Ionization detection methods include flame ionization detection (FID) and, among others, electron-capture, photoionization, and thermionic-specific — also called nitrogen–phosphorus — detection. Ionization detectors interact with solutes eluted from gas chromatography (GC) columns to produce a current that varies in proportion to the amount of solute present. FID is sensitive to molecules that are ionized in a hydrogen–air flame, including most carbon-containing compounds (with some notable exceptions; see Table I). The other detectors rely upon substance-specific ionization mechanisms and respond only to certain heteroatoms such as halogens, nitrogen, or sulfur for electron-capture detection; nitrogen and phosphorus for nitrogen–phosphorus detection; or to specific chemical structures such as aromatics for photoionization detection. As ions are formed inside the detector, they are impelled by an electric potential toward an electrode, producing a minute current on the order of picoamps (10

  A) for FID. This current is converted to a voltage, filtered, and amplified as required.  

Although these ionization detectors share several features, their implementations differ significantly. Detector performance characteristics (sensitivity, minimum detectable quantity, linearity, and selectivity) are affected strongly by the type of ionization mechanism, the internal electrode arrangement, and the electronics. 

The internal arrangement of a typical flame ionization detector is shown in Figure 1. Carrier gas from the column enters at the bottom of the detector and is mixed with hydrogen combustion gas plus optional makeup gas in the area below the flame jet. This mixture is then combined with air and burned just above the jet tip. A negative polarizing voltage is applied between the jet tip and a collector electrode; as electrons are formed, they are accelerated across the jet tip–collector gap by the electric field and sent to an electrometer. Depending upon the FID design, either the collector or the jet tip is kept at ground potential; Figure 1 shows a grounded collector design. Air, carbon dioxide, and water exhaust gases are vented from the top of the detector body. In some flame ionization detectors, a glow-plug operates momentarily to ignite the flame.

Table I:  FID relative sensitivities for various compounds and classes			

Figure 2 shows a schematic of detector electronics. From left to right: a 200-V polarization voltage is applied across the flame jet and the collector. Electrons formed in the flame by combustion of hydrocarbons are collected under the influence of the electrical field, and the resulting current is converted to a voltage by an electrometer that can have one or more operating ranges. The voltage is amplified and high-frequency components are filtered out. The detector signal is converted to discrete digital samples by an A/D converter and additional signal processing is applied as required. This is a typical implementation; there are many other possibilities. Several amplification ranges are available typically, ranging from about 1 pA (input)/mV at the highest output sensitivity to about 10 nA (10 X 10

  A)/mV at the lowest output sensitivity. Some GC systems use a logarithmic amplifier that covers the entire dynamic range. The detector signal is filtered to remove unwanted high-frequency noise. Noise is produced by instabilities in the flux of ionizable compounds in the carrier gas, by the flame itself, by the electronic circuit, and by induction of stray electromagnetic signals (from cell phones, for example). The FID electrometer and amplifier circuits impose an electronic limit on the response speed, and additional, more sophisticated signal processing is carried out in the GC firmware as well as the data system. For most capillary GC peaks, a response time of about 200 ms is appropriate and will reject the majority of detector noise while delivering better than 95% peak-shape fidelity. A response time of 50 ms or lower is required for fast capillary peaks (width at half-height << 1 s) as encountered in high-speed or comprehensive GC X GC separations. A too-fast response will not affect peak shapes, but will pass extra noise through the system and potentially worsen minimum detectable quantities. The A/D conversion rate affects the signal fidelity as well. In general, the sampling rate should be twice the maximum frequency of interest in the signal. For example, a response time of 200 ms corresponds roughly to 5 Hz and, thus, would be sampled at 10 Hz or greater. A recent installment of "GC Connections" discussed signal processing and peak shape in more detail (1). 

Figure 1: Flame ionization detector cross-section. 1  electrometer connection, 2 = effluent exit, 3 = igniter coil, 4 = igniter power connection, 5 = polarizing voltage supply connection, 6 = air input, 7 = column connection, 8 = hydrogen input, 9 = flame jet, 10 = collector electrode. (Derived from a figure courtesy of PerkinElmer Instruments, Shelton, Connecticut).			

Overall FID sensitivity depends upon the combustion gas flow rates, the carrier gas flow rate, the flame jet exit diameter, the relative positions of jet and collector, and — to a lesser degree — the detector temperature.  

Combustion gas flow rates must be set correctly for proper FID operation; follow the manufacturer's recommendations for air and hydrogen flow settings. In general, the air:hydrogen ratio should be approximately 10:1. A hydrogen flow rate of 30–45 mL/min with a corresponding air flow of 300–450 mL/min is common. FID sensitivity will be reduced as hydrogen flow deviates above or below optimum, as illustrated in Figure 3. The linear dynamic range is affected by hydrogen flow as well: higher flows tend to reduce the linear dynamic range. There is little reason to operate a flame ionization detector far off the manufacturer's gas flow settings; they have been carefully optimized for that specific detector. Air flow is less critical than the hydrogen flow, but too much air will destabilize the flame causing noise and possible flame-out. Not enough air will reduce sensitivity and shorten the linear dynamic range.

Figure 2: Flame ionization detector electronics.			

Hydrogen is highly flammable and can cause a serious explosion if allowed to build up in an enclosed space such as the GC oven. Never turn on the hydrogen flow without a column or blank fitting attached to the detector's base to prevent hydrogen from leaking into the oven.  

Standard FID jets have exit diameters of approximately 0.5–0.7 mm, which is suitable for most applications. A smaller jet of about 0.3 mm i.d. is sometimes used with capillary columns to gain sensitivity (about 1.5X); rarely, problems can be encountered with solvent-peak flame-out. A narrow FID jet is not recommended for packed-column use because stray column packing support easily can clog the jet passage. Conversely, narrower jets prevent the tip of a capillary from accidentally protruding into the flame. 

 The carrier gas flow rate is an important consideration for detector sensitivity. For packed or micropacked columns, carrier flow normally will be greater than about 8–10 mL/min. If the packed column flow is under 40 mL/min, then the standard hydrogen flow need not be changed. When the packed column flow rate exceeds 40 mL/min, it might be necessary to increase the hydrogen flow somewhat to achieve a stable, sensitive flame. Larger jet diameters (0.7 mm) are also beneficial at higher carrier flows. The choice of carrier gas — other than hydrogen — such as helium, nitrogen, or argon does not affect detector operation significantly. 

Figure 3: Effect of hydrogen flow on relative FID sensitivity. This is a representation of typical results.			

Different flow considerations apply for capillary columns. Chromatographers can choose to operate columns with inner diameters of 0.53 mm and greater well above their optima at relatively high carrier flow rates of 10–20 mL/min. Normally, a flame ionization detector does not require any special attention under such conditions. When capillary columns with inner diameters of 0.32 mm or less are used, or when wide-bore columns are operated closer to optimum flow rates of less than 10 mL/min, flame ionization detectors can benefit from the addition of makeup gas to the carrier stream before entering the jet area. Makeup gas has two important effects. First, it maintains optimum carrier gas flow through the jet and keeps the detector operating at the best sensitivity and linear dynamic range. Second, for some detectors makeup flow sweeps out the area under the jet and inside the detector base, alleviating any peak broadening that might be produced as capillary peaks encounter larger-diameter internal passageways. Follow the instrument manufacturer's instructions regarding makeup gas and its flow. 

Hydrogen, while sometimes used as a carrier gas for packed columns, is commonly used with capillary columns. Hydrogen carrier gives a wider range of optimum linear velocities or flows, is less expensive than helium, and can be generated from water on-demand with an appropriate hydrogen generator. With FID, it is convenient to compensate for the added carrier gas hydrogen by reducing the detector hydrogen flow correspondingly. For a column flow rate of 5 mL/min, for example, detector hydrogen flow should be reduced by 5 mL/min so that the total hydrogen flow through the jet is at the optimum level. 

Column oven temperature might affect carrier flow rates, depending upon the carrier supply operating mode. When changing the oven temperature, the total hydrogen flow through the detector should remain constant. Electronic pressure programming systems accomplish this by determining the carrier gas flow rate as the oven temperature changes and adjusting the detector hydrogen flow accordingly. If running in constant-pressure mode, the carrier gas flow will drop off as the oven temperature increases; the detector hydrogen flow would be increased accordingly. If running in constant carrier gas flow mode, then the FID hydrogen flow would remain constant as well. 

