Simultaneous Flame Ionization and Absorbance Detection of Volatile

Press, Inc.: West Palm Beach, FL, 1979. (22) Boublik, T., Fried, V., Hala, E., Eds. The Vapour Pressures of Pure Substances;. Elsevier Science Publish...
0 downloads 0 Views 224KB Size
Anal. Chem. 1997, 69, 3465-3470

Simultaneous Flame Ionization and Absorbance Detection of Volatile and Nonvolatile Compounds by Reversed-Phase Liquid Chromatography with a Water Mobile Phase Carsten A. Bruckner, Scott T. Ecker, and Robert E. Synovec*

Center for Process Analytical Chemistry, Department of Chemistry, Box 351700, University of Washington, Seattle, Washington 98195

A flame ionization detector (FID) is used to detect volatile organic compounds that have been separated by wateronly reversed-phase liquid chromatography (WRP-LC). The mobile phase is 100% water at room temperature, without use of organic solvent modifiers. An interface between the LC and detector is presented, whereby a helium stream samples the vapor of volatile components from individual drops of the LC eluent, and the vaporenriched gas stream is sent to the FID. The design of the drop headspace cell is simple because the water-only nature of the LC separation obviates the need to do any organic solvent removal prior to gas phase detection. Despite the absence of organic modifier, hydrophobic compounds can be separated in a reasonable time due to the low phase volume ratio of the WRP-LC columns. The drop headspace interface easily handles LC flows of 1 mL/ min, and, in fact, compound detection limits are improved at faster liquid flow rates. The transfer efficiency of the headspace interface was estimated at 10% for toluene in water at 1 mL/min but varies depending on the volatility of each analyte. The detection system is linear over more than 5 orders of 1-butanol concentration in water and is able to detect sub-ppb amounts of o-xylene and other aromatic compounds in water. In order to analyze volatile and nonvolatile analytes simultaneously, the FID is coupled in series to a WRP-LC system with UV absorbance detection. WRP-LC improves UV absorbance detection limits because the absence of organic modifier allows the detector to be operated in the short-wavelength UV region, where analytes generally have significantly larger molar absorptivities. The selectivity the headspace interface provides for flame ionization detection of volatiles is demonstrated with a separation of 1-butanol, 1,1,2trichloroethane (TCE), and chlorobenzene in a mixture of benzoic acid in water. Despite coelution of butanol and TCE with the benzoate anion, the nonvolatile benzoate anion does not appear in the FID signal, allowing the analytes of interest to be readily detected. The complementary selectivity of UV-visible absorbance detection and this implementation of flame ionization detection allows for the analysis of volatile and nonvolatile components of complex samples using WRP-LC without the requirement that all the components of interest be fully resolved, thus simplifying the sample preparation and S0003-2700(97)00114-5 CCC: $14.00

© 1997 American Chemical Society

chromatographic requirements. This instrument should be applicable to routine automated water monitoring, in which repetitive injection of water samples onto a gas chromatograph is not recommended. Predominantly, analysis of hydrophobic compounds with reversed-phase liquid chromatography (RP-LC) uses stationary phases requiring 50%-100% of an organic solvent in the mobile phase. The need for a high percentage of organic solvent in the mobile phase depends on the nature of the sample and stationary phase, often a highly porous silica derivatized with octadecylsilane, i.e., C18. The large surface area of porous silica translates into a large amount of stationary phase, resulting in a high phase volume ratio (volume of stationary phase to mobile phase). To elute hydrophobic analytes in a reasonable time, the mobile phase strength must be increased by adding a large amount of organic solvent modifier. There is an interest in reducing this hazardous organic waste generated during the separation process. Recently, we reported that RP-LC separations of hydrophobic compounds are possible with a mobile phase composed of only water.1 Performing RP-LC with a water mobile phase has been termed WRP-LC, for water-only reversed-phase liquid chromatography. The stationary phase for WRP-LC is based on a nonporous silica substrate with either an absorbed or a bonded hydrophobic material. More robust bonded-phase WRP-LC columns have also been recently developed and applied in the analysis of components extracted from sand with subcritical water,2 thus eliminating the need for hazardous organic solvents in the analysis procedure altogether. WRP-LC is of interest from a detection and quantitation standpoint as well, providing benefits for chemical analysis. Because of the low UV cutoff for water, a UV absorbance detector can be used in the 200 nm region, where detector sensitivity to compounds is improved. In addition, WRP-LC columns should offer an improved method of interfacing LC systems to a flame ionization detector (FID), a common detector for gas chromatography (GC). Work has been done in recent years to use GC-type detectors for LC, in order to take advantage of their improved sensitivity and selectivity.3 Much of the research centers around improving direct introduction systems, which nebulize or evapo(1) Foster, M. D.; Synovec, R. E. Anal. Chem. 1996, 68, 2838-2844. (2) Young, T. E.; Ecker, S. T.; Synovec, R. E.; Hawley, N. T.; Lomber, J. P.; Wai, C. M. Talanta, in press. (3) Kientz, C. E.; De Jong, G. J.; Brinkman, U. A. T. J. Chromatogr. 1991, 550, 461-494.

