The worldwide shortage of the popular solvent acetonitrile started in the early fall of 2008. Many laboratories were caught unaware of this pending shortage and did not address the problem quickly enough while others who sensed the problem stocked up on the available supply at the time. Unlike methanol and other solvents that are manufactured for industrial use, acetonitrile is available only as a byproduct of another chemical production. Surprisingly, even the solvent companies, who supply high-grade acetonitrile, experienced a shortage of feedstock for their purification and, hence, could not obtain sufficient acetonitrile to address their customer needs, and the scramble was on. Prices have skyrocketed. Since last October, internet blogs and on-line chromatography forums have been discussing and debating the issue, even referring to the shortage as a "conspiracy of solvent companies." Depending upon the laboratory and the country reporting, there was more than a sufficient amount or none at all. Many ideas to handle the "short-term" shortage were floated around, some of which were practical, some of which were off-the-wall. Many of the ideas were generated by commercial entities, some of whom slanted the possible solution to suit their business needs. In this article, I will present an unbiased look at some of the more practical aspects of coping with the acetonitrile shortage. Currently, as of this writing, there is no prediction when normal supplies will resume.  

Acetonitrile is a solvent that has favorable properties for many chemical applications. It shows minimal chemical reactivity, low acidity, and is reasonably low boiling. Important for reversed-phase high performance liquid chromatography (HPLC) and hydrophilic interaction liquid chromatography (HILIC) applications is its miscibility with water, wide range of achievable polarities with water mixtures, low viscosity (resulting in low pressure drop, even with water binary systems), and low ultraviolet cutoff (down to 192 nm). For HILIC, large volume percentages of acetonitrile in water are favored and thus, its use can be more affected by the acetonitrile shortage than reversed-phase chromatography. Particularly hard hit are those who do preparative- and process-scale LC purification using acetonitrile eluents. Acetonitrile is also a very popular solvent for sample preparation techniques such as dilution, protein precipitation of biological fluids, and as a good eluent for solid-phase extraction (SPE). Acetonitrile also finds use in solid-phase oligonucleotide–DNA synthesis, for peptide synthesis, and in the manufacturing of drug substances and drug products, sometimes in copious amounts. So there are many reasons why there is concern over limited supplies. 

When first reported, there were a couple of reasons why the acetonitrile shortage arose so quickly. One of the first reports was that during the 2008 Summer Olympics, Chinese factories where acetonitrile was produced as a synthetic by-product were shut down to minimize air pollution. One of the plants was reportedly the largest producer of acetonitrile in China. Some of the factories have not returned to production. Secondly, the September 2008 Hurricane Ike in the Texas Gulf Coast temporarily shutdown one of the major producers in Texas. This facility also had planned a shutdown in production for a factory update and expansion in January 2009. To add to the problem, this company is reported to be in financial difficulty. But the general consensus is that the worldwide demand for acrylonitrile has decreased severely due to the economic downturn.  

It turns out that acetonitrile is a minor by-product of the manufacturing of acrylonitrile using the Sohio process. About 2–4% of the Sohio process results in the formation of acetonitrile that is then removed and purified for further use. Most manufacturers of acrylonitrile incinerate this by-product as fuel rather than trying to reclaim it. So, there are a limited number of acrylonitrile producers who reclaim the acetonitrile.  

Acrylonitrile has been used widely in plastic products such as polyacrylonitrile and acrylonitrile-butadiene-styrene (ABS) resins. These products are used in automobile manufacturing (bumpers and other parts), carpeting, acrylic and carbon fibers, luggage, small appliances, telephones, computer housings, and other industries. Obviously, the reduced purchase of cars, the housing slump, and purchase of other consumer products has decreased the demand for acrylonitrile-based plastics and production has been curtailed. There also have been reports of a decrease in demand due to acrylic fibers losing market share to polyester fibers. With a large amount of pharmaceutical manufacturing moving to China and India, some of the acetonitrile production will stay in Asia, and that could be another factor in influencing the worldwide availability of acetonitrile down the road. 

Thus, because the amount of acetonitrile is proportional to the reduction in acrylonitrile production, there is currently a shortage of this widely used solvent, relative to a year ago. There are other synthetic routes to produce acetonitrile but many of them are not as economical as that obtained from the acrylonitrile process. However, there have been some reports that opportunistic entrepreneurs in China are considering starting production of acetonitrile by alternative methods.  

What Are The Consequences of The Acetonitrile Shortage?

Referred to as "The Great Acetonitrile Shortage" or "The Great Acetonitrile Drought" thereby creating "Acetonitrile Anxiety," obviously, when a commodity product falls in supply and the demand still exists, there are going to be cost consequences. The current prices for high-quality and HPLC-grade acetonitrile have gone through the roof with quotes of over $1000 for a 4 X 4-L case fairly commonplace. Price gouging seems to be in fashion. Prices have risen by factors of 6–8 since the summer of 2008. Distributors and suppliers are allocating their available supplies to favored customers based upon their past purchases. In some cases, laboratory chemical suppliers have not been selling this solvent at all, especially to new customers. In some parts of the world, supplies are not even available and deliveries as long as six months have been quoted. In other places, supplies seem to be recovering and prices moderately reduced. The general feeling is that acetonitrile prices will never return to 2008 levels.  

Besides price gouging, another consequence of the shortage is that a number of speculators are showing up on internet blogs offering large quantities of bulk acetonitrile of questionable or unknown purity. Most avoid such transactions that also could raise suspicions when importing large amounts of solvent from abroad.  

Some feel that the acetonitrile crisis is a "wake-up" call and that laboratories should now develop short-term and long-term strategies to come up with alternative solvent systems. The movement to "greener" solvent systems or miniaturized liquid-phase chromatographic systems could be a positive outcome.  

Those most concerned about the acetonitrile shortage are those who work in regulated laboratories, particularly in the pharmaceutical industry. The United States Food and Drug Administration has received many inquiries related to the shortage of acetonitrile, most of which revolve around substituting an alternative solvent to use to validated methods. Their comment on this issue can be found in reference (1). Basically, they have maintained their position that "regardless of the changes a firm makes to address the (acetonitrile) shortage, appropriate method validation and compliance with relevant good manufacturing practices (CGMPs) are necessary" (1). 

The most obvious consequence is that many laboratories are looking for a continued supply to maintain their current methods. In the absence of meeting their acetonitrile demands, they are looking for alternatives that can solve their analytical or synthetic problems with a minimum disruption. We will now explore approaches that can be used to eliminate or minimize the use of acetonitrile in their laboratory. We will not address those who use large amounts of acetonitrile in their manufacturing or synthesis processes. 

Methods to Reduce HPLC Use of Acetonitrile Based Upon Changing Column Dimensions or Particle Size

Shorter columns with the same internal diameter: 

 Depending upon the resolution of your important analytes of your current method, you might be able to reduce the length of your column without sacrificing the method resolution requirements. Because column length (

), when you reduce the column length and keep the flow rate constant, the amount of solvent required is proportionally reduced. Thus, for an isocratic separation, cutting the column in half, separation time and solvent required is halved. However, there is a proportional loss in theoretical plates (

), but because resolution is proportional to 

, one does not sacrifice as much resolution as you might think and the separation might be sufficient.  

However, for gradient elution, if one cuts the column in half, the gradient time must be reduced proportionally also. Otherwise, your method can be compromised. When changes are made in column length, column internal diameter, flow rate, the change from the initial composition to the final composition, and gradient time, conditions must be adjusted to satisfy equation 1. According to equation 1, which refers to the apparent 

*), a change in length must be accompanied by a proportional change in gradient time (

So, if the column length is decreased by a factor of two, then 

 must be reduced to ½ to obey equation 1. If, in addition, one would choose to speed up the separation by an increase in flow rate by a factor of two, then 

 would be further decreased by a factor of two. Such a twofold change in conditions can have a detrimental effect on resolution but it could be tested easily. 

 The use of smaller particles packed into shorter columns is the current trend in higher throughput LC. In this case, a reduction in length is accompanied by a proportional reduction in particle size to maintain separation efficiency and perhaps resolution as well. Because 

, there is a loss of plates in going to shorter columns but 

 is the average particle diameter) and therefore, plates are increased to maintain the overall resolution. Figure 1 provides the separation of some xanthines as an example of increasing speed by simultaneously reducing the column length from 25 cm to 5 cm and particle diameter from 5 μm to 1.8 μm while keeping the flow rate and the column diameter the same. The separation time (

) and the solvent consumption are reduced proportionally with each reduction in 

. For this example, the solvent consumption in Figure 1a was 12 mL, in Figure 1b, 4.5 mL, and in Figure 1c, a total of 2.3 mL. Here, the resolution remained more or less the same so there was no sacrifice in method performance but a fivefold solvent savings.  

There is one caveat: as one decreases the particle size, there is a slight shift in the minimum of the van Deemter curve so that flow rate might have to be increased slightly to take full advantage of the particle size reduction. Thus, the separation time could be further decreased. The new submicrometer sized particles are particularly attractive here because separation times and reduction in solvent can be up to an order of magnitude compared to longer columns packed with larger particles. I have updated my listing of suppliers of 2-μm and sub-2-μm columns for high-throughput LC (Table I). With a visit to the websites of these companies, one can get method translation equations, calculators, and software to perform the necessary calculations for changing experimental parameters.  