 The sensitivity of a flame ionization detector does not depend strongly upon its temperature, provided some conditions are met. The appropriate detector temperature is determined by the greater of the following two conditions: a minimum temperature of 150 °C for stable detector operation and a minimum temperature of approximately 20–50 °C above the highest column temperature. The detector produces a large amount of water vapor, which can condense in the cooler upper areas around the collector if the detector base temperature is less than about 150 °C; this condensed water vapor can produce noise and baseline drift. On the other hand, the detector base must be hot enough to prevent condensation of peaks as they are eluted from the column, so it should be kept somewhat warmer than the highest operating column temperature.  

If a capillary column is installed with its end inserted into the detector base up to the jet and is operated at oven temperatures approaching the column's maximum rated temperature, then it is possible that the end of the column will be overheated in a detector base that is another 20 °C warmer. Such overheating can produce excessive detector noise from decomposing stationary phase, cause solute adsorption onto subsequently exposed column surfaces, and reduce column life. A capillary column detector adapter that positions the end of the separation column in the oven and conducts carrier gas flow along a glass-lined tube or via a piece of deactivated fused silica into the detector jet will help alleviate such problems. 

Carrier and combustion gas purities, pressures, and flow rates as well as detector and column temperatures are all important considerations when setting up a flame ionization detector. There are several steps to follow. First, be sure that all gases are of sufficiently high purity and that their supply pressures are stable enough to provide reliable operation. Then, with the instrument turned on but unheated, set the required gas flows. Finally, heat the injector, detector, and column to their operating temperatures and ignite the flame. These steps are detailed in this section. 

Flame ionization detectors are quite sensitive to hydrocarbon impurities that can be present in gas cylinders or connecting tubing. Hydrocarbon impurities in the combustion gases will cause increased detector noise levels as well as higher baseline signal levels. Hydrocarbon filters are recommended for installation at the external GC bulkhead fittings for air, hydrogen, makeup gas, and of course, the carrier gas. It is not necessary to remove oxygen from the FID hydrogen stream, but an oxygen filter on the carrier line is recommended highly as well, so be sure to trap oxygen if hydrogen flame gas is used as the carrier gas as well. 

Hydrogen for FID alone should be of 99.995% purity or better. If used for carrier gas, then 99.999% or better purity is preferred. There are several excellent commercial hydrogen generators that can produce sufficient carrier-grade hydrogen to supply dual flame ionization detectors plus one or two carrier channels with split injectors. If an electrolytic hydrogen generator is used, be sure the water you add is free of hydrocarbon impurities.  

Air for FID should contain less than 100 ppb of hydrocarbon impurities. Aside from standard compressed gas tanks, a variety of suitable purified air generators are available with capacities ranging from a couple of chromatographs up to an entire laboratory's worth. Older air compressors, or so-called "house" air supplies, should not be used with gas chromatographs except to supply operating pressure for pneumatic valve actuators. 

Carrier-gas purity is also important for proper detector operation — with or without makeup gas. Makeup gas impurities affect the detector in much the same manner as combustion-gas impurities. Even without makeup gas, impurities in the carrier gas eventually can pass through the column and onto the detector. In temperature-programmed operation, such impurities might appear as broad ghost peaks during a run or as a steadily rising baseline similar to column stationary-phase bleed. In isothermal operation, impurities might appear as a slowly rising baseline with increasing noise, often over a period of hours to days. Unfortunately, a heavily contaminated gas chromatograph often proves difficult to clean up. Even though the column can be baked out or replaced, impurities can remain in the internal gas lines, valves, and regulators after the contamination source is corrected. The best procedure is to assume that there can be a gas purity problem right from the beginning and install appropriate filters. Bear in mind that the best filter is one that is never needed because the incoming gas is consistently pure. On the other hand, assume that there will be a problem with incoming gas purity at some future time. Filters are an excellent insurance policy against contaminating an instrument. 

Connecting tubing from the gas source to the instrument also can sometimes cause problems with contamination. Be sure to use copper or stainless steel tubing specially cleaned for chromatographic applications. Never use plastic tubing because significant amounts of plasticizer or monomer can be present. In addition, all plastic tubing is pervious to atmospheric oxygen. Leaky fittings also can be a contamination source: they allow some atmospheric gases to enter the instrument gas stream inside. Avoid leakage by ensuring that all fittings and ferrules are in good shape and not over-tightened. It is better to cut off a few inches of tubing and install a new nut and ferrules than to try to seal a leaking connection by overtightening. 

 Two situations arise when setting FID flow rates, depending on whether the gases are electronic pressure controlled (EPC) or manually controlled. With EPC systems, the flows are set on the instrument keypad. Do not assume, however, that the flows are correct — regular flow calibration is highly recommended. I like to measure the detector flow rates anyway. Be careful to enter the related settings that control the carrier gas mode of operation (constant pressure, constant flow, or constant velocity) and makeup gas flow. Also, bear in mind that in some GC systems, the flow rates depend upon the incoming gas pressure — if the pressure changes, then the flow controllers should be recalibrated. 

For manually controlled detector gases, as well as when directly measuring detector gas flow rates, it is easiest to operate with the column connection in the oven blocked off with a blank ferrule or plug fitting. If the column is installed, then carrier flow should be enabled for capillary column installations, in which the column end is in the detector. In this situation, the operator will need to correct measured combustion gas and makeup flows for the column flow rate. Attach a calibrated flow meter to the detector's exit with the appropriate adapter and turn off the air, hydrogen, makeup, and carrier gas flows at the instrument. Be sure to set the tank regulators to recommended pressures and turn on any in-line shutoff valves. The hydrogen flow is the first to set. Turn on the hydrogen and set the correct flow rate, following the adjustment instructions in the manual. After turning on the flow, be sure to wait a minute or so for air to be purged out of the hydrogen lines for a more accurate reading. 

Next, set the makeup flow, if used. Turn off the hydrogen flow and then turn on, measure, and adjust the makeup flow. If the hydrogen cannot be turned off conveniently, then subtract the measured hydrogen flow to find the makeup flow rate. Be careful, however, when using an electronic flowmeter. If your meter has settings to select the type of gas being measured, then it will produce inaccurate readings for gas mixtures. This is not a problem for a simple soap-bubble flowmeter, although readings should be corrected for ambient pressure, temperature, and the vapor pressure of the soap-bubble solution. The details of using a bubble flowmeter are found in reference 3 as well as in many instrument manuals and other chromatography books. 

Third, set the air flow rate. This might require a larger-volume flowmeter to accurately measure the 10-fold higher flow. Again, it is best to turn off the hydrogen and makeup flows, but you can correct the measured air flow rate if necessary.  

Finally, if it is not already on, then set the carrier gas flow. If you want to measure the carrier gas flow rate directly at the detector, then turn off the air, makeup, and hydrogen flows. Adjust the carrier gas flow controller, pressure regulator, or EPC system as required. Once column flow is established, and not before, the column and detector can be heated to their operating temperatures. 

Accurate direct measurement of capillary-column flow under about 5 mL/min requires a suitable, low-volume flow-measuring device. For an EPC system, remember that with a split inlet system in constant or programmed flow mode, the system maintains column flow by calculating and setting the pressure drop required to produce the desired column flow based upon the oven temperature, the type of carrier gas, and the column dimensions entered by the operator. If the entered dimensions do not reflect the actual dimensions accurately, then column flow and velocity errors will result. If there is any doubt, see the instrument manual for a procedure to set, measure, and correct the dimensions on the basis of the measured column average linear carrier gas velocity. 

While the instrument is heating up, turn the combustion gas and makeup flows back on if necessary. You can ignite the flame as soon as the detector temperature has passed 100 °C. Most flame ionization detectors require users to temporarily reduce the air flow during ignition. Like a choke on an automobile, this reduced air flow creates a momentarily rich mixture that is easier to ignite. Some instruments have built-in igniters that are operated by pushbutton or from the keypad, while others have manual igniters that must be held over the detector as an internal glow wire is heated electrically. Some rely on a piezoelectric igniter. In any case, ignition is most often accompanied by an audible "pop." 

 Do not lean over the FID to see the flame (it is invisible), and always wear appropriate eye protection. Do not allow any clothing to come near the detector exit. 

After the flame appears to have been ignited, check for combustion water vapor by holding a cold, shiny object such as a mirror or the polished end of a wrench directly over the FID exit — you should observe "steam" condensing on the cold surface. If you do not, the flame probably has not ignited or has gone out immediately. 

Flame ignition problems have several causes. Foremost is an incorrect flow setting — or possibly you forgot to turn on one of the flows. Make sure that all flows are correct and that the gases are connected correctly at the back of the instrument. Flame ionization detectors will produce a very loud "pop" on ignition if the hydrogen and air lines are reversed, but usually the flame will go out immediately. Be very cautious in such cases because a large, invisible hydrogen flame that extends several centimeters above the detector can result from reversed connections. 

