HPLC systems Overview An hplc system is just a way to get a sample onto the end of a column, to pump solvents through the column until the analytes drop out the end, and to detect them as they appear. There is a tremendous amount of black-magic handed down from chromatographer to chromatographer, but most can be happily forgotten if you keep in mind what the instrument is actually doing. Tubing Tubing comes as steel or peek plastic (generally). Steel withstands a higher pressure so it s usually used throughout the manufacturer s hardware. Don t change it for Peek without careful thought; manufacturers often rely on it to hold the maximum pressure of the pump (at least as far as the end of the pump) in order to carry out their tests after maintenance. Some chromatography systems (Dionex) use peek throughout because they are designed for applications where small amounts of metal ions can seriously harm the column. Peek withstands most normal temperatures, but goes brittle in contact with halogenated solvents. Peek fittings can rarely survive much beyond 300bar. Peek is flexible and easy to cut, so it s ideal for most typical chromatography applications, especially reverse-phase, where solvents are not too aggressive, and the pressure unlikely to exceed 300bar. Cut it with a properly-designed cutter; uneven cuts lead to dead volumes and leaks (see fittings below). Cutter for stainless steel tubing Cutter for peek tubing Ceramic wafer for cutting silica tubing Training_HPLC_system_3953563.doc page 1 of 13
Most manufacturers seem to use the same colour coding for the diameter of peek tubes: Colour size in size in mm inches Red 0.005 0.127 yellow (green for Agilent steel tubing) 0.007 0.178 Blue 0.01 0.254 orange 0.02 0.508 Red tubing is suitable for flows from 50µL.min -1 to 1000µL.min -1. The larger sizes up to blue are best used on hplc systems where the flow rate will go up to 5 or 10mL.min -1. If you use tubing that is too wide, you will get broader peaks because analytes will diffuse more easily up and down the tube, and the rate of flow is slower. If you use tubing that is too narrow, the back-pressure will become excessive. Some systems use plastic-coated silica capillaries. These are cut with a ceramic wafer; gently rub the smooth edge of the wafer across the capillary, and then twang the end of the capillary so it snaps off. Fittings Nearly all hplc fittings have the same general design, but the actual measurements vary between manufacturers. The simplest approach is to use Peek fittings, which slide on the tube to match the hole into which they re screwed. If you must use metal, then do up the fitting once, firmly, in place, and never try to use that tube anywhere else. The bevelled ferule will become swaged onto the tube, and won t move again. It probably won t fit accurately in another hole, particularly of a different manufacturer. Always push the tube completely into the hole before doing up the fitting. With Peek fittings I find it helpful to slide the fitting up the tube slightly, push the tube into place, and then screw in the fitting while holding the tube firmly. A B C Fitting A is correct. Fitting B has too short a tube; it includes a dead-volume (the blue bit) and will cause peak-broadening. Fitting C has too long a tube, and will leak like a sieve. This doesn t often happen with Peek fittings, but is common in steel swaged fittings that have been screwed into the wrong manufacturer s port. Training_HPLC_system_3953563.doc page 2 of 13
If you use hexagonal peek fittings with a spanner, be awfully careful not to break them. It is not easy to remove the broken remnant from the hole. Solvents These are going to be pumped through a column made of very fine beads, and then through an analyser. It follows that you mustn t have anything that will disturb the pump, get stuck in the column, or cause trouble to the analyser. Particulate stuff Particles get stuck in check-valves, the column, filters, or anywhere else. Buy your solvents as hplc-grade, ready-filtered solvents, and keep lids on to avoid dust. Don t bother filtering them again; Fisher Chemicals are probably much better at filtering than you! Buy your water in the same way if you use a lot, or filter it yourself if not. If you are adding solid buffer salts, filter after dissolving the solid. Water encourages algal growth, so if you do choose to buy it instead of filtering it, remember it doesn t last for ever. Do use filters on the ends of the solvent lines, but don t rely on them as your sole means to stop dirt from getting into the pump. If you use salty buffers, remember that they may precipitate in solvents. Before changing a line to solvents, wash copiously in pure water. You may need to change the solvent line filters. Absorbant things Anything that your detector can detect, will cause a baseline. This is why acetonitrile is sold both as hplc grade and