Table I: Commercial two and sub-2 Î¼m totally porous HPLC columns*			

The new superficially porous packings can provide some of the same advantages with the added benefit of lower pressure drops because column lengths are generally 5 cm and the particles are in the range of 2.7 μm, but the efficiencies noted are in the neighborhood of the sub-2-μm columns.  

Reduction of the column internal diameter:

 An easy approach to reducing solvent consumption is changing the internal diameter of the column with or without decreasing the column length. According to equation 1, if one reduces the diameter and desires to keep the same separation time, the flow rate must be reduced proportionally to the inverse of the radius ratios squared. Thus, if a column internal diameter is reduced from 4.6-mm to 3.0-mm i.d. at constant length, then the flow rate would be reduced by (3.0/4.6)

  to a factor of 0.43. Thus, if one originally had a method using 1.0 mL/min and switched to the 3.0-mm i.d. column, the flow rate would be reduced to 0.43 mL/min and there would be resultant solvent savings of 57%. With 3.0-mm columns, almost any modern HPLC instrument could be employed because instrumental extracolumn volumes and gradient delay volumes are sufficiently small enough not to cause method variances. Decreasing column diameter with a concurrent decrease in flow rate (to maintain constant linear velocity) allows the separation time to remain constant, and users should see a very similar chromatogram as the original.  

One can continue to reduce the column internal diameter and achieve further reductions in solvent usage. For example, changing from a 4.6-mm i.d. column to a 2.0-mm i.d. column (referred to as microbore or semimicro columns) will result in approximately 80% solvent savings without affecting the separation characteristics. Such a change can be seen in Figure 2, which shows a separation performed on a 4.6-mm column and a 2.1-mm column, both with a 150-mm length. The flow rate was reduced from 1.0 mL/min to 0.2 mL/min. Note that the sample size was reduced from 5 μL to 1 μL because the cross sectional area of the 2.0-mm column is 1/5. In the case of internal diameter reduction, the response increases fivefold if the sample size is injected. To keep the same response, the sample mass injected was decreased proportionally because a concentration-sensitive detector was used. If the same sample volume (at the same concentration) is injected onto the smaller internal diameter column, an increase in sensitivity should result (if no band spreading results from the larger injection volume), which makes microbore columns ideal for sample mass-limited situations.  

When one decreases the internal diameter of the column 

, the column void volume decreases with the inverse radius ratio squared and peak volumes become smaller. Table II shows the approximate peak volumes at three different 

 values for three different column internal diameters at constant length. Note that for the smallest internal diameter column, 2.1 mm, the peaks with low 

 values are only a few tens of microliters in volume. For conventional liquid chromatographs, optimized for 4.6-mm i.d. columns, when peak volumes become less than 60 μL, some optimization of extracolumn effects (those volumes outside of the column packing) might be required (see highlighted values in Table II). Some of the hydraulic volumes that might need deceasing are injection volume, tubing volumes from the injector to the column inlet and from the column exit to the detector inlet (0.12–0.13 mm i.d. tubing is recommended), and the detector flow cell volume itself might need to be reduced if peaks with small 

 values are to measured. Because the flow rate for 2.0- or 2.1-mm i.d. columns is now much smaller, the gradient delay volume (sometimes referred to as the dwell volume which is the volume from the point of mixing of the gradient solvents to the head of the column) must be decreased. Otherwise, the gradient might take a considerable time to reach the head of the column and extrapolations from the larger column might not be entirely comparable.  

Table II: Solvent saver columns can be used on most LC systems without modification (based on peak volume)			

As the column internal diameter becomes further reduced to 1.0 mm and below, the system volume must be totally reworked. For this reason, most commercial chromatographs desired for capillary columns have extremely low extracolumn effects and gradient delay volumes including smaller flow cells as standard. In addition, they are designed to work well at flow rates below 100 μL/min.  

To illustrate the effect of column internal diameter on solvent usage, peak volumes and flow rates, Table III provides some characteristics of such systems. Today, even nanoLC instruments that operate with extremely low flow rates (a few hundred nanoliters per minute) are available commercially but mostly are applied to proteomics, where samples are mass limited and mass spectrometry is primarily used for detection. These types of systems are the ultimate in solvent savings. With chip-based systems, a liter of acetonitrile could last a month or more!  

Table III: Reducing column internal diameter reduces mobile phase use			

Smaller internal diameter columns are often harder to pack as efficiently as larger diameter columns, so there can be some differences in plate number that might have some impact on extrapolations of separation data from larger internal diameter to very small internal diameter columns.  

In the case of gradient elution, equation 1 still holds true. With a change in both length and internal diameter, both the flow rate and the gradient time will need to be adjusted. Remember, the column volume can be determined by 

* constant with the new column and the same gradient conditions, substitute the values for new column dimensions in equation 1. 

Reduction of column internal diameter, column length, and particle diameter:

 Of course, one can combine these approaches to provide even more dramatic solvent savings. By such a combination, however, the column void volume can become quite small, necessitating optimization of the instrumental contributions as discussed previously. 

Methods to Reduce HPLC Use of Acetonitrile Based Upon Changing Solvent or Stationary Phase 

Use of other organic modifiers in reversed-phase HPLC:

 The most obvious change and one of the easiest to implement is to continue to use one's current reversed-phase column and change the organic modifier to a solvent more readily available. The most popular choice is methanol that can be obtained in high-purity suitable at a reasonable cost and is useful for most HPLC applications. Other solvents that have been considered are tetrahydrofuran, isopropanol, ethanol, and n-propanol. These solvents also are available in high purity for HPLC applications. Of these less commonly used organic modifiers, only tetrahydrofuran has some drawbacks, such as chemical stability — that is, peroxide formation for the unstabilized variety — or the presence of stabilizer, butylated hydroxyl toluene (BHT) that can interfere with UV detection and low UV transparency. However, tetrahydrofuran has some excellent solvation properties for polar analytes and is quite effective when blended with small amounts (~5–10%) of methanol. In addition, tetrahydrofuran can be blended to get approximately the same solvent strength (ε

) as acetonitrile. A mixture of 50% methanol, 25% tetrahydrofuran, and 25% acetonitrile works as a substitute for 100% acetonitrile 

Methanol has some disadvantages for HPLC use when compared to acetonitrile. Due to its higher viscosity, the column pressure drop will increase. The well-known viscosity versus % methanol in water curves show a maximum viscosity at approximately 45% methanol while acetonitrile–water mixtures show a much lower increase in viscosity as the % acetonitrile increases. So when gradients are run with water–methanol binary solvents, pressures can increase and can approach the maximum operating pressure of the LC system. When substituting methanol for acetonitrile and operating at the same experimental conditions, the higher viscosity of methanol will give lower efficiency and can lead to slightly broader peaks. Methanol shows a higher UV cutoff (~210 nm) than acetonitrile (~192 nm) that limits its usefulness in the low UV region. Probably the biggest factor that one might observe when switching from acetonitrile to methanol is the different analyte selectivities that can occur (see Figure 3, reference 3). The most glaring change was the elution order of these compounds when using acetonitrile as the organic modifier (Figure 3a) compared to methanol (Figure 3b) as a modifier. Not only did the elution order change but peaks 4 and 6 that were well resolved with the acetonitrile–water solvent system were unresolved when methanol–water was used as the mobile phase. Because methanol is a weaker solvent than acetonitrile, note that its percentage had to be increased to have similar retention times. In Figure 3b, one can see that the methanol concentration in the binary mixture was 58% compared to the 39% acetonitrile shown in Figure 3a. 

For binary mixtures of water and organic modifiers, a useful nomograph is shown in Figure 4 (4). This figure compares the approximate solvent strength of binary mixtures of water with the popular solvents acetonitrile, methanol, and tetrahydrofuran. Vertical lines in this figure intersect mobile phases having a similar solvent strength. For example, the vertical line at 40% tetrahydrofuran in water intersects the methanol line at 63% and the acetonitrile line at 53%. Thus, to get approximately equivalent retention times (or at least a similar solvent strength), these amounts of organic modifier would have to be blended with water. Thus, tetrahydrofuran is a more powerful modifier than acetonitrile that in turn is more powerful than methanol. If one closely compares the chromatograms of Figures 3a and 3b, it can be seen that the nomograph is only approximate. In addition, this nomograph does not predict selectivity changes, as was noted in Figure 3, nor are the nomographs suitable for ternary mixtures.  

A slightly different format was used in Figure 5. Here, a plot that correlates the composition of several water-miscible solvents that could be used to replace acetonitrile is shown (5). This graph adds ethanol, isopropanol, and dimethyl sulfoxide (DMSO) that could be used as modifiers. Both isopropanol and DMSO are quite viscous, so one must carefully observe the column pressure when switching to such solvents. Again, this plot should be used as a starting point in the evaluation of a modified method. 

Articles

Column Watch

Ron Majors provides an unbiased look at this present crisis and gives alternative solutions.