Continued ignition difficulty might be due to a defective igniter or other hardware problem. To check a built-in igniter, first turn off the hydrogen flow. Then press the igniter button while indirectly observing the inside of the detector with a small angled inspection mirror. For a manual igniter, observe the internal element; you should see an orange glow, or with a piezoelectric igniter you will see the spark. If not, then check the igniter connections and replace the igniter element if needed. 

Other hardware problems that cause difficult ignition include a broken or cracked flame jet, poor detector or column installation causing leaks around the detector body, or a poorly fitting flow-measurement adaptor plug that gives inaccurate flow measurements. If the detector had been operating well and then suddenly quit, check for a blocked jet tip by measuring the hydrogen flow. If required, replace or remove and clean out the jet carefully with a cleaning wire, following the manufacturer's maintenance procedures. 

Sometimes the flame can blow out just after injection; the solvent peak can be large enough to interrupt the flame. If this occurs often, change to a flame jet with a larger internal diameter if possible, and adjust the hydrogen flow to more closely match the carrier flow rate, being mindful of a possible sensitivity compromise. If the problem persists, you should try reducing the amount injected, using a lower carrier gas flow rate, or both. If you are using a 0.53- or 0.75-mm i.d. capillary column, the problem might be due to the proximity of the column exit to the flame jet. It might be helpful to withdraw the column somewhat or to install a glass-lined detector-column adapter or piece of deactivated fused silica between the column tip and the flame jet. 

Flame ionization detectors generally are reliable once they are properly set up. Operators can check a few key areas immediately when previously good detector performance falls below the minimum required for the application. Flame ionization detectors are subject to two broad trouble categories: contamination and electronics. Of these, contamination is by far the more common. 

Everything that passes through a flame ionization detector is burned in the hydrogen flame. For carbon-based substances within normal levels, carbon dioxide and water are formed. Large amounts of chlorinated compounds or carbon disulfide, however, are not burned as efficiently as hydrocarbons. These materials can produce significant quantities of carbon particles (soot) as well as hydrogen chloride in the case of the chloromethanes and carbon tetrachloride. Carbon particles tend to aggregate between the jet and the collector, forming an electrical leakage path, and the result is a high, noisy baseline. Hydrogen chloride from chlorinated solvents can be tolerated in small quantities, but after extended exposure in combination with the water of combustion hydrochloric acid, will begin to corrode the detector's inner surfaces, producing electrical leakage paths and a high, noisy baseline. 

Another common contamination source is stationary phase bleed from the column into the detector. Although this is not generally a problem for most capillary columns, packed columns as well as thick-film capillaries can emit significant amounts of stationary phase during their life, especially at elevated temperatures. Siloxane polymers produce silica when burned in a hydrogen flame. In a flame ionization detector, these silica particles tend to adhere strongly to the jet and collector surfaces inside the detector. These, in turn, can reduce the detector's sensitivity and increase the background signal level. 

To check for detector contamination, shut off the combustion-gas flows and turn off the power to the instrument. After the instrument has cooled sufficiently, remove the detector covers and examine the outside of the detector body near the detector exit. It should be clean and completely free of colored deposits. Look down into the detector. Again, the surfaces should be clean and free of deposits. If you observe some colored material inside the detector, remove the collector electrode for a closer look. A black deposit indicates carbon formation. White or gray deposits are typical of silica contamination, and green or blue-green deposits or corroded areas are a sign of excessive acid formation. 

Light deposits of silicon dioxide or carbon usually can be removed from the collector by gentle scrubbing with distilled water and surfactants or in an ultrasonic bath. Be sure to first remove the collector electrode from any attached electrical connections. Ceramic insulators from inside the detector also can be cleaned in this manner. In general, follow the manufacturer's recommended maintenance procedures. Detector parts that have been corroded should be replaced as cleaning is usually ineffective. 

Articles

GC Connections

Flame ionization detection (FID) is the most commonly used gas chromatography (GC) detection method. Flame ionization detectors respond to a wide assortment of hydrocarbons, have a large dynamic range, are...

The Flame Ionization Detector

The minimum peak size for reportable results for a liquid chromatographic (LC) method depends upon the application and the signal-to-noise ratio (S/N). The signal for the analyte peak and the baseline noise both contribute to S/N. This month's installment of "LC Troubleshooting" takes a look at the measurement of S/N, some techniques for determining the detection limits for a method, and some tricks to improve S/N. 

With today's data systems, S/N often is a number calculated as part of a data report for an analyte and can be based upon a mathematical treatment of the data, such as a root-mean-square determination of noise. Manual measurement of S/N can be performed, as illustrated in Figure 1. Select a section of baseline free of peaks, expand the plot scale so that the noise is easy to determine, and draw two lines tangent to the noise. The vertical distance between the two lines is the noise for the chromatogram. The signal is measured from the midpoint of the noise to the top of the peak. S/N is obtained by dividing the signal by the noise. Use whatever measurement units are convenient — millivolts, absorbance units, millimeters — they will cancel out and leave a unitless quantity.

Most readers of this column work for the pharmaceutical industry, in which sample types fall into two major categories. The first is pharmaceutical analysis, which focuses on analysis of the active pharmaceutical ingredient (API) in a standard (called the drug substance) or in a formulated product (called drug product analysis). In cases for which the analysis is for quantification of the API in either drug substance or product, precision and accuracy limits in the 1–2% range are required. (It should be noted that strictly speaking, we are talking about 

accuracy, but common usage prevails for the terms precision and accuracy for these parameters.) The limits are more lenient for impurities, degradants, and minor components in the same samples. The second type of analysis is measurement of the drug in a biological sample, such as plasma, and is called bioanalysis (or just bioanalytical). Bioanalytical work has a much wider tolerance than pharmaceutical analysis — 15–20% precision and accuracy in most cases. (Similar precision and accuracy can be acceptable for trace analysis of pharmaceutical impurities or degradants.) This increased allowance for method variation is because S/N in bioanalysis is much smaller than pharmaceutical analysis due to lower drug concentrations and often more background. 

If the acceptable method precision differs by an order of magnitude between bioanalytical and pharmaceutical analysis, it is obvious that an acceptable peak in one application can be unacceptable in another. It really has to do with the method's tolerance for error and S/N. I like to use a rule of thumb to help me determine how small a peak can be and still generate usable data: 

where %RSD is the percent relative standard deviation (sometimes called the coefficient of variation [CV]). Equation 1 gives us an idea of what S/N is required for a desired method precision. For example, if the %RSD for a method's precision can be no larger than 2%, this means that S/N = 25 is required (this assumes that S/N is the primary contribution to %RSD, which might or might not be true). For bioanalytical methods, S/N = 2.5 would translate into %RSD ≈ 20%. 

Let's see how S/N relates to limits of detection and quantification. Some workers consider S/N = 3 to be the limit of detection (LOD) for a method and others use S/N = 10. From Equation 1, this converts roughly to 15 and 5% RSD, respectively. I know of companies that define the LOD as the sample that generates peaks ≤ 30% RSD and limit of quantification as ≤ 10% RSD. These would correspond to S/N ≈ 1.7 and 5, respectively. It is clear that the criterion that determines the lower limits of a method can vary by an order of magnitude, depending upon the user and the application. The advantage of using S/N to define the LOD and lower limit of quantification (LLOQ) is that such determinations can be based upon a single injection, rather than several injections required for a statistical result. Thus, one can confirm S/N quickly during a system suitability test without running many samples. 

For the 1–2% precision and accuracy required by pharmaceutical methods for potency, it can be seen that large S/Ns are needed. Fortunately, for this type of method, the chemist rarely is sample limited, so injection of a higher concentration sample usually is convenient. For methods in which impurities or degradants must be reported, one generally needs to quantify any peaks larger than 0.1% of the main peak. What kind of confidence can you place in the results for peaks at this level? Some rough calculations allow us to predict precision for a method using UV detection.  

UV detectors are specified to be linear to 1.0 absorbance unit (AU), and some models are linear for larger signals. If we inject a peak that is 1.0 AU tall, what S/N can be expected at the 0.1% level? A typical UV detector specification is for noise of ±5 X 10

  AU with a dry flow cell. With mobile phase flowing through the cell, one should be able to get within about 10-fold of this value, let's say 5 X 10

  AU — about twice the noise level, not S/N > 10, so Equation 1 suggests RSD > 25% for these conditions. And a noise level this low might not be reasonable with a real sample. Is it a hopeless situation? Perhaps it is if we use peak height for quantification, but use of peak area generally will solve the problem. Although peak height and peak area should be proportional, by the time a peak is large enough to be 1 AU tall, it often is exhibiting some column overload. In such cases, the peak is broadened from overload, so the area increases faster than the height. This means that an impurity at 0.1% of the peak area of a 1.0-AU tall peak will be taller than 1 X 10

  AU. Even so, it requires quiet baselines to quantify peaks at the 0.1% level. 