gradient grade. If you have a gradient of increasing acetonitrile, and it contains something that absorbs UV, you will see a rising baseline. It might not absorb enough to cause problems in isocratic runs, but it will make it hard to see small peaks. Buy a solvent good enough for your application, and don t let things get into it. Make sure that solvent lines are clean before putting them in your solvent, and keep bottles done up. Gases There are two problems with gases in the solvent mix. Firstly they are components of the mix (in effect, they are a different solvent). They may cause things to elute at different times. This is why sugar chromatographers worry about dissolved CO 2. More generally, gases tend to come out of solution, and make bubbles. They do this especially well in the pump head. As the piston moves back to draw solvent in, it is pulling against the inlet filter, and reducing the pressure. Any dissolved gas is drawn out of solution. The bubble so formed simply expands (gases are very springy), so the pump does not draw a full syringe-full. As the pump piston pushes back in, the Training_HPLC_system_3953563.doc page 3 of 13
bubble gets compressed, so the piston goes a long way without much liquid coming out. The result is that the pump supplies the wrong volume, and the pressure fluctuates drastically. To remove gases, either degas the solvent before you start, or use an online degaser. If you degas before you start, remember that the solvents will pick up gases with time. Teflon tubing is not gas-proof, and gases will diffuse into the solvent through the solvent line tubes; you can buy special gas-proof tubing for solvent lines. There are various ways to degas. The best involve applying a vacuum to the solvent, while stirring or sonicating. Some people sonicate without vacuum to degas. The principle behind this is that sonication causes cavities to appear in the liquid, which are at low pressure. Gas leaves the solvent to fill the cavity, which becomes a bubble, and rises to the surface. In theory this may be fine, but by experiment I m fairly convinced that half an hour in an average sonicating water bath makes no difference at all to a 500mL bottle of water. Inline degasers split each solvent line into a fine mesh of tiny tubes, much like bloodvessels split into capillaries. These are all confined in a box held under vacuum, so gases will diffuse through the tube walls and be pumped away by the vacuum pump. They are usually very efficient, but do increase the volume of tubing between pump and solvent bottle, so you ll have to purge lines slightly longer when changing solvents. Miscibility Not all solvents are miscible. If you transfer a solvent line from one solvent to another in which it can t mix, you will get a horrible mess that is difficult to clean. If in doubt, consult tables in chromatography catalogues, or chemical data books. The following table lists mixtures that are OK in any proportion: water methanol acetonitrile hexane ether isopropanol ethanol chloroform isopropanol, methanol, ethanol, acetonitrile, acetone any listed solvent except hexane any listed solvent except hexane not acetonitrile, methanol, water (or acetic acid) any listed solvent except water The magic solvent that can be mixed with any others like isopropanol, miscible with all others listed any listed solvent except water When you change solvents, you can use isopropanol as an intermediate to avoid miscibility problems. It is a good idea to make sure that all lines on a pump are in solvents that can mix, even if you re not using a particular line. One moment of carelessness pumping the wrong line can make a horrible mess otherwise. If you do get an immiscible mix in a system, pumping through a lot of isopropanol can be a method to get rid of it (it will wash out most other solvents). Training_HPLC_system_3953563.doc page 4 of 13
Pumps Pumps vary hugely. Old hplc systems had a single pumping head, which couldn t pump continuously. They relied on good pulse-dampeners to keep the flow going while the pump head refilled. Newer systems invariably have several pumping pistons to maintain a more constant flow, but they still have pulse dampeners. In the system below, piston A moves forwards at the correct rate, while piston B refills. When piston A reaches the end of its stroke, it moves back fairly rapidly, while piston B pumps very rapidly. Piston B pumps fast enough to refill piston A, plus enough excess to keep the normal flow going. Gradient proportioning system Without one-way valves, the pistons wouldn t create a forward flow. These valves, usually called check valves are often as simple as a small ruby ball sealing against a hole. Some systems use electronically-operated inlet valves to improve accuracy of pumping. Passive, ruby-ball check valve Electromagnetic active inlet valve Training_HPLC_system_3953563.doc page 5 of 13