The Continuing Acetonitrile Shortage: How to Combat it or Live with It

This is the fourth "LC Troubleshooting" column in a series looking at different aspects of the calibration process for liquid chromatography (LC) methods. We first (1) considered whether or not to force a calibration curve through the origin (

 = 0). The next stop (2) was a discussion of some techniques to determine the limits of detection and quantification. Last month (3), we saw how %-error plots could help us visualize possible problems with calibration curves. The present discussion focuses on three different calibration models: external standardization, internal standardization, and the method of standard additions. Next month, we will look at the technique of curve weighting. 

The use of external standards is the simplest, and likely the most common method of calibration for quantitative LC methods. The technique simply compares the detector response between known concentrations of analyte with the response for samples containing unknown concentrations. A calibration curve (also called a standard curve or sometimes a "line") is generated by injecting a series of calibration standards. For well-behaved methods, as demonstrated by validation studies, and narrow ranges, for example, ±10% in concentration, a single-point calibration can be used. In this technique, the response (area) for a known concentration of reference standard is calculated (area/concentration) to generate a calibration factor. This value is divided into the area for an unknown concentration and the result is the concentration of the unknown. 

More commonly, calibrators are prepared that cover the expected sample concentration range, and the response of these calibration standards is used to generate a calibration curve. This is demonstrated with the data of Table I. The data of Table I simulate the use of a method for which all the calibrators are injected both at the beginning (data set 1) and end (data set 2) of a batch of samples. Calibrators were prepared at 1, 2, 5, 10, 20, 50, 100, 200, 500, and 1000 ng/mL and injected with the batch of samples. The results are combined in Table I. Using the data system software or spreadsheet software, such as Microsoft Excel, a calibration curve can be plotted, as is shown in Figure 1. Here, concentration (

-intercept, so the curve can be forced through zero (

). (See the discussion of reference 1 for more information on zero-intercept decisions.) This regression equation then is rearranged (

/403.7) to calculate the concentration of unknown samples. For example, a sample that generates a peak of 36,827 area counts (last line, Table I) would have a concentration of (36,827/403.7) = 91.2 ng/mL. A common sense double-check of the calculation shows that in Table I, 36,827 area counts would fall between 50 and 100 ng/mL, so the result seems reasonable. 

Figure 1: External standard calibration plot from data of  Table I.  

External standardization works well when sample preparation steps are simple and the injection volume precision is good. For example, if the sample comprises a pharmaceutical tablet that is weighed, dissolved in an aliquot of injection solvent, filtered, and injected with a modern autosampler, external standardization is appropriate. When many sample preparation steps are included in the method or there is a question about autosampler precision, an internal standard can improve the precision and accuracy of a method.  

With internal standardization, a second compound, often related to the analyte but never found in the sample, is added at a known concentration to every sample and calibrator. For example, one might pipette 20 μL of a 10 μg/mL solution of internal standard (IS) into 1-mL aliquots of sample and calibrator at the beginning of the sample preparation process. This would mean that each sample would now contain 200 ng/mL of internal standard. If the proper internal standard is chosen, it should track the sample through the sample preparation process, correcting for sample losses due to incomplete extraction, sample loss, or minor differences in reconstitution volume. It is the ratio of the analyte to internal standard that is the critical measurement in an internally standardized method. 

The calibration curve data are generated by injecting calibration samples of different concentration that all contain the same concentration of internal standard. The ratio of analyte area to internal standard area is calculated and plotted as the 

-value against the concentration of the calibrator. Table I shows the internal standard areas and analyte/IS ratios for the same data used for the external standard experiments. The calibration curve is plotted in Figure 2. The 

 (0.00011), so the curve is forced through zero. The same unknown used previously (last line of Table I) generates an analyte/IS ratio of 0.3645. The regression equation is rearranged to 

/0.004037), which allows calculation of the unknown concentration of (0.363551/0.004037) = 90.1 ng/mL. This is approximately the same value as that achieved by external standardization. 

Figure 2: Internal standard calibration plot from data of Table I.  

Now that we've see the two most common standardization methods for LC, which one should be used? In most cases, precedent will have been set already with similar methods, so the choice should be clear. In other cases, it might not be obvious which calibration type to choose. The simplest way to make the decision is to check the results empirically. Prepare calibrator samples that contain internal standards, as described previously, and analyze them. Process the data using both the external standard and internal standard technique. Next, "back-calculate" the calibration samples against the calibration curves and determine the %-error by which each point deviates from the regression line. Sometimes this is called calculating the residuals. For the data sets discussed previously, the errors for each curve type are shown in columns 7 and 8 of Table I, with the ratio of the errors in the right-hand column. The results can be compared in several ways. You can compare the results visually at each concentration or the ratio of the errors. Alternatively, you can calculate the average of the absolute values (2.74% for external standard and 1.22% for IS) or the sum of the absolute values (54.8% and 24.5%). In each case, you can see that the internal standard method reduces the error by about twofold. This supports the use of internal standardization for this method. This suggests that there is some kind of physical sample loss or inconsistent volumetric recovery in the sample preparation process that is compensated by the use of an internal standard. If the results were comparable or the external standard method gave smaller errors than the internal standard method, the simpler external standardization technique should be used. 

When either the external or internal standard calibration method is used, it is usually best to prepare matrix-based standards. This means that the calibration standard is prepared in a solution that represents the sample extract in all ways except the presence of the analyte. For pharmaceutical samples, the blank matrix might be a placebo of the formulation. For bioanalytical samples (for example, drugs in plasma), a drug-free plasma sample might be used. For a pesticide measurement in soil or water, pesticide-free soil or water might be used. By using a matrix-based standard, the likelihood of signal suppression or enhancement by the matrix is reduced. A blank matrix sample usually is run to confirm that there are no interfering peaks present in the matrix. 

Sometimes, however, it is impossible to obtain an analyte-free matrix. For example, in the measurement of an endogenous component of blood, such as insulin, it might be impossible to obtain blood without the analyte. Or for the analysis of a waste stream, it might be impossible to formulate a blank sample — that is, the waste stream containing everything except the analyte. When a blank sample cannot be obtained, the method of standard additions can be used to determine the concentration of the analyte in an unknown sample. 

Figure 3: Standard additions calibration plot from data of Table I. 

A series of calibration standards is prepared at several concentrations. The standards are then added to aliquots of the sample. For the example of Table I, calibrators were prepared and spiked into sample aliquots to result in samples that contain 0, 1, 2, 5, and 10 ng/mL of added calibrator. Next, the samples are analyzed and the results plotted, as in Figure 3. Note that the plot has a significant 

-intercept (dashed line in Figure 3); this represents the response for the analyte content of the unspiked sample. To determine this concentration, the regression line is extended to the left until it intercepts with the 

-intercept represents the negative value of the concentration in the unknown sample. This is calculated by taking the regression equation (

 = 1055/402.4 = –2.6. The negative of this value (2.6 ng/mL) is the concentration of analyte in the unknown sample.  

We have considered three types of calibration plots this month. You will most commonly encounter the external or internal standard techniques. The choice of the technique can be based upon similar methods in your laboratory, customary usage in your industry, or by empirical testing. Usually, internal standardization, because of its additional complexity, is reserved for methods that require extensive sample preparation, which can result in physical loss of sample in the process. The method of standard additions usually is reserved for cases in which a blank matrix is not available, and therefore may be rarely, if ever, encountered in many laboratories. 

The number of calibration standards used in a method also will vary depending upon the application and the custom for the laboratory or industry. If the calibration plot is linear and passes through the origin, a single-point calibration curve comprising one standard concentration can be justified. However, this technique often is reserved for methods that cover a narrow range in concentrations, such as ±10%. More commonly, calibration standards are formulated at several concentrations that span the expected concentration range of the samples to be analyzed. As a general rule, it is best to bracket the concentration range with calibration standards; extrapolation of data beyond the calibration range, while sometimes justified, adds the potential for error to the method. 

"LC Troubleshooting" Editor John W. Dolan is Vice-President of 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 

  For an ongoing discussion of LC trouble-shooting with John Dolan and other chromatographers, visit the Chromatography Forum discussion group at 

LC Troubleshooting

Part IV of this series focuses on three different calibration models: external standardization, internal standardization, and the method of standard additions.

Calibration Curves, Part IV: Choosing the Appropriate Model

The oxymoron in the title illustrates a predicament in analytical practice. Since the early 1990s, we have witnessed great strides in mass spectrometry (MS), improvements that bring sophisticated capabilities to a diverse user base. Yet along with this success has come the inevitable learning process. So where are we now? What do the early innovators see as their legacy, and what can we expect according to those who now have 15 or 20 years of experience behind them? 

Pharmaceutical practice was the leading edge of a movement allowing a fully trained mass spectrometrist to give nonspecialists access to advanced capability without the need for extensive training. To it we owe such terms of art as "open access" and "walk-up MS," which characterize a distributed rather than centralized analytical capability. The belief underpinning this distributive capability was that medicinal chemists and others could perform high-order analytical processes without the need for extensive training on rather complicated instruments. It is not automation, or a machine performing repetitive tasks with little or no human intervention that frees chemists to pursue more creative activities or those requiring discriminating judgment. Instead, open access and walk-up systems provide a menu of processes from which chemists presenting a variety of samples choose analytical tasks that run automatically. 