The previous discussion makes it clear that low detection and quantification limits require S/Ns  consistent with the method precision and accuracy requirements. The only way to lower the LOD and LLOQ is to reduce the noise, increase the signal, or both. Let's look at some of the options available. 

Reduction of baseline noise can be accomplished in several ways. These include the following. 

Signal averaging: The detector time constant and data system sampling rate determine the amount of smoothing that takes place for the signal. General guidelines are to set the detector time constant to about one-tenth of the width of the narrowest peak of interest. Similarly, the data system generally defaults to 10–20 data points across the peak. Larger time constants and slower data rates average the signal more, resulting in a reduction of noise, but excessive averaging will reduce the signal for peaks of interest as well. However, changing these values from the defaults can reduce noise without compromising the signal — it is worth checking.

Temperature control: Variation of the column temperature and the temperature of the mobile phase entering the detector can create undesirable noise. For best results, use a column heater, insulate the tubing connecting the column and detector, and protect the detector from drafts. Although the overall laboratory temperature can be constant, the local temperature can vary significantly, especially if the LC system is near a heating vent.

Reagent and solvent purity: Use high-performance liquid chromatography grade (HPLC-grade) solvents for the lowest background signals. Also use high-purity reagents for sample preparation. Match the injection solvent with the mobile phase.

Additional pulse damping and mixing: Most LC systems are plumbed to obtain small system dwell volumes (the volume between the solvent mixer and the head of the column). Additional mixing volumes and pulse-dampening devices often will reduce baseline noise, but at the expense of increased dwell volumes. Dwell volume differences can be a problem for transfer of gradient elution methods, but only affect system washout and solvent changeover for isocratic methods. Improved mixing for gradient methods often can be obtained by premixing the A and B solvents. For example, a 5–95% gradient generally will have a quieter baseline than the pure solvents are mixed online if 5% B is mixed into A and 5% A into B with a gradient setting of 0–100%. Isocratic methods will have quieter baselines if the solvents are mixed manually.

Sample cleanup and column flushing: Sample cleanup steps will reduce the amount of extraneous material that gets introduced onto the column and generally will result in lower baseline noise. If the column is flushed with strong solvent at the end of each run, strongly retained materials will be eluted from the column, and it will reduce background noise.

Increase of the analyte signal while maintaining a constant noise value will increase S/N. When considering trace analysis and detection limits, peak height is a more important parameter than peak area, so anything that can be done to increase the peak height generally is beneficial. 

Wavelength selection: For UV detection, operating at the maximum absorbance for each peak will maximize the signal. Many methods are setup with a wavelength chosen as a compromise for the response of all sample components, so one or more components can give stronger signals if another wavelength is chosen. Nearly all modern detectors are under the control of software, so it is easy to change the detection wavelength during a run so that the optimum wavelength is used for each peak. All organic compounds have a strongly increasing absorbance as the wavelength is reduced below 220 nm, so lower wavelengths often give larger signals than higher wavelengths. However, background noise and response of interfering peaks is likely to increase at low wavelength too, so it is a good idea to evaluate the overall gain in S/N when a change in wavelength is made.

Better detector or modify analyte: The UV detector is the most popular detector for LC work, but several other detectors are available that offer increased selectivity. For example, a fluorescence detector can produce a large signal increase and a reduction in the background noise for samples that fluoresce or that can be derivatized to fluoresce. Electrochemical (amperometric) detectors and mass spectroscopic detectors are other examples of detectors that can give huge increases in signal for some compounds without the same increase in noise.

Inject more sample: In many methods, sample availability is not a limiting factor, so a larger mass of sample can be injected to increase the analyte signal. One can take advantage of on-column concentration for both isocratic and gradient methods by using an injection solvent that is weaker than the mobile phase to enable injection of much larger sample volumes than are used normally. If you increase the sample mass on column, remember to check to be sure that large peaks are not overloaded.

Reduce the peak width: For a given peak area, narrower peaks will be taller, thus improving detection limits. Volumetric peak widths can be reduced in isocratic methods by reducing the retention time (use of stronger mobile phase). With gradient methods, steeper gradients will reduce peak widths. Other tricks are to use smaller-particle columns, shorter columns, and narrower internal diameter columns, either alone or in combination. If you go this route, the detector time constant and data system collection rate might need to be modified (see the previous discussion on reduction of noise).

Improvement of detection and quantification limits for LC methods is a result of taking advantage of all the tools you have at your disposal. Increasing the analyte signal size, reducing the baseline noise, or a combination of both will be required to get lower method limits. 

"LC Troubleshooting" Editor John W. Dolan is Vice-President of BASi Northwest Laboratory of McMinnville, Oregon; a Principal Instructor for LC Resources, Walnut Creek, California; and a member of LCGC's editorial advisory board. Direct correspondence about this column to "LC Troubleshooting," LCGC, Woodbridge Corporate Plaza, 485 Route 1 South, Building F, First Floor, Iselin, NJ 08830, e-mail 

LC Troubleshooting

The minimum peak size for reportable results for a liquid chromatographic (LC) method depends upon the application and the signal-to-noise ratio (S/N).

The Role of the Signal-to-Noise Ratio in Precision and Accuracy

Evolution in science and technology is usually gradual and major inventions and achievements have their forerunners — observations or techniques that seem to be close to the concept of the later invention. This is also true about chromatography. Today, it is without any dispute that Mikhail Tswett was the true inventor of this method (1). However, before his groundbreaking activities from 1903 to 1906, there were others who, intentionally or unintentionally, carried out some separation in a moving flow. An example is Runge, who, about one-half a century before Tswett and a century before the invention of paper chromatography by Martin's group, presented dynamic separation on filter paper, although not for the purpose of separation but to illustrate reactions (2).  

Classical chromatography has been based upon adsorption, and adsorption processes have been used in laboratory as well as in manufacturing before the introduction of chromatography. Some of these processes were carried out in a moving stream and their aim was usually the removal of some undesired compounds, in other words, some kind of fractionation or separation. Such processes also have been introduced in petroleum processing: one example is the production of Vaseline from crude oil. However, we do not conceive these manipulations as chromatography. 

Toward the end of the 19th century, the rapidly evolving petroleum production also initiated speculations concerning the origin of petroleum, and attempts have been made to explain the reason for the obvious difference in the characteristics of crude oils occurring at different locations. One prominent American chemist involved in such work was David T. Day, whose theory (the "filtration hypothesis") was based upon the assumption of migration of the oil in the earth through various strata and selective retardation of certain compounds or compound groups during this migration. To prove the validity of his assumption, Day carried out some model experiments and demonstrated that indeed, some fractionation can be achieved in this way (whether it had anything to do with the origin of petroleum is another matter.) His experiments were repeated in Germany by Carl Engler and their publications also inspired some activities by other petroleum chemists. The technique of Day and Engler somewhat resembled chromatography, however, it was a kind of dead-end street, and there was no cross-influence between their work and Tswett's. Still, we can consider their approach as a forerunner of chromatography. Their activities are the subject of our present discussion. 

David Talbot Day (1859–1925), a graduate of The Johns Hopkins University, Baltimore, Maryland, was associated from 1885  on with the U.S. Geological Survey, Washington, DC. First, he pioneered in systematizing information on the mineral resources of the U.S. Then, in the 1890s, he turned his attention to various aspects of the newly emerging petroleum industry. Finally, in 1920 he established a private laboratory in California, investigating the possibilities of recovering oil from shales. 

By the last decade of the 19th century, oil production in Pennsylvania and Ohio was well established. Although chemical investigation of petroleum was still in its infancy, relying mainly on measuring its physical characteristics, it was noted that the characteristics of crude oils from different locations are different: one is heavier and contains more higher-boiling components than the other. Various explanations have been suggested for this difference, among them the so-called filtration hypothesis of Day, first described in a paper presented in 1897 [3]) He assumed that originally a primary oil existed at a certain location; this oil then migrated through various rock formations with help of diffusion. The various strata retarded more or less of the heavier fractions and, thus, the oils which passed through them and then accumulated at their final locations had different compositions. 

