Trace Analysis by Liquid Chromatography - Analytical Chemistry (ACS

James F. Lawrence. Anal. Chem. , 1980, 52 (11), pp 1122A–1135A. DOI: 10.1021/ac50061a718. Publication Date: September 1980. ACS Legacy Archive...
0 downloads 0 Views 8MB Size
Trace Analysis by Liqui In the past decade, liquid chromatography (LC) has probably experienced a greater development than any other analytical technique. Most of the initial applications were carried out in the pharmaceutical field where LC was ideally suited to monitoring drug preparations for quality control and purity, especially those compounds that were unsuitable for analysis by gas chromatography ( G O due to their thermal instability or poor volatility. Since then, LC has gained acceptance in many other areas of analytical chemistry, for both organic and inorganic substances. However, applications to trace analysis have been much more limited. Many workers initially believed that one of the main advantages of LC over GC was that the former required much less sample preparation. But experienced analysts, especially those who have attempted the determination of trace quantities of organics in samples such as foods, soil, natural waters or biological tissues and fluids, know that this is not true. In fact, LC with UV detection generally requires at least as much sample extraction and cleanup as GC with electron capture detection. This misconception concerning sample preparation illustrates the need to understand the advantages and limitations of LC as applied to trace analysis. There are many factors that must be considered when one is attempting trace analysis by LC, and one of the first questions should be "Is LC the best approach to t a k e ? " At the moment, there are a large number of chromatographers who try to do all analyses by LC even though, in many cases, GC is far superior. A good analytical chemist will employ the best technique available to him to successfully solve an analytical problem by the simplest means possible. LC vs. GC Most analysts entering the field of LC probably have had previous expe-

rience with GC. Although the concepts of "chromatography" remain the same, there are important differences between the two. In GC, the two major variables used to achieve a separation are the type of stationary phase and column temperature. In LC the separations may be described as solutesolvent interactions where the solute molecules establish an equilibrium between stationary and mobile phases. For LC analyses one has the choice of different stationary phases and mobile phases. Column temperature is usually ambient, and any deviations from this are used for special cases. At present, there are over 150 commercially available stationary phases for GC. In LC the choice of stationary phases is far more limited. For adsorption chromatography, silica gel or alumina is usually used. There are perhaps four or five commercially available types of polar bonded phases useful for normal-phase partition chromatography and about seven or eight for reversed-phase chromatography. Research in this area is continuing at a good pace. However, at the moment, the vast majority of LC analyses are carried out on either silica gel or the Cs or Cis reversed-phase materials. From this it can be gathered that the mobile phase plays a far greater role in LC than in GC. It is this variable that gives LC its great flexibility. Such types of chromatography as adsorption, normal-phase partition, reversed-phase partition, ion-exchange, ion-pair, exclusion, ligand-exchange and affinity chromatography make possible the separation of closely related species by a much wider variety of selective interactions. In effect, LC should be suitable for any type of compound t h a t is soluble in a liquid suitable for a mobile phase. In comparison to GC, LC is much less flexible in its detection systems. There exist a number of very sensitive general and selective detectors for GC. Many a t t e m p t s have been made to

1122 A • ANALYTICAL CHEMISTRY, VOL. 52, NO. 11. SEPTEMBER 1980

0003-2700/80/A351-1122S01.00/0 © 1980 American Chemical Society

Report James F. Lawrence Food Directorate, Health Protection Branch, Tunney's Pasture, Ottawa, Ontario K1A 0L2 Canada

Chromatography convert several of these for use with LC systems. However, because of the problems associated with the phase change (i.e., converting the liquid stream into a vapor), this approach to LC detection has not become popular. Thus, the most widely used detectors to date are those that carry out measurements in liquid streams. These include detectors based on absorbance, fluorescence, electrochemical oxidation or reduction and refractive index differences, the last being the least sensitive or selective, thus generally unsuitable for trace analysis. In GC, to determine several compounds over a wide polarity or volatility range, temperature programming is employed. The equivalent in LC is gradient elution where the mobile phase composition is altered continuously during the chromatographic run. This normally requires two pumps and the necessary electronics to program them. However, electronically controlled mixing devices have been developed for solvent mixing and gradient elution with a single pump. Such an apparatus should become popular since it avoids the cost of a second pump. The reproducibility of gradient runs and time required to return to initial conditions are somewhat superior in GC compared to LC. It is of course easier to make reproducible temperature programming runs in GC since the stationary and mobile phases are not altered. However, in LC where the mobile phase composition actually changes, the initial chromatographic conditions require a longer time to be reestab-