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997 3465

rate the liquid stream near the detection element. Total eluent introduction schemes are rarely used with general response detectors such as an FID,4 because the analyte signal is obscured by the organic modifier in the LC eluent. As a result, elementspecific detectors such as the electron capture detector,5-7 the thermionic detector,8-10 and the flame photometric detector11 have seen the most use for LC. Previously, in order to utilize the FID as an LC detector, the organic modifier had to be removed prior to analyte detection. Existing systems achieve this by depositing the eluent stream on, for example, a moving wire or disk.12,13 Moderate heating evaporates the solvent as the deposited eluent is transferred to the detection zone. High boiling point compounds are then either transported directly into the flame or pyrolyzed prior to flame ionization detection. A drawback of these systems is that the solvent evaporation step also removes volatile analytes. Because WRP-LC uses a water mobile phase, we turn our attention to the development of an interface that relies on repeated water drop formation. There is recent interest in developing analytical methodologies based on drop formation. The small volume and reproducible geometry of drops makes drop use feasible for various detection schemes. We have dynamically measured low concentrations of surface-active species in water by optically determining the maximum size of growing drops.14,15 In addition, by measuring the pressure inside growing drops containing either water or surfactants in water, a combination of surface tension and surface adhesion information was obtained and used to characterize complex samples.16 Drops have been used as windowless optical cells for photometric detection.17 Due to the renewable nature of the drop surface, drops have also been used as a gas sampling interface, becoming enriched by various permanent gases as a gas stream flowed by the drops.18 In this article, we report the coupling of a WRP-LC column to an FID, using a novel interface comprised of a small-volume dynamic headspace sampler based on eluent drop formation. As eluent from a conventional bore LC column forms drops at a capillary tip, helium flowing past the drop is enriched with the volatile components in each successive drop. The formation of drops, as opposed to a stream of liquid, increases eluent residence time in the small-volume interface in order to ensure significant transfer of volatile species to the gas phase. The analyte-enriched helium stream is sent to an FID for detection. Since the mobile (4) Krejci, M.; Tesarik, K.; Rusek, M.; Pajurek, J. J. Chromatogr. 1981, 218, 167-178. (5) de Kok, A.; Geerdink, R. B.; Brinkman, U. A. T. J. Chromatogr. 1982, 252, 101. (6) Maris, F. A.; Nijenhuis, M.; Frei, R. W.; de Jong, G. J.; Brinkman, U. A. T. Chromatographia 1986, 22, 235. (7) Zegers, B. N.; de Brouwer, J. F. C.; Poppema, A.; Lingeman, H.; Brinkman, U. A. T. Anal. Chim. Acta 1995, 304, 47-56. (8) Gluckman, J. C.; Novotny, M. J. Chromatogr. 1984, 314, 103-110. (9) Kientz, C. E.; Verweij, A.; de Jong, G. J.; Brinkman, U. A. T. J. Chromatogr. 1992, 626, 59-69. (10) Kientz, C. E.; Verweij, A.; de Jong, G. J.; Brinkman, U. A. T. J. Chromatogr. 1992, 626, 71-80. (11) Karnicky, J. F.; Zitelli, L. T.; van der Wal, S. Anal. Chem. 1987, 59, 327333. (12) Veening, H.; Tock, P. P. H.; Kraak, J. C.; Poppe, H. J. Chromatogr. 1986, 352, 345-350. (13) Dixon, J. B. U.S. Patent 4215090, 1980. (14) Lima, L. R., III; Synovec, R. E. J. Chromatogr., A 1995, 691, 195-204. (15) Young, T. E.; Synovec, R. E. Talanta 1996, 43, 889-899. (16) Olson, N. A.; Synovec, R. E.; Bond, W. B.; Alloway, D. M.; Skogerboe, K. J. Anal. Chem. 1997, 69, 3496-3505. (17) Liu, H.; Dasgupta, P. K. Trends Anal. Chem. 1996, 15, 468-475. (18) Liu, S.; Dasgupta, P. K. Anal. Chem. 1995, 67, 2042-2049.