In 1900, a famous international exposition (forerunner of present-day world's fairs) was held in Paris, France, and there, the U.S. had a representative pavilion with a special display of the American petroleum industry: Day was in charge of this exhibit. In conjunction with the exposition, the First World Petroleum Congress was held in August and there, Day presented a paper on his hypothesis. Its title was "The variation of the characteristics of crude oils from Pennsylvania and Ohio" (4). After describing the characteristics of different samples, he mentioned that he also carried out some "filtration experiments" by conducting crude oil through fuller's earth, resulting in some fractionation: the lighter fractions moved faster, before the heavier fractions, and this process imitates the movement of oil in the earth. However, he apparently did not give any actual data, nor did he describe the experimental setup (at least it is not included in the published text of his paper). After referring to a "German scientist" (most likely to Engler, but no name was given) who stated that it would be important if somebody could develop a method enabling the separation of individual hydrocarbons present in the crude oil, Day made the often quoted (actually misquoted) statement that "the filtration method offers good hope . . . [and] I believe that before the next winter, I will be able to accomplish complete separation." (Translation of the text of his lecture published in French.) 

We do not know whether Day's statement reflected his optimism or an anticipation of future results. However, the realization of this hope definitely was beyond his capabilities and it took decades until his expectation could finally be accomplished by a different generation of petroleum chemists. 

In the years following the Paris lecture, Day carried out some additional investigations but did not publish them; we only know of a lecture presented in 1903 at a meeting of the Geological Society of Washington, DC (briefly reported by Mendenhall [5]). In the following decade, he had two more papers: one co-authored with Gilpin (6) and a more detailed paper in which he extended his theory to the oils of California, Louisiana, Mexico, Oklahoma, Texas, and Wyoming, but again, without actual data (7). Finally, in 1922, Day published a handbook for the petroleum industry (8) and in it, he briefly discussed adsorption processes; as an illustration, he presented a table summarizing the results of Gilpin and Cram (9). Otherwise, however, he did not deal anymore with such investigations. 

As mentioned earlier, Day was a graduate of The Johns Hopkins University and apparently he maintained some contact with his former alma mater, serving as an advisor in some research projects carried out by Professor Joseph E. Gilpin (1866–1924) and his graduate students. I know of three research papers by Gilpin's group published in 1908–1913 (9–11). These give a very detailed report of their investigations, with dozens of tables. However, they never achieved any real separation of individual components but measured only the physical constants of the individual fractions. Table I presents the results of one typical experiment (9). 

The papers of Gilpin and associates also described the system used in the investigations and we can assume that essentially it was the same as originally used by Day (Figure 1). Glass or tin tubes of 3–6 ft length and 1–1.25 in. diameter. The fraction character did not translate properly.]were filled with fuller's earth, and these were placed vertically in open dishes containing the petroleum samples. The sample ascended in the tube very slowly: the experiment reported in Table I took one full day but even longer times were not unusual. The experiments might have been accelerated by applying vacuum to the top of the tubes (see Figure 1). At the end of the experiment, the oil-laden filling of the tubes was slowly and carefully removed stepwise and cut to fractions, and the oil in the fuller's earth fractions was removed by water displacement. The oil recovered in this way represented only a fraction of the original sample: about 40% remained permanently adsorbed on the fuller's earth packing. (It should be noted that the so-called fuller's earth has served at that time as the most widely used adsorbent, both in the industry and laboratory work. It consists mainly of hydrated aluminum magnesium silicate and its name reflects one of the main industrial uses by textile workers — the so-called 

 — to remove grease and oils from cloth. In the U.S., the most important fuller's earth deposits were in Florida. Therefore, the material also was called Florida earth.) 

 Figure 1: Experimental setup used by Gilpin and associates (10). The crude oil samples were placed in tin cans of about 1-L volume (A). The glass or tin tubes filled with fullerÃÂ´s earth (B) were 3â6 ft long with 1â1.25 in. diameter. To accelerate the upward movement of the oil, sometimes vacuum (D) was applied to the top of the tubes through a manifold (F) and a large buffer volume (C); the manifold was connected to the tubes by small rubber tubes fitted with pinchcocks (E).			

We can conclude that Day's prediction in his Paris lecture never materialized, and neither he nor Gilpin ever separated individual hydrocarbons. In fact, the actual aim of their investigations — and this is clear when reading their papers — was only to prove the validity of the "filtration hypothesis," that some fractionation could have happened during passage of crude oil from one place to the other. 

Carl Engler (1842–1925) graduated from the Technical College (later Technical University) of Karlsruhe, Germany, and from 1876 until his retirement in 1919, he was associated with this school as a professor of chemical technology. He was instrumental in developing this university as one of the most important European technical schools of higher learning. For four decades Engler was the most important German scientist involved in petroleum research, with significant contributions to both technology and testing: even today we speak about the Engler viscosimeter and Engler distillation, and his five-volume petroleum handbook has served for decades, even after his death, as the bible of chemists, engineers, and geologists. 				

Table I:  Fractionation of a crude oil sample from Venango County, Pennsylvania, according to Gilpin and Cram (9)			

Engler apparently was present in Paris at Day's lecture and was very interested in his "filtration hypothesis." Therefore, after returning home, he started systematic investigations to study the process and to see whether adsorbents other than fuller's earth would also give similar results. Because Day gave no description of his experimental setup, Engler had to devise his own (Figure 2a). In Engler's system, the upward travel of the oil was facilitated by gravimetric pressure; also, he let the whole oil pass through the tube and collected portions of the effluent emerging from the top of the tube. He also devised another setup (Figure 2b) in which the fractions could be sampled at various heights of the tube filled with the adsorbent. In this way, the distribution of the individual fractions along the tube could be checked. 

Figure 2: Experimental setup used by Engler and Albrecht (12). The oil sample was flowing from a reservoir (A) by gravimetric pressure through a glass connecting tube (B) to the bottom of the tube filled with fullerÃÂ´s earth (C). The tubes were 80â90 cm (2).			

Engler, together with E. Albrecht, published his results in 1901 (12). Similar to Day (and later Gilpin), he recorded only the changes in the physical characteristics of the individual fractions, without any attempt at further separation of individual compounds. Table II presents typical results of Engler and Albrecht (12). In addition to fuller's earth, they also tried other adsorbents ("various sands") but the results were essentially the same. They also carried out experiments with other samples — for example, a 1:1 (v/v) ethanol–aniline mixture — but again, only the specific weights of the individual fractions were reported. 

Engler also considered the theoretical background of the (partial) fractionation occurring on fuller's earth. Apparently influenced by the then fashionable "capillary analysis" of Goppelsroeder (see, for example, the discussion in reference 13), he attributed the partial separation of the lighter fractions to capillary action. It took almost a decade until Ubbelohde  clarified that fractionation on fuller's earth is due to selective adsorption and has nothing to do with capillary action (14). 

The system of Engler and Albrecht clearly was more advanced than the one used by Day and Gilpin: it was faster and collection of the effluent was more convenient than the tedious work of dividing the fuller's earth packing into fragments and recovering the oil from these. Because the activities of Gilpin and his graduate students came seven years after Engler's publication, one might ask the obvious question, why did Gilpin not adapt Engler's system for his own investigations? Most likely the reason for this was that Gilpin's primary aim was (just as that of Day) to investigate changes occurring during passage of oil 

 (modeled by the fuller's earth) through various geological formations, and not obtaining individual compounds. Another question one might ask is: because Gilpin's publications followed Tswett's, why did he not cite him? Most likely Gilpin was not aware of Tswett's work, but even if he had known about it, he would not consider any relationship to his investigations. 

Table II:  Fractionation of American crude oil sample according to Engler and Albrecht (12)			

We do not know of any further work of Engler along this line, however, a few petroleum chemists reported on some follow-up investigations. We should briefly mention three here. 

Leo Ubbelohde (1876–1964) studied at the University of Berlin, Germany, and from 1907 on (from 1910 on as a professor) was associated with the Technical University of Karlsruhe, Germany. In 1933, he moved to the Technical University of Berlin–Charlottenburg. He had been very active in associations in the field of petroleum research and production. From 1907 to 1914, he served as the Secretary General of the International Petroleum Commission and in 1933 he established the German Society for Mineral Oil Research. He was also active in the fields of textiles and natural fats. We are interested here in one of his papers from 1909 criticizing a paper of Rakusin. In this paper, he stated among others, that the theoretical basis of the "filtration experiments" of Day and Engler had nothing to do with capillary action but was based upon selective adsorption (14). 

Commercial exploitation of the oil deposits along the Caspian Sea, near Baku (in present-day Azerbaijan) began in 1872 and by the beginning of the 20th century it represented the largest oil field in the world. (Today, it is practically unknown that these fields were developed by brothers of Alfred Nobel and the income from oil production significantly contributed to the immense wealth of the family.) A number of important Russian chemists have been associated with the petroleum industry and they also carried out intensive research. However, although they published in well-known Russian and foreign scientific journals, a historical evaluation of their activities is still missing. Essentially, we know about them only from the review papers of Herbert Weil (15–17) and the discussion of Camin and Raymond (20). 