A wide range of automated laboratory practices has evolved to include all industries. Symyx Technologies, Inc. (Santa Clara, California), which first developed automation to advance its own research, is an example of a company that grew with the opportunity and necessity. Its initial market objectives were not drug discovery but material science programs: for example, producing petrochemical and polymer catalysts that led to high-profile collaborations with ExxonMobil and Dow. The company's expertise in polymer chemistry led to founding and then spinning off Ilypsa Pharmaceuticals, recently acquired by Amgen (Thousand Oaks, California). Though research remains a significant part of Symyx's operations, it nevertheless enjoys a reputation as a third-party integrator of laboratory workflows, with major pharmaceutical customers as well giants in other industries to its credit (

While I was researching this topic, a colleague made an interesting point. Cecilia Mazza (Senior Market Development Manager, Chemical Analysis, Waters Corporation, Milford, Massachusetts) noted that the terminology we use to describe aspects of workflow capability remains fluid as its adoption and integration evolves. She observes that it is sometimes difficult to ascertain the extent of a workflow when terms like "on-line," "at-line," and "off-line," which we blithely assume to be standard, become blurred with terms such as "near-line."  

Open access began with the tools available at the time (gas chromatography [GC] and solid analysis probes) to suit a need (increasing throughput while decreasing repetitive task necessities for humans). As laboratory technology advanced, bringing with it ever wider use of the mass spectrometer, and as the nature of compounds investigated became increasingly diverse, the need persisted. 

The idea to investigate this subject began with a discussion with one of the first people I met in MS practice, Jeffrey P. Kiplinger, who was then at Pfizer and is now president of Averica Discovery Services (Worcester, Massachusetts). Kiplinger, one of the early adopters and developers of the open-access concept, collaborated with software engineers at Fisons Instruments (Manchester, UK) division that was renamed VG Biotech in 1995 and, later, Micromass. Waters purchased Micromass in 1997 and continued the company's manufacture of mass spectrometers. 

Figure 1: Effect of introducing Open Access MS to the discovery chemistry environment at Pfizer in the early 1990s. Shown are plots of samples analyzed per year by technique (top) and the mean time required for analysis per year (bottom). (Taken from an internal Pfizer presentation courtesy of Jeffrey Kiplinger.)			

"As an organic chemist with a technology bent, I backed into a mass spectrometry career. From postdoc work with Maurice Bursey and Alan Marshall, I took my first real job in 1988, with Pfizer, to run a mass spec service group for a discovery chemistry department of about 75 chemists. Pfizer was then the world's 20th largest pharmaceutical company — before my interview I had to learn how to spell 'Pfizer' and find an annual report in the library to learn what the company made." 

"At Pfizer, my job was to set up a new capability, FAB-MS [fast-atom bombardment], which promised the capability of acquiring spectra on many of the new chemical leads the company was working on — compounds with higher molecular mass and more polar functionality than a lot of common small-molecule drugs. While the new high-resolution, double-focusing magnetic sector instrument was being ordered and awaiting installation, I received a one-year education in how mass spec contributed to the work of drug discovery." 

"My associate Dick Ware, universally known as 'Tupper,' and I acquired spectra on tiny samples of final synthesis products intended for registration in the compound file. These were purified compounds of known structure — the chemists had NMR spectra of every intermediate compound on the synthesis path, as well as the final product — and we were providing the final verification. We dissolved the samples and, using a fine syringe, added a microliter to the end of a capillary tube. Then we used tweezers to place the tube at the end of a long, cannon-like, direct-insertion probe. The probe went into the mass spec's ion source through two vacuum locks, and after a pump-down [evacuation] cycle, we heated the probe and boiled off the sample, acquiring perhaps as many as a half-dozen +EI spectra. A little more than half of those spectra showed a molecular ion; the remaining spectra (with no significant molecular ion present) gave fragment ions that we hand-annotated with structures (this was where our knowledge of McLafferty's and Turecek's 

 was important). With an external calibration, we could achieve mass accuracy to ±10 ppm on about 80% of the structures, so the chemist performing the synthesis could publish or patent." 

"As mind-numbing as this work sounds — 12 spectra a day, little creativity, no productive activity one day each month because of source cleaning — it was fun learning how drug discovery worked, sharing the load with really smart scientists, and working in a well-funded lab. Many of the chemical series we worked on gave similar spectra, and over time we learned the characteristic fragment ions these series produced. FAB ionization gave us a "molecular ion" [MH+] much more frequently, and the chemists much preferred it to the older electron ionization [EI] data. When I asked several of them why, their answer surprised me: 

'We really only look at the molecular ion. If the ion is wrong, it means we did something wrong. If it's right, it's confirmation that the product is the right stuff.' 

"Apparently, all the extra work, spectral annotation and mass measurement, were needed less than mass confirmation. Over a few weeks, it became apparent that as little as one in 50 submissions required careful inspection by the "trained mass spectrometrists." The rest were fully useful if they showed the expected molecular mass peak." 

Figure 2: AIDE automation of the NMR process workflow at Pfizer. (Courtesy of Wei Wang, presented at Association of Laboratory Automation 2004, San Jose, California.)			

"As a postdoc at Ohio State University, I published some work jointly with Professor Leo Pacquette's group. Dr. Pacquette was then the best-funded organic chemist in the department. His large group even had its own mass spec, a Hewlett-Packard GC–MS system that everyone in the lab used without the help of a mass spectrometrist. One grad student kept the instrument operational, and everyone simply put samples in the open autosampler tray and typed the entry into a sample list. They used a short capillary column and a rapid temperature gradient. The system proved indispensable for checking the purity of solvents and starting materials and even produced usable spectral data on a variety of products. When the student chemists couldn't confirm identity with GC–MS, they resorted to NMR analysis. At the time, my biggest lesson was that a mass spectrometrist was not always necessary in obtaining analytical data. Many of my colleagues didn't learn that lesson until much later." 

"Back at Pfizer, by the summer of 1990 we were running as many as 1000 samples per month [using an initial 'open access' GC system], and chemists were reporting useful structural data on a surprising number of compounds. We ordered a second system. Other than an initial 30-minute training session and regular maintenance, this GC–MS system ran without intervention by analytics staff. We never saw the data generated unless asked a question, and so to my knowledge, that instrument was the first open-access mass spectrometer (we called it "walk up") in the pharmaceutical industry." 

"In 1991 I had some discussions with a Pfizer counterpart from the U.K. Research division in Sandwich, Frank Pullen. We were encouraged by the success of the walk-up GC–MS experiment, and wondered whether the newer liquid interfaces such as thermospray (TSP) and particle beam (PB) might offer a way to acquire spectra on compounds that were too large or thermally labile to pass through a GC column. By 1992, a Fisons Trio-1000 TSP-MS system in the U.K. was operational along with an HP-5989 with a PB–ammonia CI interface in the U.S. Both gave molecular mass peaks on 80%–90% of discovery chemists' compounds. Flow injection analyses were performed on a one-minute cycle. No ionization method then seemed stable enough with liquid chromatography (LC) flows or gradients to give smooth data over a complete LC run, and the 20-minute high performance liquid chromatography (HPLC) gradients of the early 1990s were too long to implement on a walk-up system." 

"The instruments were custom built, and they included innovations unavailable on commercial production instruments. Such modifications are incorporated in today's instruments. But at the time, they were six-to-nine-month development projects for the manufacturers, and they represented some cost to Pfizer. They included a software interface for easy sample login, an ability to add samples to a running queue, and an automated idle or shutdown routine so that an unattended instrument didn't continue to run. No one had anticipated an unattended mass spectrometer, and no one had previously thought of controlling an LC system from the MS data system." 

"At the 1993 ASMS meeting in San Francisco, these two instrument system approaches were presented at a workshop on automated MS technology. At the time, these instruments, having each run 40,000 to 60,000 samples over the preceding year, were probably the most productive individual mass spectrometers ever built. The notion that unsophisticated users could operate mass spectrometers and that spectra could be returned directly to a chemist without interpretation by a mass spectrometrist was regarded as risky and establishing a poor precedent. Some members of the audience were clearly threatened. I wrote down one quotation, which I won't attribute: 

"Putting mass spectrometry in the hands of the end-user is simply irresponsible!"

"When we published our experiences with this new approach in early 1995, I titled the paper 'Putting Mass Spectrometry in the Hands of the End User' (1). Next to our paper in the 

Journal of the American Society of Mass Spectrometry

 was a similar experiential account by Lester Taylor, Bob Johnson, and Robert Raso that first used the term 'open access' to describe a similar approach at what was then Burroughs-Welcome Pharmaceuticals (now GSK) (2)."

Kiplinger's perspective is instructive: "Neither Hewlett-Packard nor the Fisons division that built the Trio-1000 MS were particularly interested in commercializing a walk-up platform. But VG Biotech picked up the project, trademarked the 'open access' phrase and ran with it. I was told that in 1996 the open access platform generated over $30 million in revenue, an unheard-of annual production for a mass spectrometer." 