3466

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

phase has no organic modifier, the interface design is simpler than those used in existing LC-FID systems. The current configuration detects the volatile fraction of an LC sample, a capability lacking in existing systems. Additionally, simultaneous UV-visible absorbance detection can provide information about the nonvolatile fraction of the sample. In this respect, the system described here provides an added capability over GC, which is limited to analysis of volatile species. This report discusses the design and investigates the performance of the drop headspace interface. The effect of liquid flow rate on the time-dependent analyte transfer to the gas phase is examined. A liquid flow rate is chosen to be compatible with conventional bore LC columns, as well as to enhance detection limits for volatile compounds. The periodicity that continuously forming drops introduce to the signal is damped with a small-volume gas mixing chamber prior to detection by FID. The linear dynamic range of the volatile organic detection system is reported. Test analytes are separated on a bonded-phase WRP-LC column. The separated analytes are monitored with UV absorbance detection, and the volatile components are sent via the drop headspace interface to the FID. Comparison of the detection limits and selectivities of the FID and UV absorbance signals will show the benefits of flame ionization detection for LC, as well as the advantages of using both detectors in series. EXPERIMENTAL SECTION Materials. Test analytes used in this study were of analytical grade (J. T. Baker, Phillipsburg NJ, and Aldrich, Milwaukee WI). Instruments. Distilled deionized water as the mobile phase was supplied using a syringe pump (LC2600, Isco, Lincoln, NE). Six ports of a 10-port electrically actuated valve were used to inject samples (Model EC10W, Valco Instruments, Houston, TX). UV absorbance detection (Model V4, Isco) was performed using a 3.5 µL cell with a 0.5 cm path length. Detection was at either 200 or 260 nm, as specified. Flame ionization detection of headspace-enriched helium was performed using a detector within a gas chromatograph (Model 3600cx, Varian Analytical, Sugarland, TX). The FID conditions were as follows: air flow at 300 mL/ min, hydrogen flow at 30 mL/min, and helium carrier at 35 mL/ min. Software written in-house controlled a data acquisition board (Model AT-MIO-16XE-50, National Instruments, Austin, TX) to acquire both the UV absorbance and the FID data. The drop headspace cell is constructed from 1 and 10 mL disposable syringe barrels epoxied end-to-end, mounted vertically with the smaller syringe in the upper position (see Figure 1). A short section of 0.040 in. PEEK tubing (Upchurch, Oak Harbor, WA) connects the upper end of the smaller syringe to a 1/16 in. stainless steel Swagelock “T”. Deactivated silica tubing (J+W, Folsom, CA) with 320 µm i.d. is inserted through the “T” and the PEEK tubing into the upper syringe so that the silica tubing terminates about 0.5 cm from the top of the upper syringe. It is through the silica tubing that the liquid stream to be analyzed is delivered to the cell. Drops form at the tip of the silica tubing and fall unobstructed into the lower, larger syringe, which acts as a waste liquid reservoir. A narrow upper syringe barrel was chosen to increase the linear flow of helium past the growing drop, which enhances analyte evaporation. The bottom end of the lower syringe is sealed with its plunger in the fully extended position and is removable for easy access to the interior. Helium, controlled with a Nupro fine metering valve (Swagelock, distributed by Supelco, Bellefonte, PA), enters the cell via a hole drilled

Figure 1. Schematic of WRP-LC system with UV absorbance and flame ionization detection, highlighting the drop headspace interface that transports volatile analytes from the LC eluent to the FID.