With respect to our subject, the work of two Russian scientists should be mentioned: they are M.A. Rakusin and V.E. Herr. Both followed the investigations of Day and Engler, studying the fractionation of crude oil by "filtration" through fuller's earth. M.A. Rakusin (18) also believed that capillary action served as the basis of this process, and Ubbelohde's quoted paper (14) is a polemic discussion of this incorrect interpretation (and some other statements of Rakusin). The other Russian chemist is V.E. Herr who, in 1909, published a paper in the German petroleum journal on "contribution to the filtration of crude oils from Baku through fuller's earth" (19). In this paper, he concentrated on the possibility to eliminate the high-molecular-weight aromatic fraction of crude oil by selective adsorption. It is of interest to quote his statement that his investigations clearly showed that "fractionation in the usual sense (i.e., separation of individual components) does not take place." 

In 1927, the American Petroleum Institute started its Project No. 6, with the aim of separation and identification of the individual chemical constituents of petroleum and petroleum products, a project lasting for 40 years (13,20). In this project, fractionation on silica gel was first used, followed by increasingly sophisticated methods and the work of Day, Gilpin, and others achieving only limited fractionation was slowly forgotten. After some decades, László Zechmeister, one of the pioneers of (classical) liquid chromatography, finally brought back their work from oblivion. In a lecture on the "History, Scope and Methods of Chromatography" presented at the Chromatography Conference of the New York Academy of Sciences held in November, 1946, he discussed among others the origin of chromatography and the activities of Tswett as the inventor of the technique. He also mentioned the investigations of Day and Gilpin and correctly stated that they should be considered as forerunners of chromatography (21). Zechmeister repeated this discussion in the introduction of his book published in 1950 (22).  

Quite unexpectedly, just a few months after the publication of this book, a polemic article was published in 

 by Herbert Weil and Trevor I. Williams (23), criticizing Zechmeister that he was unjust to Day by naming Tswett as the inventor of chromatography. According to them, this distinction belongs to Day. Soon after this brief article, the first of Weil's series on the history of "industrial petroleum chromatography" was also published (15) and in it, he further elaborated on this question. In these papers, Weil and Williams falsely interpreted Day's 1900 lecture in Paris, claiming that in it, some form of chromatography had been described in detail. Therefore, they stated that August 20, 1900 (the day of Day's lecture in Paris) should be considered as the "birthday of chromatography." 

Zechmeister politely tried to answer Weil and Williams, emphasizing that instead of diminishing Day's merits, he actually resurrected it from oblivion (24). However, Weil and Williams continued this polemics, repeating their claim that Day "did in fact discover and clearly describe a system of adsorption chromatographic analysis" (25). Of course, we have seen in the previous discussion that Day did 

 describe his system (this was done only eight years later by Gilpin) and that neither he nor his followers attempted to separate individual hydrocarbons. The papers of Weil and Williams clearly represented a misquotation, or misinterpretation, of Day's lecture. 

Very wisely, Zechmeister stopped this polemic and did not continue to argue with Weil and Williams. Today, the role of Tswett as the true inventor of chromatography is universally recognized; at the same time, however, we should also recognize Day and Engler as 

, first pointing out the possibility of fractionation of petroleum by selective adsorption. Their work, using Zechmeister's words, "might have developed into systematic chromatography" (21). 

The polemics between Zechmeister and Weil and Williams seemingly ended with the polite answer of Zechmeister. However, this is only true about western literature: Soon, the priority of the invention of chromatography was blown up to a political question, proof or denial of the achievements of the Soviet Union. In the January 2006 issue of our magazine, we shall deal with this second controversy which could be characterized as a milestone in human stupidity. 

In the past, I have published two detailed discussions of the activities of Day, Gilpin, Engler, and associates: as part of a book chapter (13) and as an article in 

 (26). The figures and tables included in the present article are from the latter. I would like to express my appreciation to the publisher of 

From 1988 to 2004, "Milestones in Chromatography" editor Leslie S. Ettre was associated with the Chemical Engineering Department of Yale University (New Haven, Connecticut), first as an adjunct professor and then as a research fellow. Previously, he had been with the Perkin-Elmer Corporation for 30 years. He is currently a member of LCGC's editorial advisory board.

(4) D.T. Day, "La Variation des Caractères des Huils Brutes de Pennsylvania et de l'Ohio." 

Proc. Cong. Int. du Pétrole, Notes, Memoires et Documents. Journal de Pétrole

, D.T. Day, Ed. (Wiley, New York, 1922). 

Milestones In Chromatography

This column deals with a few chemists active during the early growth of the petroleum industry who carried out chromatography-like investigations even before Tswett...

Early Petroleum Chromatographers

Since the beginning of modern high performance liquid chromatography (HPLC) in the late 1960s, users have required continually new and improved columns to tackle more difficult separation problems or to improve their overall productivity and sample throughput. Column researchers and manufacturers have responded to these needs with the development of more efficient and more reliable packing materials. One of the areas in which improvements have been made is in particle size reduction. Figure 1 shows a series of 

 curves that I developed in the early 1970s that showed the influence of the particle size of silica gel on column efficiency (1). Known by theoreticians for years, this data systematically showed that the use of smaller size particles resulted in more efficient columns. At that time, all we had to use were irregularly shaped particles; spherical microparticulate silicas had not yet been developed. To make an easier comparison, Figure 2 provides a plot of this data for 

 at 1.0 cm/s linear velocity versus average particle size of the silica packing. This log–log plot suggested that even smaller irregular particles would result in further performance improvements but at the time, smaller particles were not available in narrow particle size distributions. 

Figure 3 gives a rough chronological order of the introduction of commercial column packings since the beginning of HPLC. The first high-performance packings were the pellicular (porous-layer bead) ion-exchange packings developed by Horvath and coworkers (2). These particles were used for the separation of nucleotides and resulted in the first commercial high-performance packing Pellosil (Northgate Laboratories, Hartford, Connecticut). Pellicular packings were rather large compared to today's particles — 40–50 μm. They had a thin porous coating that allowed rapid solute mass transfer into and out of the packing, resulting in improved chromatographic efficiency relative to the large porous particles that were generally used for liquid chromatography (LC) separations at the time. However, these pellicular packings had a big disadvantage — low surface area and, thus, very low sample capacity. For more details on this historical perspective, consult reference 3. 

At the time, column researchers knew that small porous particles (less than 20 μm) would provide even better efficiency and maintain the high capacity of the earlier porous packings. Some earlier work by Piel (4) in 1966 and Bidlingmeyer and Rogers (5) in 1969 showed promise, but the particles that they used were commercially available Cabosil (Cabot, Billerica, Massachusetts) fumed-silica packings that were sub-1.0-μm sizes, were fairly inert, were difficult to handle, and required extremely high pressures to operate. In short, these particles were unsuitable for the current needs at the time. Small porous particles in the range of 10 μm were not available in narrow particle-size distributions and in commercial quantities and packing procedures for micrometer-size particles into usable columns were not available. This all ended when Merck (Darmstadt, Germany) produced 5–10 μm narrow cuts of their thin-layer chromatography (TLC) grade silica gel and made them commercially available, and high-pressure slurry techniques were developed to reproducibly pack them (6). The first microparticulate column, MicroPak Si-10 silica gel, was introduced in 1972 by Varian Associates (Walnut Creek, California). 

Figure 1: Van Deemter plot for silica gel packings of decreasing average particle diameters (1). Columns: Merck LiChrosorb Si-60 silica gels of various diameters; Corasil II was included as an example of a silica pellicular packing (Waters, Milford, Massachusetts) mobile phase: 90:9.9: 0.125 (v/v/v) hexaneâmethylene chlorideâisopropanol. Test solute: N,N-diethyl-p-aminoazobenzene. (Reprinted with permission of Preston Publications, Niles, Illinois.)			

As a natural progression, as smaller particles were developed for HPLC, 5 μm became the standard particle diameter in the late 1980s. Later, high-performance 3-μm particles were reintroduced in the early 1990s after some initial performance problems of 3-μm commercial columns in the 1980s. Most recently, the sub-2-μm barrier was broken with the introduction of the Zorbax Rapid Resolution HT columns (Agilent Technologies, Palo Alto, California) in 2003. As depicted in Table I, at Pittcon 2005, column suppliers showed a number of high-performance columns packed with sub-2-μm particles (7). 