Figure 1 . Chromatograms of terbacil in corn (2.0 ppm) and potato (0.2 ppm). (a) = LC-UV (254 nm) and (b) = G C EC. Injections were made from the same sample extracts. Arrow indicates terbacil peaks

lished. Also, retention values (k') are less reproducible from one run to another in LC than in GC. Each system is susceptible to detector interferences resulting from temperature (GC) or mobile phase (LC) changes. These can be especially troublesome when doing residue analysis where high detector sensitivities are required. Therefore, in both types of chromatographic systems, isothermal (GC) or isocratic (LC) separations should be employed wherever possible. Since most analytical labs have more GC than LC equipment, methodology is usually directed to GC. It is the opinion of this author that LC need be used only for those compounds that are not easily determined by GC. There is little value in developing LC methodology for compounds such as organochlorine pesticides, for example, since the GC methodology is so well developed. Figures 1 and 2 compare GC-EC and LC-UV results of two pesticides in foods. The chromatograms each represent results from the same sample solutions. It can be seen in Figure 1 that for terbacil, a uracil-type herbicide, the LC results are far superior to the GC ones. However, at the same time, the opposite is true for the wild oat herbicide benzoylpropethyl as shown in Figure 2. Thus, it cannot be said that one technique is better than the other. LC and GC may complement one another and should be employed in that manner.

Trace Analysis vs. Formulations Analysis Probably the greatest use of LC at present is for formulations analyses. The reason is that the solutes are present at concentrated levels in relatively pure form. Commercial preparations of drugs, cosmetics, pesticides, coloring agents, etc. are often monitored for quality control by direct LC analysis with little or no cleanup. Trace or residue analysis on the other hand is far more difficult to

ANALYTICAL CHEMISTRY, VOL. 52, NO. 11, SEPTEMBER 1980 • 1123 A

mm of H PLC chromatographers use Burdick& Jackson solvents* ...and with good reason Superior quality makes them best for H PLC. Low UV absorbing impurities Low non-UV absorbing impurities which may otherwise produce artifacts Low residue on evaporation Low level of impurities detectable by gradient elution reverse phase H PLC

Consistent and rigorous quality control eliminates "surprise" solvent impurities. Quality is controlled not only by liquid chromatography but also UV spectrophotometry and gas chromatography—including flame ionization and electron capture detectors. Our specifications are not ideals. They show the maximum allowable limits. Typical lots are normally well below our specifications. Twenty years of attention to the highest quality have resulted in unique techniques for purification and meticulous bottling methods which guarantee product purity. B&J doesnt guess what kindoi purity you need. B&J delivers solvents of a singularly universal high quality. Call 616/726-3171 today for information about these and our many other fine solvents for your HPLC work! Methanol Methylene chloride Acetonitrile UV Chloroform Tetrahydrofuran UV Water *From "HPLC Market Study" conducted by Centcom, Ltd. for Analytical Chemistry magazine.

BURDICK & JACKSON LABORATORIES, INC. 1953 SOUTH HARVEY STREET • MUSKEGON, MICHIGAN 4 9 4 4 2 CIRCLE 31 ON READER SERVICE CARD

1124 A • ANALYTICAL CHEMISTRY, VOL. 52, NO. 11, SEPTEMBER 1980

Figure 2. Chromatograms of benzoylpropethyl in corn (0.2 ppm). (a) = LC-UV (254, nm), (b) = GC-EC. Injections were made from the same sample extract (diluted in the case of the GC-EC results). Arrows indicate the peaks carry out due to the nature of the samples and low concentrations of solute. Materials such as food, biological tissue and fluids, soil, and natural waters create great problems for LC analyses. T h e removal of a drug or pollutant from such matrices inevitably brings with it a host of compounds with many similar properties (solubility, polarity, etc.). T h e analyst's problem becomes one of how to remove as many of these as possible for success-

How to simplify HPLC using low-pressure valves. Rheodyne's Technical Note 3 shows eight ways of using lowpressure switching valves to simplify method development and routine analysis in high-pressure liquid chromatography. These economical 2-position and 6-position Teflon valves can be inserted before the pump or after the column. Among uses described in detail are switching between reservoirs to select the correct mobile phase for a particular routine analysis. Switching between several different solvents during method development when seeking maximum selectivity. Switching to a flushing solvent. Switching effluent back to the reservoir to conserve solvent And switching effluent to a fraction collector.