into the side of the upper syringe, below the level of the growing drops. A portion of the helium stream flows upward past the drop, picking up the organics evaporating from the drop. The analyteenriched helium then flows through the annular gap between the silica and PEEK tubing to the “T”, and from there through a 95 cm length of 320 µm deactivated fused silica transfer tubing to an FID. The analyte-enriched helium flow to the FID was adjusted to 35 mL/min. The drop headspace cell is designed to sample the headspace of individual drops. To prevent vapor from the fallen drops in the reservoir from traveling to the FID, a second portion of the helium stream flows downward to the waste reservoir, away from the growing drop. The downward flow prevents the upward migration of waste watervapor. This secondary helium stream is vented from the cell by means of a fixed restrictor, a 20 cm length of 250 µm i.d. silica capillary connected to the side of the lower syringe. Helium flow to the vent is 80 mL/min, and the pressure in the cell is estimated at 6 psi above ambient. Waste liquid is continuously drained from the lower syringe. In experiments where a constant liquid flow of 1 mL/min was sent to the headspace cell, a 10 cm length of 0.007 in. PEEK tubing acting as a fixed restrictor was attached to the lower syringe below the water line. When liquid flows were periodically changed, this fixed restrictor was replaced with a variable restriction device, built inhouse. The drop headspace cell was external to the gas chromatograph and was used at ambient temperatures around 20 °C. Experimental Conditions. Variable Flow Rate Study. To study the effect of liquid flow rate on the extraction performance of the headspace cell, a steady-state amount of sample was injected into the water stream leading directly to the drop headspace cell, using a 0.63 mL sample loop. Analyte-enriched helium vapor was routed to the FID. FID Response to Volatile Organic Compounds at Drop. When determining the FID response to organics at the drop, a steadystate concentration of sample was injected in the manner described above. Additionally, the cycling in vapor concentration resulting from the continuously forming drops was damped by adding a gas mixing cavity to the enriched helium stream prior to the FID. The mixing cavity was comprised of a 1/4-3/16 in. brass Swagelock union, the ends sealed with two 7/16 in. GC injector port septa (Septa 77, J+W). Capillary transfer tubing was inserted through both septa, one coming from the drop headspace cell

Figure 2. Effect of liquid flow rate on the evaporation rate of 1 ppm toluene from growing drops at the drop headspace interface, as monitored by the FID: (A) 33, (B) 250, and (C) 1000 µL/min. Four drops are shown at each flow rate. The baseline was zero in all cases.

and the other leading to the FID. The internal volume of the mixing cavity was 0.4 mL, and the cavity was mounted in the GC oven at 60 °C. The cavity volume is specifically set to dampen concentration pulses for a helium carrier flow of 35 mL/min and a liquid flow of 1 mL/min. Liquid Chromatography. An oligomeric phase prepared from (3,3,3-trifluoropropyl)methyldichlorosilane (Gelest, Tullytown, PA) was bonded to nonporous glass beads with a 6.61 µm mean diameter (Powder Technologies, Burnsville, MN), based on procedures detailed elsewhere.2,19,20 The beads were packed in a 25 cm × 4.6 mm i.d. stainless steel column. A 20 µL sample injection loop was used, and the water mobile phase flow was set at 1 mL/min. Separations were first monitored by UV absorbance, and eluent from the UV absorbance detector was directed to the drop headspace cell, where the analyte-enriched vapor was sent to the FID. Because of post-UV absorbance detector transfer tubing volumes, the FID response lagged the UV absorbance response by 4 s. RESULTS AND DISCUSSION The LC-FID interface discussed here operates by sampling the headspace of the LC eluent, which appears as continuously forming drops from a capillary tip. This is a dynamic sampling system not only because helium is flowing past the drop but also because the drop is growing. The changing drop geometry results in a time-varying concentration of evaporated organic analytes in the helium carrier in the vicinity of the drop. The rate of organic transport into the gas phase is dependent on analyte volatility and water solubility. For the drop headspace cell, transport to the gas phase is also a function of the following properties: analyte migration rate to the drop surface, drop surface area exposed to the helium stream, helium linear flow past the drop, cell temperature, and analyte replenishment rate in the drop due to continued drop growth. Figure 2 illustrates the time dependence of the FID response to the headspace of water drops containing 1 ppm toluene for different liquid flow rates. As can be seen in Figure 2A, at a low flow rate of 33 µL/min (14 s/drop), the evaporation rate of toluene quickly matches its replenishment rate in the drop, as can be seen from the apparent steady-state FID response during (19) Rehak, V.; Smolkova, E. J. Chromatogr., A 1980, 191, 71-79. (20) Scott, R. P. W. Silica Gel and Bonded Phases; Wiley: New York, 1993.