"By the mid-1990s the drug discovery world was in the thrall of technology and investing resources at an unprecedented rate. High-throughput technologies for screening, chemistry, and analysis consumed so much of the industry's attention that for a time, it seemed companies forgot to use them to discover drugs. The pressures on chemists to produce and on technologists and analysts to enable production were huge, and they generated some memorable conflicts. Mass spectrometrists told stories of users who sabotaged each other by removing queued samples and substituting their own. Analysts told of users running black, tarry samples or vials half-filled with silica from scraped thin-layer chromatography (TLC) plates (clogging systems). Some groups named and shamed abusers on a 'wall of shame.' Others simply gritted their teeth and hired more support staff to keep systems operating." 

"Meetings in the 1995–2000 era focused on high-throughput technologies, now mostly linked to LC–MS. We debated the proper ratio of supporting analytics staff to chemists to run an effective discovery program. Technologies for remote instrument monitoring were presented that allowed us to see our machines from home, to check them on the weekends. For a time it was a hot role to play in pharma — developing new productivity enhancement tools that would ultimately increase the ability to discover new drugs by 100-fold or more." 

Of course, not all of that happened, while remote monitoring is commonplace now. "Bubbles pop, and we've returned to a saner way of thinking." Nevertheless, the value to a chemist of being able to acquire useful information is a constant. "Throughout the history of organic chemistry, understanding what happens in a reaction has always been a bottleneck. The little leap of understanding that took place in about 1990 was that the molecular mass of the product is the clearest and simplest indicator of the success of a synthesis. The development of open-access LC–MS, a work of nearly 10 years of partnership between pharma customers and technology companies, was driven by the value inherent in this realization. It's hard to imagine a modern chemistry program without an available open-access mass spec system." 

For the most part, Kiplinger has moved on to new challenges while today's practitioners improve on the benefit of open access. What do they have to say about it? In the following sections, we will look into where the open-access concept is headed, taking into consideration economics, technological advances, and — to the extent we can — the vicissitude of human nature. 

In March 2009, with the help of William Farrell, Senior Principal Scientist in High Throughput Analysis and Purification, Pfizer Global Research (La Jolla, California) and a director for the Society of Small Molecule Science, which organizes the CoSMoS conference each year (

), we assembled a roundtable composed of people with scientific understanding and extensive practical knowledge in the area of open access. John Van Antwerp, recognized for his role in developing and implementing open access in laboratories throughout the Americas for Waters Corporation joined in the discussion with a group of leaders in the practice of distributed high end technology. 

Farrell has more than 20 years chromatographic experience, specializing in supercritical fluid chromatography (SFC) and HPLC and hyphenated MS techniques. Since 1998 he has been working on implementing SFC and HPLC for the analysis and purification of combinatorially derived medicinal compounds. He has been a major driving force behind the development of automated preparative SFC systems and has pioneered the use of SFC for high throughput applications. We wanted to address the question of whether the instruments and their software allow seamless agreement between need and function. John Isbell, Ph.D, Director of Analytical Sciences and DMPK, Genomics Institute of the Novartis Research Foundation (GNF), La Jolla, California, worked for 10 years in mass-directed purification, including a couple of years in the Kassel group during which time he worked on a custom, parallel, two-channel PrepLCMS that was ultimately made obsolete by the MUX Purification Factory (Waters Corporation). He moved to GNF to set up a HT-purification group around the MUX Purification Factory and demonstrated its ability to purify 10,000 samples/month. John Harman, Associate Director, Research, Neurocrine Biosciences, Inc. La Jolla, California, has been leading efforts since 1997 in mass directed purification starting with open access followed by open access by mass directed purification until 2006 when the model switched to centralized or service purification. Muhammad Alimuddin , Senior Scientist, Pfizer Global R&D, La Jolla involved with the Discovery Platform Chemistry/Open Access has been administering open access systems for more than 9 years. Initial experience was in open access administration of LC, LC–MS, and GC–MS in a CRO laboratory. At Pfizer he expanded the open access portfolio by introducing reversed phase mass directed prep purification, chemical stability assay and photo stability assay using LC–MS, assessing residual solvents by GC–MS and high resolution accurate mass using an LC–time-of-flight (TOF) system. Wei Wang, Ph.D, Associate Research Fellow, Pfizer Global Research, La Jolla, California, started as an NMR spectroscopist at Princeton University in 1995, and spent the last 10 years applying his skills of physical and analytical chemistry, as well as analytical informatics at Pfizer. Wang brought complementary insights to the discussion with his experience with automating open access NMR. His current interests include two different platforms of high-throughput, capillary flow or capillary tube NMR on Varian and Bruker. 

The group determined that, in some cases, the need that first gave rise to open access persists. In pharmaceutical practice, because analytical support is primarily for medicinal chemists, and mass spectrometers involve significant costs in both initial outlay and training, their use must be warranted. Medicinal chemists are well trained in NMR practice as part of their schooling and, as Wei Wang pointed out, open access or walkup NMR offers the same advantages of open access LC–MS, namely, greater access to expensive instruments, faster turn around time and reduced overall costs. 

NMR lends itself to open access development through inherent simplicity and cooperative human behaviors. Factors like cost and complexity may keep LC and MS out of the hands of many during their school years. As Farrell points out,  

"I'm sure there are very progressive academic labs but . . . the vast majority . . . know how to run [manual] columns, TLC, and how to perform NMR analyses. Other technology is foreign to [students]. If every NMR were converted to flow-probe NMR, you would run into an equal number of problems as in HPLC." 

Wei Wang adds "Tube-based NMR tends to have fewer problems but there is still room to improve its robustness and ease of use." 

Wang and colleagues at Pfizer developed a walk-up workflow for NMR, with a collection of designed user interfaces and utility programs that incorporated third-party vendors' processes into a customized process. Their goal was to use automation to increase productivity by reducing manual, repetitive tasks, increasing reliability, decreasing response and down times, and simplifying user interfaces to help users obtain better information faster. They derived an acronym, "AIDE" from the exercise: 					

Integrate various software and technologies into one "turn-key" solution.

Develop our own process to meet users' needs.

Educate users to utilize our facility effectively.

Often, despite training coupled with experience, the most expedient approach is to centralize the practice, as Neurocine's John Harman discovered. He found open-access practice for purification generally inefficient. When deciding how to proceed, he and his team ignored instrument usage time because "the cost of a machine is nothing compared to, unfortunately, the intangible cost of time." After analyzing runs for a year, he found chemists were successful in purifying their molecules only about 40% of the time. Yet according to the chromatography, they should have been successful 70% of the time. Thus, the chemists were essentially throwing away 30% of their samples simply because they were not preparing them properly, choosing the wrong gradient, or setting the wrong threshold. 

The consequence was to centralize the purification practice. Harman identified super users and trained them for 6 h. He observed that users compete with each other and used the new knowledge to their advantage. When someone needs assistance they come to the central group for help. Lack of training in analytical chemistry leads to the incidents Kiplinger alluded to, which earned certain individuals their membership on the "wall of shame." 

John Isbell at GNF uses a different approach to ensure purification remains a distributed capability in open access. GNF has "medicinal chemists who police their own instruments." Isbell relates, "We never asked the chemists to do this. What actually happened is that the experienced chemists took it upon themselves to enforce department policies, reminding others what constitutes proper usage. The medicinal chemists made all the rules and policies for all the walk-up instruments. A key element in the success of this approach is attributed to the very low turnover at GNF." 

Other steps have proven useful, too. A company with two sites approached work submissions differently. One group took the approach that the job of chemists was to make compounds, not run LC–MS. So although the chemists worked without constraint, their administrator spent 30% of the time unclogging injectors from submitters using sample solutions made at "20 mg/mL in 90% water." 

The twin site had as many systems but required using only the vials supplied to submit samples. They supplied 4-L bottles of methanol and trained the operators simply "if you had enough material to perform NMR on your sample dilute it by 1000 add one drop of water and filter to check for solubility issues before doing LC–MS." 

The difference between the groups was a 70% up-time at the first site and a 99% up-time percentage at the second. A significant difference underscoring the approach, according to our experts, rests on the size of a company. Peer pressure works at a smaller company with limited turnover, but a larger, more diverse company might benefit from common-sense training. 

Particularly for small groups, solubility and other assays can be automated where those tasks are sufficiently repetitive and well-characterized. For example, a group can develop a chiral separation using extensive automation specifically for the purpose of creating a menu choice for open-access submitters. Thus, whereas chemists in contract research organizations are adept at using each piece of laboratory equipment, switching as needed from one project to another, those working in major pharmaceutical operations are being driven toward automation as a way to make more efficient using their time. 

Muhammad lists various new techniques developed as open access: assessing residual solvents by GC–MS; high resolution, accurate mass using an LC–TOF instrument; reversed-phase, mass-directed, preparative purification (5–1000 mg crude loading); chemical stability assay using LC–MS; photo stability assay using LC–MS. He predicts "there is demand and we may expand more in providing physical properties of compounds like Log D, p

, HLM based on a paper published last year . . . we are exploring the possibility of determine the Log D in flow mode to make it open access"(3). A thorough view of postcombinatorial thinking as we look forward to automated processes in pharmaceutical applications is presented in a comprehensive article by John Isbell (4). 