Send for Tech Note #3 All techniques are fully described in this well-illustrated 6-page technical note. Contact Rheodyne, Inc., RO. Box 996, Cotati. Calif. 94928. U.S.A. Phone (707) 664-9050.

RHEODYNE THF LC CONNECTION COMPANY

Figure 3. Comparison of absorbance (a) and fluorescence (b) for the detection of aflatoxins B,, B2, G-, and G2 fui LC analysis. How this is approached depends primarily upon the nature of the compound of interest. For example, if the compound of interest were strongly fluorescent, one would try to employ a fluorescence detector for the analysis. The cleanup would be directed toward reducing the fluorescent background. It is possible to have many coextractives present in the final sample solution but if they do not fluoresce, they will not interfere in the analysis. However, it must also be kept in mind that when too much coextracted material is present it can interfere in the chromatography by distorting peaks or altering retention values. Even though chromatograms may appear relatively clean because of selective detection, the repeated injection of "dirty" samples can lead to contamination and to decreased column life. To keep the chromatographic system functioning well as long as possible, it is good practice to inject as little sample as necessary to obtain a reliable result. This is especially important for trace analysis and of lesser importance for formulations analysis. Pump and Injector Requirements The minimum requirements for pumps are that they deliver essentially pulseless flow over the range 0.5-2.0 mL/min at high detector sensitivity. The noise, of course, depends upon the type of detector but generally a pump that produces less than 1% noise above the detector noise at 0.01 absorbance units full scale (AUFS) on a UV detector is satisfactory. Most popular pumps in use today, including

CIRCLE 186 ON READER SERVICE CARD 1126 A • ANALYTICAL CHEMISTRY, VOL. 5 2 , NO. 1 1 , SEPTEMBER

1980

the dual and single piston reciprocating pumps, easily meet this requirement. The most useful injection ports are those incorporating a syringe-loop configuration. These permit injection of different volumes up to the volume of the loop. Also, highly reproducible injections can be made when the full loop volume is used. They are easy to use and are preferred to stopped-flow injection or injection via a septum. Although all syringe-loop injectors are rated for operation at 3000 psi or more (which is more than adequate for most LC applications), those capable of operating at 6000 psi permit the use of higher flow-rates for rapid mobile phase changes or column conditioning or cleaning. Detectors The prime requirement for LC detectors that are to be used for trace analysis is high sensitivity. This is very important when one is considering the quantities of sample extract that must be injected to produce a reliable peak for the component of interest. If detector sensitivity is poor, then more sample equivalent must be injected, and if done on a regular basis this will lead to shortened column life. Sensitivity becomes increasingly important if extremely low levels (e.g., parts per billion), are to be determined. For most applications, detectors sensitive to 1-50 ng of substance are required. This of course eliminates the refractive index detector for application to trace analysis. Selectivity, another important factor in detection, can be considered to

Figure 4. Comparison of GC-EC and L C - U V for the analysis of metoxuron (ethylated) in soil (1.0 ppm). (a) = 1 mg equivalent of sample injected, (b) = 200 mg

be the ability of a detector to discrimi­ nate between the compound of inter­ est and t h e coextracted material. Usu­ ally advantage is taken of some elec­ tronic or optical property of the analyte. Figure 3 compares the simulta­ neous absorption and fluorescence de­ tection of some aflatoxins. All four compounds can be detected by ab­ sorption with similar sensitivities; however, by fluorescence, Gi and G2 are more sensitive. T h e merits of each detection system can only be judged after application to actual sample analyses. A detector may be capable of detecting subnanogram quantities of a compound, but t h a t does not necessar­ ily mean it is capable of detecting low concentrations in a sample where many coextractives are present. Figure 4 clearly shows this by comparing G C - E C with L C - U V detection of the same sample extract containing the herbicide metoxuron (ethylated) (1). Here, the EC detector was 100 times more sensitive to the compound than the UV detector. However, as can be seen when quantities are injected such t h a t the sample backgrounds are simi­ lar, the L C - U V system is slightly su­ perior in terms of minimum detect­ able levels (mg/kg) in an actual sam­ ple. Detector selectivity and sensitivi­ ty as well as cleanup go hand in hand in determining the minimum detect­ able levels in real samples. Usually, more sample cleanup is required as detector selectivity decreases. This applies also to the UV detector when operated at different wavelengths.

Figure 5. Comparison of the resolution of two peaks at differ­ ent column efficiencies, (a) 200 theoretical plates; (b) 500; (c) 1500; (d) 3200. (Calculated for the first peak.)