The pressure of the chromatography system is important in holding a check valve closed. As a result, most pumps work really badly unless they re pumping against a decent back-pressure (30bar minimum). It s not a good idea to rely on an hplc pump to create a solvent mixture unless it s pumping through a pressure regulator (we usually use 50bar). You may also need a pressure regulator if you use very short columns. Simple chromatography is isocratic ; it uses a single solvent mix throughout. You can make this by hand and pump it with a single pump. However, most chromatography uses gradients, and even isocratic chromatography is often most easily dealt with by letting the pump prepare the mixture for you. Therefore virtually all hplc pumps are equipped to take and mix several solvents. There are two basic ways to do this. Low pressure pumps mix the solvents (often 4; a quaternary pump) in a small chamber before the pump heads, at low pressure. Then a single pump draws its solvent from this mixed pool. Each solvent enters the mixing chamber through an electronically-controlled valve. To prepare a mixture, the valves are simply opened for the appropriate percentage of the total filling time. Low pressure (in this case quaternary) High pressure (in this case binary) High pressure pumps have multiple pumping units, each handling a different solvent. They mix their flow after the pump. The mixture is controlled by setting each pumping unit to work at a different rate. High pressure systems are much more expensive, but generally more accurate and reliable. Pumps usually also have some means of detecting the pressure, and a pulse-dampener. This is a liquid-spring. There are various designs, often a diaphragm or a long wobbly tube. The spring is pumped-up as the pump runs, and evens out the pressure slightly by continuing to supply a flow if the pump momentarily stops, for instance while pistons change direction. Pressure pulses are bad for hplc-systems. They can cause problems for detectors, and create extra noise in the detector. They are also not good for columns. Training_HPLC_system_3953563.doc page 6 of 13
The pressure detector may be incorporated into the pulse dampener, or separate. You should always know where your pressure detector is. There s a tendency to assume that the pressure you see on the computer screen is the pressure wherever you are looking in the system; but if a tube has got blocked between the pump heads and the pressure detector, you could have no pressure reading at all, but a truly phenomenal pressure in the pump head itself. Priming Some pumps don t work when they re empty, and therefore can t fill themselves. These need to be primed. Follow the manufacturer s instructions. Assuming you ve inherited the use of an instrument whose manuals have long since disappeared, you may be lucky enough to find a.pdf file somewhere as an electronic version. If all else fails, the check-valves are designed to allow forward flow only, so turning the pump on and sucking solvent from the outlet of the pump with a syringe is usually a good way to prime it. Purging The pump will not supply the correct solvents when all its tubes are full of the wrong things. After changing solvents you will need to purge the solvent lines. Usually there is a valve somewhere that can be opened to direct the pump output to waste and allow you to pump at a much higher rate than you could get through the rest of the system. Autosamplers The autosampler has to get the sample into the flow of solvent so it can reach the column. There are numerous ways to do it; you should look at the literature that came with your autosampler. The next pages illustrate two methods. In the first, the needle assembly is never exposed to high pressures, and merely acts as a syringe, transferring the sample from the vial to a sample loop. The sample loop is then moved into the solvent stream to complete the injection. In the second, the needle assembly normally remains in the high pressure flow-path. It is removed from the flow-path temporarily to draw sample from the vial. Training_HPLC_system_3953563.doc page 7 of 13
Low pressure injection system Needle takes sample from vials in racks (ends of racks B, C, D, E visible) and squirts into sample loop (orange on plumbing diagram ) via needle-seat. Main flow is shown in red, low-pressure needle-flow in blue. Valve-channels are shown in green. (Column) Solvent allowed to warm here before entering column From syringe To waste From pump Training_HPLC_system_3953563.doc page 8 of 13
High pressure injection system Needle Needle seat Lots of tubing Metering device (Column) Training_HPLC_system_3953563.doc page 9 of 13