 "MS — The Practical Art" Editor Michael P. Balogh is principal scientist, MS technology development, at Waters Corp. (Milford, Massachusetts); a former adjunct professorand visiting scientist at Roger Williams University (Bristol, Rhode Island); cofounder and current president of the Society for Small Molecule Science (CoSMoS) and a member of LCGC's editorial advisory board.

MS &#151; The Practical Art

Columnist Michael Balogh looks at the questions: Where are we now? What do the early MS innovators see as their legacy? And what can we expect in the future?

Open Access, Automation, and the Promise of Simple Complexity

The 57th ASMS Conference on Mass Spectrometry was held at the Pennsylvania Convention Center, 1101 Arch Street, Philadelphia, Pennsylvania, from May 31st through June 4th, 2009. 

The conference began early for some attendees; with several one-and two-day short courses beginning on Saturday May 30th from 9 a.m. to 4:30 p.m., Sunday followed with more short courses and informative tutorial lectures from 5:00 to 6:30 p.m.   

Williard H. Harrison gave a tutorial lecture titled, "The Art of Communication in the Sciences." Harrison is Dean Emeritus and a professor in the Department of Chemistry at the University of Florida, Gainseville. He received the Ph.D. degree from the University of Illinois, where his research interests in analytical, optical, and mass spectroscopy took flight. Professor Harrison is a member of the honor society of Phi Kappa Phi, has published 109 research papers, delivered 92 lectures, and presented 138 contributed papers, a number of which were on the topic of glow discharges. Needless to say, Harrison had a wealth of knowledge to offer on the topic of communication in the sciences. 

David E. Clemmer, a professor for the chemistry department at Indiana University, gave another tutorial lecture titled, "Ion Mobility in Mass Spectrometry." Professor Clemmer received the Ph.D. in chemistry from the University of Utah and did postdoctoral work at the Himeji Institute of Technology (Japan) and Northwestern University. Professor Clemmer has published more than 135 papers and his work has been recognized with several awards, including the Fresenius Chemistry Award. He has also been a member of the U.S. Defense Science Study group. 

Following the tutorial lectures was the official opening session and Plenary Lecture, which featured a talk from Jerald L. Schnoor, of the University of Iowa, entitled, "Biofuels and Global Climate Change." This topic was especially fitting for the June conference as gas prices and temperatures are on the rise once again this summer. Schnoor received the B.S. degree in chemical engineering from Iowa State, the M.S. degree in environmental health engineering, from the University of Texas, and the Ph.D. degree in civil engineering from the University of Texas. Throughout his career, Schnoor has been very active in environmental issues. He testified several times before Congress on the environmental effects of acid deposition and the importance of passing the 1990 Clean Air Act. As Editor-in-Chief of 

, Schnoor guided the leading journal in the world in both environmental engineering and environmental science. Professor Schnoor has published (as author, co-author, or editor) six books and over 150 research articles in archival journals, in addition to serving as lead editor of a series of texts and monographs for John Wiley & Sons. His vast array of knowledge and his passion for environmental issues was evident in his lecture. Professor Schnoor was an eloquent speaker and surely set the bar high for the rest of the week with his thought-provoking presentations. 

Monday through Thursday were full program days consisting of oral sessions, poster sessions, and workshops. A few popular oral sessions from the week were, "Chromatography-Mass Spectrometry Approaches to Increase Throughput in Discovery PK Assays;" "Mass Spectrometry in Environmental Toxicology;" "Accurate Mass LC–MS Approaches for Characterization and Quantification of Drugs and Metabolites;" "Mass Spectrometry in Homeland Protection;" "LC–MS Assessment of Human Metabolism in Compliance with "MIST" Guidance;" and "Novel LC–MS Techniques for Regulated Bioanalysis of Drugs." These are just a few of the great topics that this year's ASMS covered. All of the sessions were very well researched and offered a lot of practical knowledge to any chromatography or mass spectrometry user.   

As always, the staple at ASMS was the hospitality suites, which featured complimentary cocktails, food, and a great atmosphere to meet and greet with fellow colleagues. Philadelphia was a wonderful setting for this conference, which definitely showed its "brotherly love" to all the ASMS conferees. Next year's ASMS conference is scheduled to be in Salt Lake City, Utah — we here at 

 can't wait to see what next year's program has to offer!  

Assistant Editor Megan Evans takes a look back at the annual ASMS Conference that was recently held in Philadelphia.

57th ASMS Conference Review

Supercritical fluid chromatography (SFC) has been practiced for approximately 50 years. Early studies used packed columns with varying degrees of success (1) and it would be fair to say that SFC did not generate much interest in the separation science community in this period. In the early 1980s interest in SFC exploded when it was reported that SFC could be easily accomplished with wall-coated open tubular (WCOT) columns similar to the columns currently used for gas chromatography (GC). 

The columns differed in two respects: SFC columns had smaller inner diameters and the polymeric stationary phase coating the walls was more extensively cross-linked (2). The emergence of SFC with a mobile phase that exhibited appreciable solvating power relative to helium and hydrogen was primarily promoted by users of GC who saw SFC as a way to increase the limited sample base afforded by GC. This technology, which used pure carbon dioxide, flourished for some time but it had several disadvantages for wide-scale acceptance by the analytical community. Specifically, the technique was traditionally limited to relatively nonpolar compounds because 100% carbon dioxide as the mobile phase exhibited poor solvating power similar to hexane; robust sample injection was not feasible; a wide polarity range of stationary phases was not available; high-quality separations required nearly an hour; gradient carbon dioxide pressure and flow rate were coupled such that flow rate changed as mobile phase pressure changed; passive, fragile restrictors at the column outlet often plugged and the separation had to be aborted; and multigram scale-up of a successful separation was not feasible (3). 

SFC on packed columns for both qualitative and quantitative purposes underwent a renaissance in interest at the beginning of the 1990s when the limitations of capillary SFC became obvious and important progress in composition gradient techniques for mixed mobile phases was achieved (4). Sophisticated commercial instrumentation allowing independent flow control under both pressure and composition gradient conditions came out in 1992 boosting the development of applications in many major areas of separation science. 

By 1997 it was clear that the future of SFC would focus more on the separation of moderately polar analytes with polar-bonded silica-based stationary phases, modified carbon dioxide and spectroscopic detectors. In other words, practical SFC, which affords a bridge between high performance liquid chromatography (HPLC) and GC would be more HPLC-like than GC-like (5). Many of these newer developments were inspired by the work of Dr. Terry Berger who spent more than a decade systematically undoing many of the misconceptions concerning packed column SFC (pcSFC) that existed in the 1980s (6). Some examples of the Berger findings include very long columns with large pressure drops were feasible for SFC; mobile phase carbon dioxide density and solvent strength were disconnected; the effect of mobile phase additives (that is, secondary modifiers) on peak shape and retention were introduced; packed column SFC was shown to be broadly applicable to small drug-like molecules; and quantitative SFC recovery of solutes without conventional cyclone trapping and aerosol generation were demonstrated. 

Even with these instrumental improvements, wide acceptance of the technology was not forthcoming because the perception was that highly polar analytes were not soluble in a carbon dioxide–based mobile phase and were, therefore, not separable (7). Historically, carbon dioxide has been treated as a nonpolar solvent, primarily because of its low dielectric constant and zero dipole moment. Carbon dioxide has also been described as a quadrupolar solvent because of its significant quadrupole moment. 

As far as its microscopic solvent behaviour is concerned, carbon dioxide has the potential to act as both a weak Lewis acid and Lewis base. Strong theoretical and experimental evidence has indicated that carbon dioxide can participate in conventional or nonconventional hydrogen bonding interactions (8). Nevertheless, researchers initially sought to extend the applicability of SFC by either using more polar pure fluids, such as ammonia, sulfur dioxide, nitrous oxide and Freon-23 or by adding a polar organic solvent that is, "modifier") to carbon dioxide to improve the solvating power of the mobile phase (9). Both these experimental routes, however, did not generally lead to successful separations.  

In HPLC, the surfaces of almost all packing materials are heterogeneous. Thus, the traditional explanation for lack of analyte elution involves stationary phase "active sites," which usually means the presence of some subset of silanols or metal ions on the traditional bonded-silica packing material. Mobile phase additives in HPLC are very polar substances that are added to the mobile phase at low concentration to improve peak shapes by either covering up, adsorbing on, or reacting with active sites. An alternative approach in HPLC to dealing with active sites involves either silanol endcapping or polymeric coating schemes that are meant to permanently cover active sites, thus eliminating the need for an additive to the mobile phase. All of these strategies have been explored with SFC because the stationary phases used by HPLC and packed column SFC were essentially the same in the beginning. 

In the 1990s virtually all peak distortions or nonelution events in pcSFC were attributed either to irreversible adsorption of the very polar analyte to the silica support or (unlike HPLC) to poor solubility of the compound in the modified mobile phase. In pcSFC, endcapping, deactivation and polymer-coated supports, however, enjoyed only limited success (10). The use of additives in conjunction with modifiers — as will be shown later — proved to be as highly successful for SFC as it was for HPLC. Although WCOT columns exhibit much less activity, the elution of even weakly polar analytes was considerably more problematic via this SFC mode because delivery of modified fluids was not experimentally feasible (11). Needless to say, separation of both ionizable and fully ionic analytes with a carbon dioxide–based mobile phase during the early 1990s was thought to be highly unlikely. In other words, SFC provided (at the time) a number of advantages over traditional HPLC, such as speed, practical use of longer columns, a normal-phase retention mechanism and reduced use of organic solvents, but the nature of SFC mobile and stationary phases was thought to not allow the elution of highly polar analytes — much less ionic compounds. 