One wavelength may be superior to another, not because of sensitivity but because of selectivity. T h e problem of choosing which type of detection sys­ tem to use is not very great since there are only a few available suitable for trace analysis. These include absorbance, fluorescence, electrochemical and transport detectors. All are sensi­ tive to low nanogram or smaller quan­ tities of many compounds. T h e y differ as to their detection mechanism, com­ patibility with mobile phases, and se­ lectivity. Chromatography Columns

T h e r e are no special requirements of chromatography columns t h a t make some more suitable for trace analysis than others. The ability of a column to resolve a number of peaks within a given time frame at a constant mobile phase composition and flow rate is re­ ferred to as efficiency, and it increases as a column resolves more peaks per unit time. This is important for both trace and formulations analysis since no m a t t e r what concentration, the so­ lute molecules must be separated from other components before being de­ tected and quantitated. Figure 5 com­ pares the effect of column efficiency on the resolution of two peaks. Col­ umns with high efficiencies permit the elution of a solute at a given retention value (k') in a minimum volume of mobile phase. T h u s , as efficiency in­ creases, the volume of mobile phase t h a t contains the solute decreases. In other words, the solute concentration

1128 A • ANALYTICAL CHEMISTRY, VOL. 52, NO. 11, SEPTEMBER 1980

increases causing an increased detec­ tor response, measured as peak height, as illustrated in Figure 5. This is an advahtage when doing trace analysis since minimum detectable levels are always measured in terms of peak height (signabnoise ratio). Peak area is not affected significantly by efficiency. One important point to consider concerning chromatographic efficien­ cy is t h a t high efficiency columns are not always required. For example in Figure 5, chromatogram (b) may be perfectly acceptable for doing formu­ lations since the two peaks are com­ pletely resolved. T h e same can be said for trace analyses as long as the peak height is such t h a t it is above the min­ imum detectable level of the detector. T h e time for a chromatographic run is dependent not only upon how long it takes the solute molecule to elute, b u t the time required for elution of any sample coextractives as well. This is important since late peaks can in­ terfere in further analyses. Such prob­ lems are well known in GC. Since the shortest time possible for a chromato­ graphic run is preferred, the solute should elute as quickly as possible without decreasing the required mini­ m u m detectable levels in the sample. How fast this is done depends upon sample interferences. In trace analysis by LC, retention values greater than k' = 2 are usually required. Usually, close to the solvent peak (ίο) there is an abundance of coextracted material. As one progresses further away from it, the chromatogram becomes cleaner.

INVOLVED WITH

Pesticides? Toxicology? Environmental Research?

Figure 6. Comparison of the direct analysis (a) of the insecticide, carbofuran (UV absorbance detection) with the same analysis after formation of the fluorescent dansyl derivative (b). Arrows indicate the carbofuran peaks

Extraction and Cleanup New Corco HPLC-Grade High Purity Solvents Are Double-Distilledin-Pyrex" for High Reliability and Repeatability with HPLC, Gas Chromatography, Spectrophotometry Corco double-distills HPLC solvents in Pyrex" —to provide the exceptionally high purity required for reliable and repeatable results in HPLC work. We've even labeled them HPLC-Grade to distinguish them from other solvents—even from our own ACS grade reagent solvents. If you're involved in pesticide analysis, toxicology work, environmental research—anywhere vou need the reliability and repeatability of true high purity solvents—count on Corco to meet your stringent requirements. Our Pyrex-double-distilled HPLC-Grade solvents are also ideal for gas chromatography and spectrophotometry. For complete details call or write Corco today—or circle the number.

CORCO CORCO CHEMICAL CORPORATION Manufacturers of Reagent&Electronic Chemicals Tyburn Road S Cedar Lane · Fairless Hills. PA. 19030 (215)295-5006 •Pyrex-registered Trademark. Corning Glass Works

Sample preparation is the main factor that distinguishes trace analysis from formulations analysis by LC. The degree and type of sample preparation depend upon several factors, beginning with the chemical and physical properties of the analyte itself. Other factors such as the type of detector and chromatography system will also influence the approach to the cleanup. The analyst must ensure that none of the solvents or other materials used in the extraction and cleanup will interfere in the final determination. For example, if UV detection at 254 nm is employed, then benzene or acetone should be avoided since traces of these in the final mixture may cause unnecessary interferences. The extraction and cleanup procedure must in general be compatible with the chromatographic system employed. For example, adsorption chromatography is well suited to existing methodology for the analysis of trace levels of pesticides. Usually, cleanup involves solvent partitioning followed by chromatography on an adsorption column such as Florisil or alumina. The final residue is dissolved in a suitable nonpolar organic solvent (preferably mobile phase) for chromatographic analysis. However, if reversed phase chromatography is used, dissolving the final sample in an aqueous solution of methanol or acetonitrile may be difficult, especially if the final residue contains some coextracted sample oils. It may be possible to dissolve the sample in pure methanol or acetonitrile but then chromatography m a y be adversely affected if t h e m o -