Figure 2: Dependence of efficiency on particle size at constant linear velocity. Data were extracted from Figure 1 at v = 1.0 mL/min. D is defined as H at v = 1.0 cm/s (1). (Reprinted with permission of Preston Publication, Niles, Illinois.)			

Interestingly, along the way of particle size reduction, the pellicular concept reappeared first with the development of the nonporous silicas (8) and then with the poroshell silicas (9,10). The nonporous silicas (and also nonporous resins) were around 1.5 μm in average particle diameter and were best for the separation of macromolecules such as proteins. They provided rapid separations but had very low sample capacity and high-pressure drops so columns had to be rather short. The Poroshell silicas (Agilent Technologies) were of 5-μm diameter, so that pressures were reduced greatly and they had higher sample capacity than the nonporous silicas and resins. Poroshell columns provide fast separations of proteins, which diffuse very slowly and, thus, show poor mass transfer characteristics relative to small molecules. Both types of packings could be derivatized with various silane functionalities to perform reversed-phase, ion-exchange, and affinity separations. 

Table I:  Commercial Sub-2 Î¼m HPLC Columns			

The purpose of this article is to focus on the smallest particle columns that appear to be directed to ultrafast separations using very short columns and to longer columns directed more for high-resolution separations of complex multicomponent samples. 

Why Do Small Particles Give Better Chromatographic Performance?

The goal of LC is to separate as many molecules as possible in the shortest possible time using a high-efficiency column packed with small particles that interact with the molecules by various chemical forces. Ideally, one would like to inject a multicomponent sample as an extremely narrow band that would then be separated into discrete narrow bands of the individual molecules at the end of the column. Working against this process is band spreading, which occurs during the transit of the molecules down the column length. Starting with the actual injection itself, band spreading occurs in areas in which there is no packing to interact with the sample molecules. One source of band spreading occurs outside of the column itself and is referred to as extracolumn effects. This extracolumn band broadening occurs in the sample loop, ports of the injector body, connecting tubing and fittings, column frits, column end fittings, as well as the detector flow cell. Another source of band broadening associated with the mobile phase occurs within the chromatography column between particles and in the pores of the particles. Finally, mass transfer occurs as the solute molecules transfer from the stagnant mobile phasewithin the pore to the stationary phase and back out again. 

Without getting too deeply into chromatographic theory, the separation efficiency 

 (or height equivalent to theoretical plate [HETP]) in micrometers as a function of mobile phase velocity is described by the van Deemter equation, shown simplistically in Equation 1. 

where A, B, and C are constants and μ is the mobile phase linear velocity (proportional to flow rate), measured in centimeters per second. The A term is a measure of the packing efficiency and is a function of packing efficiency and particle size. The B term is a function of longitudinal diffusion, or diffusion in the mobile phase, and the C term is a function of the mass transfer between the stationary and mobile phase as well as within the mobile phase. Within the C term, there is also a proportional dependency of the particle diameter squared. 

Figure 3: History of HPLC particle development.			

Figure 4 shows a diagram of the additivity of the three terms in the van Deemter equation. Note that the B term is dominant at low flow velocities, while the C term is dominant at high flow velocities. The minimum of the van Deemter curve represents the ideal flow velocity where maximum column efficiency is obtained. It is a compromise between the B and C terms. Figure 4 is an idealized representation of the curves shown in Figure 1. The A and C van Deemter terms are influenced by the particle size. Smaller particles tend to reduce the value of 

, which means that the column is more efficient — that is, it provides more theoretical plates per unit length. Small particles tend to allow solutes to transfer into and out of the particle more quickly because their diffusion path lengths are shorter. Thus, the solute is eluted as a narrow peak because it spends less time in the stationary and stagnant mobile phase where band broadening occurs. 

Figure 4: Diagram showing the additivity of the A, B, and C terms of the van Deemter equation for an HPLC column.			

One advantage of using smaller particles is that the column can be shortened and the plate count remains the same or nearly so. A shorter column means a faster separation can be achieved because separation time is proportional to column length. A shorter column run at the same linear velocity as a longer column also uses less solvent. 

Another fallout of the decrease in particle size is that the van Deemter curves tend to flatten out at higher linear velocities and the minimum shifts toward the right. Figure 5 shows a series of van Deemter curves for 5-, 3.5-, and 1.8-μm bonded spherical silica columns. One can easily see that the column packed with 1.8-μm particles gives a flatter curve at high linear velocity than the 5-μm column. Thus, one can run faster flow rates (linear velocities) and peaks maintain their efficiency yet the separation time decreases proportional to the increase in flow rate. 

Figure 5: HETP versus linear velocity for small particle spherical bonded silica columns (19). Columns: 50 mm X 4.6 mm Zorbax Eclipse XDB-C18 (30 mm, 1.8 Î¼m); mobile phase: 85:15 (v/v) acetonitrileâwater; flow rates: 0.05â5.0 mL/min; temperature: 20 Â°C; sample: 1.0 Î¼L octanophenone in eluent.			

Are There Any Downsides to Reducing the Particle Size?

There are a number of experimental parameters one should be aware of when reducing the particle size. One is the column pressure. Equation 2 shows the dependence of column head pressure on a number of experimental parameters including the particle size. Note that the pressure is inversely proportional to the square of the particle size.  

So when the particle size is halved, the pressure goes up by a factor of four. However, often for fast and ultrafast separations, the column length is also reduced so the pressure increase is not nearly as high as one would expect because pressure is proportional to column length. Of course, if longer lengths of columns, say 100 or 150 mm, are required to achieve higher plate counts, then higher pressure pumps might be required. Currently, there are commercial HPLC systems with upper pressure limits as high as 2 X 10

  psig. It should be noted that the total pressure that the HPLC system experiences is the sum of the column backpressure and the instrument backpressure. The latter results when small internal diameter capillaries are used in the flow paths to reduce extra column effects and the gradient delay volume. As the flow rate increases, the back pressure due to these capillaries increases proportionally. 

Another experimental parameter that bears watching when reducing the column length and especially when reducing the internal diameter is the extracolumn band broadening. Some of the modern ultrafast LC columns are only 15-mm long with an internal diameter of 2.1 mm. Such a column has a total void volume of around 33 μL. Many conventional HPLC instruments were developed for typical 150 and 250 mm X 4.6 mm analytical columns, where the total column void volumes are 1.6 and 2.6 mL, respectively. Peaks on the 15 mm X 2.1 mm column when packed with 1.8-μm particles are often only a few microliters wide (for low-

 peaks), which implies that extracolumn band broadening must be minimized if the true advantages of these small columns are to be realized. Some modern liquid chromatographs have been designed or modified to minimize the extra column effects. Upgrade kits are available to modify some older HPLC instruments to work satisfactorily with sub-2-μm columns. Extracolumn band broadening and similar effects are more thoroughly discussed in reference 11. 

Figure 6: USP assay for loratadine tablets on three columns with different particle sizes (18). Instrument: Agilent Series 1100 with thermally controlled column compartment; columns (USP L7): (a) 150 mm X 4.6 mm, 5-Î¼m Zorbax XDB-C8; (b) 100 mm X 4.6 mm, 3.5-Î¼m Zorbax XDB-C8 Rapid Resolution; (c) 50 mm X 4.6 mm, 1.8-Î¼m Zorbax XDB-C8 Rapid Resolution HT; mobile phase: 36.8:31.6:31.6 (v/v/v) 0.01 M dibasic potassium phosphate (pH 7.2)âmethanolâacetonitrile; flow rate: 1.0 mL/min; temperature: 30 Â°C; sample: single loratadine tablet, disintegrated and prepared as described in USP; sample size: (a) 15 Î¼L; (b) 10 Î¼L; (c) 5 Î¼L.			

Some other experimental parameters that must be taken into account when running fast and ultrafast separations are as follows: 					

Detector time constant (peaks can be only a second or two wide for short, narrow-bore columns run at high flow rates; and a time constant that is too slow will make the peaks artificially broad because the detector cannot keep up with the rapidly changing signal)

Data sampling (acquisition) rate (data system needs enough data points, usually 10–20, across a peak to define the peak to integrate area, determine retention time, and so forth)

Autosampler cycle time (for separations less than a minute or two, throughput can be slowed down by a slow autosampler)

Gradient delay volume (if the volume from where the gradient forms to the column head is too large, fast separations will be compromised because gradient has not reached column in time).

The terms "fast" and "ultrafast" are relative terms. If your separation was already 25 min on your present column and was reduced to 7 min, you would consider the latter time fast. If you were faced with hundreds of samples each requiring 7 min of separation time including gradient regeneration, then performing the same tasks in 2 min would be a big improvement and perhaps enable you to complete your task in a day or so instead of a week. 				