Suffice to say, this situation has drastically changed because experimental approaches using additives have been quite successful. Effective additives are generally too polar to be miscible with carbon dioxide. Instead they are added to the primary modifier and then the modifier plus additive is pumped as a single ternary fluid thereby augmenting the solvating power and flow of pressurized carbon dioxide. In general, ionizable analytes cannot be eluted — or else they are eluted with severely distorted peaks without an appropriate additive in the mobile phase. Today, it is apparent that the use of additives dramatically extends the range of solute polarity amenable to carbon dioxide–based SFC. This review focuses on publications that describe the application of packed column SFC to both ionizable and fully ionic compounds. 

Berger has suggested that additives perform at least four different functions (12): cover active sites on the solid support; change the polarity of the stationary phase; suppress ionization of the analyte; enhance ion pair formation with the analyte; and raise the polarity and solvating power of the mobile phase. The most obvious role for additives is to change the mobile phase polarity and solvent strength. In this regard, solvatochromic dyes have been used to measure the solvent strength of tertiary mobile phases (13). Small concentrations of additive were sometimes (but not always) found to increase the apparent polarity or solvent strength of the mobile phase. Ion suppression is another rather obvious role for additives, especially if the additive is clearly more acidic (or more basic) than the acidic (or basic) solute. 

Figure 1 shows the separation of lovastatin with and without trifluoroacetic acid (TFA) in the mobile phase. TFA is theorized to suppress the ionization of the carboxylic acid group to produce a much sharper peak. In general, the first small addition of additive improves peak shape and sometimes shifts retention. The addition of higher concentrations of the same additive tends to have little further effect. One difference between acidic and basic additives has been noted. Multifunctional acidic additives have been effective in suppressing peak tailing, but multifunctional basic additives have degraded peak shape. 

Stationary phase surface coverage by additives is less well defined for conventional columns because the small particle surface is made up of both exposed silica solid support and bonded stationary phase. Surprisingly, carbon dioxide has been shown to strongly adsorb onto column packing. Maximum adsorption of carbon dioxide occurred near the critical temperature and critical pressure (14). Parcher and colleagues have suggested that a thick film of carbon dioxide may act as part of the stationary phase. Modifiers were also shown to strongly adsorb to the stationary phase (15). Because the total amount of adsorbed material increased under these conditions, the adsorption of modifiers was stated to not displace adsorbed carbon dioxide. The quantity of additives that adsorb onto various stationary phases has been measured (16). 

Low to moderately polar stationary phases, such as octyl and cyanopropyl, adsorbed only small amounts (0.4–0.6% of a monolayer) of additive. Under the same conditions sulfonic acid and diol columns adsorbed much larger amounts of additive creating surface coverage up to 21% of a monolayer. Finally, certain additives can be very strongly held by the stationary phase thereby changing the stationary phase polarity, vide infra. In this case, additives can cause an increase, a decrease, or no change in solute retention. In these experiments, the sudden introduction of an additive does not result in an immediate change in retention time because the initial process is very similar to a titration with an endpoint. 

Chromatographers readily agree that SFC with polar-bonded stationary phases and carbon dioxide–based mobile phases is normal phase chromatography without most of the problems usually associated with classical normal-phase HPLC such as the strong dependence of the retention time on the concentration of small amounts of very polar constituents in the mobile phase. As such, the orthogonality of SFC to reversed-phase LC is an attractive asset for purity assessment (17). Equilibration is very fast, usually requiring flow turnover of little more than three to five column volumes. Composition, pressure and temperature can all be programmed with rapid recovery to initial conditions between analyses. 

Solvent characteristics of the fluids used in SFC are not unique to supercritical fluid conditions (12). The characteristics are present irrespective of whether the fluid is defined as a liquid, a dense gas, or supercritical fluid. Many times, the operator cannot tell when the definition (or name) of the mobile phase changes from supercritical fluid to subcritical fluid during a typical separation. As a result, the nomenclature has often been confusing and loosely applied in this area. SFC today usually uses carbon dioxide above or near its critical temperature of 31 °C and critical pressure of 73 bars, combined with an organic modifier such as methanol or ethanol and an additive. The differences in SFC and subcritical fluid chromatography have been overstated in the past. Diffusion coefficient, viscosity, density and solvent strength all show minimal changes when the mobile phase changes from super- to subcritical. When outlet pressure is elevated and internal pressure and temperature are controlled, the resulting techniques are similar and the behavior of conventional GC and HPLC are completely and seamlessly bridged to SFC (18). 

SFC has become a powerful tool for both high throughput qualitative and quantitative analyses at various stages of drug discovery and development. Often used in parallel with other separation techniques, SFC has been extensively employed in purity assessment, structure confirmation–elucidation, degradation profiling and formulation assay. The remainder of this review will describe some of the studies where ionic analytes have been chromatographically separated using a carbon dioxide–based mobile phase. The descriptions will draw upon various issues that have been noted earlier in this paper. The primary goal of the manuscript is to demonstrate to the reader that pcSFC is readily applicable to highly polar analytes. 

Historic Applications: Ion Suppression Against Ion Pairing

The claimed (but erroneous) chromatographic separation of charged species via SFC dates back to 1988. An advertisement showed a capillary SFC chromatogram of quaternary ammonium salts obtained with carbon dioxide as the mobile phase. After further review by Sandra and David, it was discovered that the quaternary ammonium salts had been converted off-line to amines and the reported separation was of the latter analytes although the title of the offering continued to mislead the reader (19). 

In the same year, the separation of five charged enantiomeric 1,2-aminoalcohols as diastereomeric ion pairs using SFC with a cyanopropylsilica stationary phase was reported. N-Benzoxycarbonylglycyl-L-proline, which was dissolved in the acetonitrile modifier along with triethylamine, was employed as the chiral counterion (20). The resulting solution was mixed with liquid carbon dioxide in a 1:4 (v/v) ratio. Triethylamine served as the competing ion as it became protonated by the carbon dioxide phase. The retention and separation efficiency was stated to be controlled by pressure and temperature as well as by the concentration and nature of the ionic mobile phase components. Variation of the nature and concentration of the counterion and the competing ion was predicted to make the system highly versatile. 

These ion-pairing principles were further tested via SFC two years later (21). Successful elution of propanolol hydrochloride from a cyanopropylsilica packed column with 25 mM sodium heptane sulfonate in methanol-modified carbon dioxide was reported. Without sulfonate in the mobile phase, the analyte was not eluted. The influence of dimethyloctylamine, tributylamine, and citric acid on chromatographic selectivity was studied. A prime limitation of ion-pair SFC was found to be the solubility of ion-pairing agent in the carbon dioxide–modifier mixture. As a rule of thumb, the paper suggested that the best choice of initial conditions when starting to optimize the separation of ionizable compounds was to use a diol phase, tributylamine, and acetate ion in methanol as the ion-pairing agent. 

Much later an assay of berberine and palmatine in Phellodendri Cortex was established using ion-pairing SFC on-line coupled with ion-pair supercritical fluid extraction (SFE) (22). Dioctyl sodium sulfosuccinate in methanol (100 mM) served as the ion-pairing agent for both the SFE and the SFC experiments both of which were completed in approximately 20 min. A silica gel column was used as an SFE trapping and SFC separation column. The modifier was 10% (v/v). 

The need for an additive to facilitate separation of ionizable analytes was challenged years later by Geiser (23). Enantioseparation of ionizable amine hydrochloride salts and free acids without the use of amine or acid additives was reported. Examples were shown for racemic mixtures of three amine salts (propranolol, thioidazine, and tramadol) and a carboxylic acid (flurbiprofen). Tandem ultraviolet and polarimetric detection in conjunction with a Chiralpak AD-H chiral stationary phase using methanol-modified carbon dioxide was used. A wide dynamic range, suitable for preparative chromatography, was demonstrated for column injections of 0.15–6 mg rac-flurbiprofen. 

Although the authors did not comment on this aspect, the study is subject to criticism in that the free base and free acid may be the actual species being separated rather than the salt as claimed. For example, the amine salts investigated are derived from either secondary or tertiary amines and in solution they may exist in equilibrium with the free amine. In other words, good evidence for the separation of the actual amine salt appears to be missing. As far as the acid, flurbiprofen, is concerned, the inherent acidity of carbon dioxide should ensure the predominance of the neutral acid form in the mobile phase. As before, proof for the separation of the conjugate base of flurbiprofen appears to be missing from the report. It should be noted, however, that for many acidic analytes the acidity of carbon dioxide is not sufficient to totally suppress ionization. For example, SFC of a variety of acidic polyphenols has been successfully studied with several acid additives such as citric acid and trifluoroacetic acid (24). While certain polyphenols eluted with sharp peaks without additive, others, which were more strongly acidic, required additive. 