In most systems, there will be a loop somewhere that holds sample before it enters the main solvent stream, and there will be a syringe that draws it up. Somewhere a valve will remove the loop from the solvent stream so that it can be filled with sample, and put it back in to wash the sample onto the column. The maximum volume you can inject depends not only on the syringe, but also on the loop! In loop-injection systems, such as the low-pressure example shown above, the most exact way to make an injection is to fill the loop totally, with an excess on both ends, and then rely on the injection valve to cut off the ends of the liquid column of sample. However, this gives you no way to change the volume you inject without changing the loop. Therefore most autosamplers use the syringe to measure volumes, and partially fill the loop. In general it is not a good idea to fill a loop more than half full if your system uses the loop as part of the injection-volume control process. As a rule of thumb, if you can easily find a sample loop with its volume marked, assume you should not be injecting more than half this volume unless you are doing full-loop injections. If you can t find a sample loop, you can assume that the manufacturer has designed the instrument in such a way that the syringe alone controls the volume reliably, and you can use any volume the instrument will accept. Needles go from vial to vial; they can carry sample from one run to the next. Most autosamplers provide needle-wash systems to avoid this. It may be as simple as dipping the needle in a spare vial of solvent. If so, don t put a lid on the wash-vial. Otherwise the lid becomes the thing that causes the cross-contamination, instead of the needle... It s important to understand your autosampler. Around our lab you will find an Agilent system (which takes from the vial exactly what it injects), a Thermo Surveyor system (which takes rather more than it injects; worth knowing if you intend to make multiple injections from a limited sample), and Waters Alliance systems (which also take exactly what they will inject, but take it from about 5mm above the bottom of the vial; catastrophic if you only have 4mm of sample in the vial). Columns Read the manufacturer s guidance on any column, and make sure you understand how the separation works. I am assuming reverse-phase for the remainder of this section. Columns come in all shapes and sizes. In isocratic separations you are relying on differing rates of movement of chemicals along the column. The longer the column, the more separate the blobs of chemical. You will need a long column. If you re doing gradient separation, you are hoping that some chemicals will fall off the column at lower solvent percentages than others. It may be possible to achieve separation with a very short column. The thing that determines whether you ve got an analyte off the column is how much solvent of an eluting strength you ve pumped past the beads to which the analyte was stuck. If the column is twice as large, it will have four times the cross-sectional area, and you will need to pump four times as much solvent to achieve the same result. Longer columns also need more solvent in general. Training_HPLC_system_3953563.doc page 10 of 13
250mm 4.6mm diameter 100mm 2mm diameter Guard holder 50mm 30mm In the past it was necessary to use quite high pumping rates as the dead volumes of the system were large. Detectors were big; if you have a flow cell of 1mL volume and pump at 1mL.min -1 you can hardly expect to resolve peaks only a minute apart. One peak will still be in the detector flow-cell as the other arrives. Modern systems are designed with low volumes. Since many detectors (including electrospray MS and UV absorbance) measure concentration, not quantity, there is no point in using high flow rates. For instance, a 4.6mm column works well at 1mL.min -1 and has a cross-sectional area 5 times that of a 2mm column. If you replace it with a 2mm column you will be able to work at 200µL.min -1, use one fifth of the solvent, and prepare one fifth as much sample (or increase your sensitivity five-fold if material is limiting). Analytical columns are expensive. They can be messed up in various ways. Particulate matter can block them. Chemical contaminants can bind to them and alter their chemistry, possibly permanently. The column material can be dissolved or degraded; this either changes the chemistry of the column, or causes voids to appear, which are really bad for peak-width. One common protective measure is to use a guard column. This is a short column of the same material, just upstream of the analytical column. Anything that could stick to/get stuck on/dissolve/degrade the main column will do it to the guard column. The guard can be washed or replaced, preserving the analytical column from damage. If you have an expensive guard column you may try washing it; take it off the analytical column first. You don t want to wash the contaminants onto the analytical column. There is much debate about how useful guard columns are. It depends on your sample. If you always use carefully cleaned samples (perhaps from solid phase Training_HPLC_system_3953563.doc page 11 of 13