Figure 7: Ultrafast separation of alkylphenones (20). Instrument: Agilent Series 1100 with 80-Hz diode-array detector; column: 30 mm X 4.6 mm, 1.8-Î¼m Zorbax SB-C18; flow rate: 5 mL/min (18 mm/s); gradient: 50â100% acetonitrile in 0.3 min; temperature: 50 Â°C; injection volume: 3 Î¼L; flow cell volume: 5 Î¼L; sample: alkylphenone test mix, 100 ng/Î¼L in acetonitrile. Peaks: 1 = acetanilide, 2 = acetophenone, 3 = propiophenone, 4 = butyrophenone, 5 = benzophenone, 6 = valerophenone, 7 = hexanophenone, 8 = heptanophenone, 9 = octanophenone.			

Basically, fast or ultrafast LC relies on the use of small particles packed into short columns run at high flow rates. Often, one can accomplish the same separation that can be accomplished on a longer column with a larger particle size but in a fraction of the time. An example of a 

) method that has been "downsized" is shown in Figure 6, where an antihistamine was chromatographed on three different columns packed with 5-, 3.5, and 1.8-μm reversed-phase C18 bonded material with the same bonded stationary phase. Note that 150 mm X 4.6 mm columns packed with 5-μm particles are still the standard in most HPLC laboratories. This isocratic 

 method required a total of 38 min (Figure 6a) on the 150-mm column. However, one can see that the same separation was achieved on a column (Figure 6c) that was only 50 mm in length but in a third of the time — in less than 13 min. Even switching to a 100-mm column reduced the total time to just over 23 min (Figure 6b). Note that the excipients were separated on all three columns so nothing is lost in the "downsizing" experiments. The gain is separation speed. 

Ultrafast separations generally refer to separations achieved in a minute or two for relatively simple samples. Figure 7 shows the 24-s separation of nine alkylphenones on a short (30 mm) column packed with 1.8-μm packing run at 5-mL/min. A ballistic gradient was used for the rapid elution of these nonpolar analytes. Because the average peak width was only 0.3 s, a diode-array detector with a fast sampling rate (80 Hz) was required. Otherwise, the fast peaks would have been distorted because the time constant would have been too slow to adequately detect them. 

Figure 8: Fast protein separation. Column: 10 mm X 4.6 mm, 1.5-Î¼m ProSphere HP ZAP! C18; mobile phase A: water with 0.1% trifluoroacetic acid; mobile phase B: acetonitrile with 0.1% trifluoroacetic acid; gradient: 25â75% B in 4 min; flow rate: 1.0 mL/min; detection: UV absorbance at 280 nm. Peaks: 1 = cytochrome C, 2 = lysozyme, 3 = b-lactoglobulin, 4 = ovalbumin. (Courtesy of Alltech/Grace.)			

Where Can I Obtain Sub-2-μm Columns to Experiment with?

Column Watch

Higher productivity and faster analyses are two of the driving forces for continued improvement in high performance liquid chromatography (HPLC) column technology.

Fast and Ultrafast HPLC on Sub-2-µm Porous Particles&#151;Where Do We Go From Here?

Baker, Timothy R.; Hoke II, Steven H.;  and Dobson, Roy L.M. Evaluating Analytical   Instrumentation Using the Golf Score Card   Tool. September, p. 1024. 

Balogh, Michael P. MS — The Practical Art.    Column. 

Bartoszuk, Malgorzata A. See Zarzycki, Pawel K. 

Biselli, Scarlett; Hartig, Lutz; Wegner,    Heiner; and Hummert, Christian. Analysis of   Fusarium Toxins Using LC–MS-MS: Application to Various Food and Feed Matrices.    April, p. 404. 

Boos, Karl-Siegfried. See Georgi, Katrin. 

Burge, Larry J.; and Sorgen, Christina M.  Evaluation of Polar-Embedded and Polar-Endcapped C12–C18 Stationary Phase Vulnerabilities Toward Highly Nucleophilic    Compounds: A Case Study. August, p. 762. 

Dolan, John W. LC Troubleshooting. Monthly   column. 

Dolan, John W., and Snyder, Lloyd R. "Adjusting   Conditions for a Routine Reversed-Phase    HPLC Assay, Part II: Changing Separation   Conditions," in Column Watch. February,  p. 118. 

Esteve Romero, J.S. See Ruiz Ángel, M.J. 

Ettre, Leslie S. Milestones in Chromatography.   Column. 

 François, Kelly. See Vercammen, Joeri. 

Gamache, Paul H.; McCarthy, Ryan S.; Freeto,   Scott M.; Asa, Darwin J.; Woodcock,    Matthew J.; Laws, Katherine; and Cole,    Roderic O. HPLC Analysis of Nonvolatile    Analytes Using Charged Aerosol Detection.   February, p. 150. 

Georgi, Katrin, and Boos, Karl-Siegfried. Control   of Matrix Effects in Bioanalytical MS–MS    Using On-Line Multidimensional Solid-Phase   Extraction. April, p. 396. 

Henry, Richard A., and Bell, David S. Important   Guidelines for Optimizing Speed and Sensitivity in Small-Molecule LC–UV and    LC–MS. May, p. 496. 

Hinshaw, John V. GC Connections. Column. 

Hoke II, Steven H. See Baker, Timothy R. 

Hummert, Christian. See Biselli, Scarlett. 

Kulhanek, Kathrin M. See Zarzycki, Pawel K. 

Lamparczyk, Henryk. See Zarzycki, Pawel K. 

Mabic, Stéphane; Regnault, Cecilia; and Krol,    Jim. The Misunderstood Laboratory Solvent:   Reagent Water for HPLC. January, p. 74. 

Majors, Ronald E. Sample Prep Perspectives.    Column. 

McCalley, David V. Choice of Buffer for the    Analysis of Basic Peptides in Reversed-Phase   HPLC. February, p. 162. 

Morrison, Ryan D., and Dolan, John W. "Peak   Fronting, Column Life, and Column Conditioning," in LC Troubleshooting. June,  p. 566. 

Ng, Paul K., and Coan, Michael H. "Chromatography Applications in Drug Discovery   of Therapeutic Proteins," in Directions in    Discovery. March, p. 280. 

Parmely, Rick. "Successful Gas Chromatography   Using Fused-Silica Columns," in Column    Watch. November, p. 1166. 

Przybyciel, Matthew. "Fluorinated HPLC Phases   — Looking Beyond C18 for Reversed-Phase   HPLC," in Column Watch. June, p. 554. 

Editorial Index

2005 Editorial Index

Varian, Inc. (Palo Alto, California) recently announced that it has acquired PL International, Ltd. (Shropshire, United Kingdom), and its subsidiaries including Polymer Laboratories, Ltd. (Amherst, Massachusetts) for approximately $42 million in cash. The transaction includes an opportunity for up to $23 million of additional payments over three years, depending upon financial performance. 			

Polymer Laboratories develops of consumables and instrumentation for advanced polymer analysis, including columns, standards, and specialist chromatography systems dedicated to gel permeation chromatography analysis.  

 Laboratory Information Management Systems			

"The acquisition of Polymer Laboratories is strategic to the advancement of Varian's goal to grow the applications-based consumable products to become a larger portion of the business," said Garry W. Rogerson, president and CEO of Varian. 

The acquisition is part of the company's strategy to strengthen its product portfolio with internal research and development efforts and strategic acquisitions of complementary products and technologies. 

Thermo contributes to Hurricane Katrina Disaster Relief Fund

Thermo Electron (Waltham, Massachusetts) recently has announced that it will match employee contributions to the Red Cross in support of the Hurricane Katrina Relief Fund. During the days following the disaster, the company's employees contributed money, food, water, medicines, and other relief to help the evacuees and those remaining in the affected areas. 

Varian acquires Polymer Labs, Thermo contributes to Hurricane Katrina Disaster Relief Fund, and French Police select Whatman for DNA collection.

LCGC North America-12-01-2005

The Flame Ionization Detector

The Role of the Signal-to-Noise Ratio in Precision and Accuracy

Early Petroleum Chromatographers

Fast and Ultrafast HPLC on Sub-2-µm Porous ParticlesWhere Do We Go From Here?

2005 Editorial Index

Peaks of Interest

LCGC North America-12-01-2005

The Flame Ionization Detector

The Role of the Signal-to-Noise Ratio in Precision and Accuracy

Early Petroleum Chromatographers

Fast and Ultrafast HPLC on Sub-2-µm Porous Particles&#151;Where Do We Go From Here?

2005 Editorial Index

Peaks of Interest

Fast and Ultrafast HPLC on Sub-2-µm Porous ParticlesWhere Do We Go From Here?