Uncertainty regarding elution of the free base versus the amine salt is prevalent in the early study by Gyllenhaal and colleagues (25). Metoprolol tartrate, metoprolol succincate, and hydrochloride analogues were initially dissolved in dichloromethane. SFC was then performed on both a diol-silica or aminopropyl-silica column with triethylamine (TEA) or TEA mixed with acetic acid. Given these conditions, separation of the free base was most likely occurring.  

Gyllenhaal and colleagues have also investigated the effects of binary additives in the mobile phase for separation of metoprolol and related amino-alcohols (26). Use of ethanesulfonic acid (ESA) and triethylamine as binary additives to methanol modifier and a diol column showed that when the additives were discontinued their positive effect disappeared virtually instantaneously. Standard conditions used were 10% methanol, containing 24 mM of acid and 18 mM of TEA, in carbon dioxide with a flow-rate of 1.5 mL/min. The column dimensions were 125 mm × 4 mm and kept at 40 °C with a back pressure of 150 bar. The effects on selectivity were stronger with trifluoroacetic acid than ethanesulfonic acid. The described approach was thought to offer a way to tune the selectivity of SFC systems when amines are analyzed without the need to change the stationary phase of the chromatographic system. Subsequently, these same workers have reported the separation of these analytes without an amine and basic additive from both an aminopropyl-silica and an ethylpyridine-silica column (27). 

Recent Applications: Ion Suppression Versus Ion Pairing 

Pinkston and colleagues brought new thinking to this area recently when they showed that low levels of volatile ammonium salts as mobile phase additives allow the elution of polar and even ionic organic materials from a Deltabond cyanopropyl-bonded silica column (28). In contrast to other common additives, the volatile ammonium salts were compatible with mass spectrometric detection. The SFC-UV chromatograms of reserpine standard with methanol and with 1.1 mM ammonium acetate in methanol are shown in Figure 2. 

Two mechanisms depending on the presence or absence of ammonium salt were suggested to have the greatest influence on retention in the carbon dioxide–methanol–ammonium salt–Deltabond cyano system. The salt effect was stated to be a result of either ion pairing, deactivation of the silica surface, or both. The effect of ammonium acetate additive in the mobile phase for the elution of seven pharmaceuticals is shown in Figure 3. None of these analytes are eluted with only a methanol modifier. 

In a closely related study, the effect of various ionic additives on the elution of ionic compounds such as 4-dodecylbenzene sulfonate and sodium 4-octylbenzene sulfonate has been determined (29). The additives were lithium acetate, ammonium acetate, tetramethylammonium acetate, tetrabutylammonium acetate, and ammonium chloride dissolved in methanol. Three stationary phases with different degrees of deactivation were considered: conventional cyanopropyl, Deltabond cyanopropyl, and bare silica. Each sulfonate salt was successfully eluted using all the additives with good peak shape under isocratic–isobaric–isothermal conditions. The column memory effect created via strong interaction of the additive with silanol sites for sodium 

-toluene sulfonate on a silica column with tetrabutylammonium acetate as additive is illustrated by the series of chromatograms in Figure 4. Different additives yielded different retention times and in some cases different peak shapes. 

The presence of the additive was crucial concerning elution of the sulfonate salts. Solid state silicon-29 nuclear magnetic resonance spectroscopy suggested that ammonium ions deactivated available silanol sites on both the Deltabond phase and the bare silica (30). Ammonium ion was also speculated to interact with cyano-sites on the bonded phase. Computational calculations concerning the charge distribution on various ammonium salts were performed in an effort to explain the elution behavior of the sulfonate salts. The acetate counterion was thought to facilitate elution of the anionic sulfonates from the positively charged stationary phase in a pseudo ion-exchange mechanism. On the bare silica where abundant silanol sites are present, modification of the stationary phases was thought to be the dominant effect. Ammonium ion–induced dipole interaction with the cyano function may be dominant with the more deactivated Deltabond cyano phase. 

Elution of cationic species such as amine salts in SFC is more difficult to prove unless the amine salt is quaternary (that is, NR

 ) (31). Otherwise, equilibrium with the neutral form may give rise to elution of only the neutral or a mixture of the conjugate pair. In this regard, the successful elution of two quaternary amine salts from a Deltabond cyano-bonded silica column with the addition of a sodium alkylsulfonate to the methanol-modified carbon dioxide–based mobile phase has been reported. An ion pairing interaction between the positively charged analyte and the anionic part of the alkylsulfonate additive was proposed. This study demonstrated the reciprocity of ion-pair SFC in that if amine salts are good additives for elution of sulfonate salts, then sulfonate salts should be good additives for elution of amine salts. This observation, however, does not speak completely to ion-pair SFC, but rather to additive deactivation of the stationary phase because Hoke and colleagues (32) have shown via packed column SFC–tandem mass spectrometry (MS–MS) that cetylpyridinium chloride in saliva is readily eluted from a 5 cm Deltabond cyano column with ammonium acetate additive in the mobile phase in approximately 12 s with a retention factor of 2. 

In some cases, workers have discovered that conversion of the neutral analyte into an ion actually enhances efficient chromatographic separation by favoring an ion-pairing mechanism. Stringham has reported that incorporation of a strong acid such as ESA in both the sample solvent and mobile phase modifier gives a dramatic improvement in the separation of chiral basic compounds (33). Removal of the acid from either the sample solvent or the mobile phase resulted in loss of the separation. The mechanism of separation appeared to involve an intact salt pair formed between the protonated basic compound and the conjugate base of ESA. Methanesulfonic acid consistently gave shorter retention times than ESA. With camphorsulfonic acid, selectivity was consistently lower than that observed with ESA. TFA in place of ESA gave either broad or absent peaks. 

Blackwell and Stringham succeeded to elute peptides such as bradykinine and oxytocin from a polymeric column using eptadecafluorooctanesulfonic acid as an additive in methanol with carbon dioxide from a divinylbenzene polymeric (that is, 500 Å) column (34). Ion pair effects were discussed as being responsible for the enhanced elution compared with methanesulfonic acid and TFA as additives. The potassium salt of heptadecafluorooctanesulfonic acid as an additive failed to elute any of the polypeptides. The role of the modifier was found to be secondary to that of the mobile phase additive. As progressively stronger acidic mobile phase additives were used, the peak profiles of the various polypeptides improved and retention decreased. 

A more recent report concerns the separation of polypeptides containing a variety of acidic and basic residues with up to 40 mers via packed column SFC–MS (35). TFA was employed as the additive in a carbon dioxide–methanol mobile phase (which was compatible with MS detection) to both suppress deprotonation of peptide carboxylic acid groups and to protonate peptide amino groups. A 2-ethylpyridine-silica column was used for the study. 

Figure 6 shows mass chromatograms of five polypeptides (angiotensin I, angiotensin II, angiotensin III, urotensin II, and sauvagine) with various mobile phase modifier–additive combinations. All three angiotensins gave the sharpest peak shapes with 13 mM TFA in methanol as modifier. Urotensin II also was eluted fastest and provided the best peak shape with 13 mM TFA in methanol. With the salt of TFA (that is, ATFA), the peptide was retained strongly. Sauvagine contains over 40 residues, having a molecular mass of 4500 Da including four basic amino acid groups and seven acidic amino acid groups. Only when 13 mM TFA in methanol was used as modifier was sauvagine eluted with a sharp symmetrical peak shape. The mechanism of separation is not known. No doubt TFA plays several important roles such as both protonation of the analyte and the ethylpyridine stationary phase. It, however, is clear that polar, high molecular weight, protonated polypeptides have been chromatographed employing some form of ion-pair (or ion-exchange) SFC–MS. 

Packed column SFC has experienced tremendous growth as an analytical and preparative technique over the past 20 years (36). Unlike GC and HPLC, where various retention models have been extensively developed for years, SFC suffers from the lack of systematic studies on retention mechanisms and useful models for solute retention. Wu has discussed three different retention models: empirical models, thermodynamic models, and extrathermodynamic (that is, linear free energy relationships) models (37). The ion pairing principle in SFC may not match any of these general retention mechanisms. Nevertheless, SFC of highly polar analytes appears to have been demonstrated on numerous occasions over the past 40 years. As more separation scientists grasp the true benefits of SFC, new applications are expected with enhanced mechanistic and thermodynamic understanding. 

The assistance of Negin Nazem during the preparation of this manuscript is gratefully appreciated. 

Larry T. Taylor is Emeritus Professor of Chemistry at Virginia Tech, Blacksburg, Virginia, USA. He serves on the editorial boards for 

Journal of Chromatographic Science, Chromatographia, 

He teaches short courses on SFC and SFE. Approximately 200 peer-reviewed publications dealing with SFC and SFE have been published. He Chaired the Scientific Committee for both SFC 2008 and SFC 2009 international conferences. 

The author looks at both the history and the recent developments in supercritical fluid chromatography (SFC).

LCGC North America-06-01-2009

The Continuing Acetonitrile Shortage: How to Combat it or Live with It

Calibration Curves, Part IV: Choosing the Appropriate Model

Open Access, Automation, and the Promise of Simple Complexity

57th ASMS Conference Review

Separation of Ionic Analytes via Supercritical Fluid Chromatography: Achieving the Impossible