size 7 µm, pore size Å; unmodified, eluent in column n-heptane 10 mm ID 7104 715711 21 mm ID 7122 NUCLEOSIL 100-7 Particle size 7 µm, pore size 100 Å; unmodified, eluent in column n-heptane 21 mm ID 7126 10 mm ID 715800 21 mm ID 7156 8 mm ID 715275.80 16 mm ID 715275.160 * On request, all preparative columns are available with any NUCLEOSIL or NUCLEODUR packing. On request, are also available with two adjustable end fittings, with 80 mm ID or with different lengths. For preparative HPLC columns for special separation problems please see the following chapter Columns for special applications Each column is individually tested and supplied with test chromatogram and test conditions 72