CIRCLE 33 ON READER SERVICE CARD

1130 A • ANALYTICAL CHEMISTRY, VOL. 52, NO. 11, SEPTEMBER 1980

bile phase contains 30% or more water. If polar solvents are used to extract trace organics from samples, much of the coextracted material will be very polar in nature. When these are injected into an adsorption chromatographic system, they might irreversibly bind to the adsorbent. After a period of time the column performance will decrease and will not be restored upon cleaning. To avoid this, it is recommended that either a pre-column (guard-column) be inserted between the analytical column and the injection port, or that a classical column cleanup be carried out. The author prefers the latter since in complex samples some type of column cleanup is normally required to remove substances that will not necessarily affect chromatography, but which will interfere in the detection. The problem of such irreversible contamination in reversed-phase chromatography is less but nevertheless possible. It must be emphasized that sample cleanup is of the utmost importance both for maintaining the integrity of the chromatographic system and for accurate reliable results. Chemical Derivatization

Chemical derivatization involves altering the chemical structure of the analyte to make it more suitable for determination by chromatographic techniques. In GC it is frequently used to increase volatility or thermal stability so that the analyte will pass through the chromatography system as a symmetrical peak. In both GC and LC, derivatization is used to enhance detector response. In fact, for LC, increased detectability is the main

reason for derivatization. Usually this involves the attachment of a highly absorbing chromophore or fluorophore. Another technique involves the conversion of the analyte to an electrochemically active substance suit­ able for electrochemical detection. There exists an abundance of reagents for derivatization of many classes of compounds. Some have found exten­ sive use while others are limited in their applications. It must be kept in mind that chemi­ cal derivatization should only be used as a last resort to solve detection prob­ lems. The additional reagents re­ quired and the increased sample han­ dling can increase cleanup time and cause increased analytical error. The advantages of using derivatives must be weighed against the problems that may be encountered. It should be made clear, however, that in many in­ stances, derivatization simply has to be used if an analysis is to be made at all. Figure 6 shows the sensitivity gained for the insecticide carbofuran in potato after derivatization. The same sample extract was used in each case except that the fluorescent dansyl (5-dimethylaminonaphthalene-l-sulfonyl-) derivative was prepared for fluorescence detection. The insecti­ cide itself is nonfluorescent. The deri­ vatization improves the detection limit 5-10-fold in the sample. Reactions may be carried out prior to chromatography in a classical man­ ner or in some cases post-column reac­ tions are possible. However, for the latter, special reagent requirements are necessary, and additional equip­ ment is needed. References (1) J. F. Lawrence, C. van Buuren, U. A. Th. Brinkman, and R. W. Frei, J. Agric. Food Chem. 1980,28, 630.

KC 18 :RPTLC A true Ci 8 reversed phase TLC plate. Better resolution. Faster. Higher capacity. Fully silanized. Close correlation to C18 H PLC columns. In inexpensive microslides and standard TLC plate formats. KCia: high efficiency, true reversed phase precoated glass TLC plates. Ο β groups Si-O-Si-C bond­ ed to a special silica gel. Minimum residual silanols. Fast developing: 2 min/cm. Rf deviation only 0.5%. High capacity allows prep TLC (milligrams applied) on the same plate. KC 18 : available with or without fluorescent indicator in microslides, 1 " χ 3" (particularly useful for solvent system development, and HPLC correla­ tions) 5 χ 20 cm and 20 χ 20 cm formats, channeled or unchanneled. Whatman quality. Immediate delivery. KC1B: For high efficiency RPTLC. Write for free samples. Whatman Chemical Separation Division 9 Bridewell Place, Clifton, New Jersey 07014 Tel: (201) 777-4825

Jim Lawrence is a research scientist with the Food Directorate of Cana­ da's Health Protection Branch in Ot­ tawa. Since joining the directorate eight years ago he has been actively involved in the development of ana­ lytical methods for pesticides and re­ lated residues in foods by both gas and liquid chromatography. He re­ ceived a Ph.D. in organic-analytical chemistry from Dalhousie University in Halifax, Nova Scotia, in 1972.

Whatman CIRCLE 226 ON READER SERVICE CARD