extraction), carefully filtered, perhaps you don t need a guard. If you can t filter your sample and are relying on spinning it well, and your sample is perhaps a horrible mixture, a guard might be good. C18 guard columns can be very cheap, costing as little as 30 spin-filters. Under these circumstances it s almost worth using the guard as a filter... Older chromatographers worry that columns can be damaged by sudden pressure changes, sudden changes of solvents, and general physical abuse. At the ends of gradients, they often change the solvent back to the starting conditions nice and gradually, and they never just turn a pump on at its full rate. All this caution is utterly unnecessary for modern silica columns. They can survive nearly anything, especially falling off the bench. I have heard of bent columns still performing within specification. Columns should be stored in the manufacturer s recommended solvent mix to prolong their life. If in doubt, I tend to use 70% acetonitrile for C18 columns. This isn t quite as good as 100% acetonitrile, but runs less risk of precipitating salts. Designing runs When you design gradient runs, it s a good idea to equilibrate the column before the next run; otherwise retention times will be short and unreliable. There are various rules of thumb about how long the reequilibration should be. I use ten column void volumes, on the assumption that 60% of the column is void. Thus, for a 100mm 2mm column running at 250µL.min -1 : 2 volume= πr length 60% = 3.142 1 100 60% = 189µL Ten column volumes are therefore 1890µL 1890 equil. time= 250 = 7.6 min 2 A shorter equilibration time will be adequate if the gradient didn t change the percentage mix of solvents very far. Solving problems There are probably more check-list guides for solving chromatography problems than there are for any other lab procedure (try the chromatography suppliers catalogues and literature). Therefore I m providing just a few tips: Gather all the information you can; things sometimes go wrong during a run, half way through the night. Look to see if you can find a log-file. See if the pump pressure is available as a signal in your data file. Go at it logically; see if you can work out what physical explanation could cause that symptom. Then test. Training_HPLC_system_3953563.doc page 12 of 13
Don t change more than one thing at once without checking what the results of your changes might be. Bits of tissue-paper make excellent leak-detectors if your system doesn t tell you automatically. When looking for the source of high back-pressure, work your way along the system logically, undoing bits and seeing where the pressure suddenly drops. When removing small parts from a complicated instrument, do try to block any obvious holes down which the part could fall. It s not fun retrieving a small screw from the drainage system of a mass spec spray chamber. When exchanging a suspect part, you might not want to throw away the old part until you know it was bad. Keep it in a separate, labelled bag until you re sure. It s amazing how difficult it is to tell which is the new, and which the old, when you have two check-valves sitting side by side on the bench. Pump maintenance The usual things that go wrong are piston seals and check-valves (i.e. the moving bits). Leaky piston seals lead to puddles round the pump-heads (or stalactites if you re working with salts). Replace seals (and scratched pistons) following the manufacturer s advice. Check-valves get stuck, stopping the pump altogether, or causing it to pump wildly wrong rates and mixtures. Manufacturers generally like to sell you a new check-valve, but you can try taking the old one out and sonicating it in a variety of solvents first. Only try this on simple ruby-ball valves; complicated electronic active valves are hopeless when they go wrong. Symptoms of pump problems are usually horrible pressure fluctuations, and sudden drastic changes of retention time. Before panicking, always check the pump is properly purged and bubble-free. No part of an hplc should be left in corrosive solvents. Water is also a bad storage solvent because things will grow in it. Isopropanol is good for long-term storage. We also often use 50% methanol; this is excellent as it doesn t corrode the system, doesn t allow algal growth, and doesn t precipitate any residual salts from salty chromatography. Autosampler maintenance Usually autosamplers need only periodic attention from the manufacturer when the valve rotor is getting worn. You may have the occasional needle-accident; this is probably a useful spare part to keep. Never argue with an autosampler needle; we have seen them go through vials, and force bits of hplc system through a vial septum without batting an eyelid. Training_HPLC_system_3953563.doc page 13 of 13