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3467

Table 1. Limits of Detection for Selected Volatile Organic Compounds, and Relevant Physical Properties

compound

boiling point, °Ca

vapor pressure at 20 °C, Torrb

solubility in water at 20 °C, wt %c

LOD, ppbd

1-butanol 1,1,2-trichloroethane butanone chlorobenzene toluene ethylbenzene o-xylene

117 114 80 132 111 136 144

4 16 71 9 22 7 5

7.81 0.11 24 0.05 0.074 0.014 0.018

9 3 2 0.9 0.5 0.5 0.5

a From ref 21. b For pure liquids. From ref 22. c From ref 23. d At detector. A 2 s smooth was applied before calculation.

the latter portion of the drop growth cycle. Drop detachment is observed as a sharp decrease in FID response. Figure 2B shows that, at 250 µL/min (1.9 s/drop), the faster toluene transfer into the drop results in an increased evaporation rate, which is seen as an elevated signal at the FID. At this flow rate, toluene is replenished in the drop faster than it is removed by evaporation, resulting in a signal that is dependent on the surface area of the growing drop. As a result, the evaporation rate does not reach steady state prior to drop detachment. At 1000 µL/min (0.46 s/drop), a flow compatible with conventional bore liquid chromatography, even more toluene per second is transferred from each drop, as shown in Figure 2C. The gas mixing volume of the drop headspace cell blurs the FID response from the individual drops, which is more noticeable at faster drop rates. The faster analyte mass transfer to the gas phase at higher flow rates results in a larger FID signal per time, improving detector sensitivity. The profile of the FID signal over the course of drop growth was identical for all analytes tested (those listed in Table 1).21-23 At faster drop rates, the residence time of individual drops is naturally shorter, resulting in incomplete transfer of volatile analytes to the gas phase. Liquid convection aids in analyte transport to the drop surface at higher flow rates, partially counteracting the effect of shorter drop residence times. By integrating the toluene signal for individual drops at different liquid flow rates and extrapolating to zero flow, it was estimated that a 10% transfer of toluene to the helium carrier occurs at 1000 µL/ min. The fraction transferred will be analyte specific. The reproducible geometry of the drop, in conjunction with a controlled helium flow and constant temperature, provides the proper conditions for reproducible transfer of volatile organics to the gas phase. In this study, we are interested in using the drop headspace cell to couple water-only liquid chromatography with gas-phase detection techniques. With this as our goal, the drop headspace sampling system was modified to provide a stream of enriched helium to the FID that does not carry the cyclic signature of the continuously falling drops, as in Figure 2. A gas mixing cavity was added to the system, immediately prior to the FID. The volume of the cavity (0.35 mL) was large enough to dampen the concentration pulses, yet small enough to prevent a measurable (21) Weast, R. C., Ed. CRC Handbook of Chemistry and Physics, 59th ed.; CRC Press, Inc.: West Palm Beach, FL, 1979. (22) Boublik, T., Fried, V., Hala, E., Eds. The Vapour Pressures of Pure Substances; Elsevier Science Publishers: New York, 1973. (23) Stephen, H., Stephen, T., Eds. Solubilities of Inorganic and Organic Compounds: Binary Systems; Pergamon: New York, 1963; Vol. 1.

3468 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

Figure 3. Large injection volume of 33 ppm butanone, directed at 1 mL/min to the drop headspace interface, producing a steady-state concentration at the drop interface at about 30 s: (A) FID response with no gas mixing cavity prior to detection and (B) FID response with a small-volume mixing cavity inserted between the headspace interface and the FID.

amount of band broadening in the liquid chromatography phase of the experiments. The effect of the mixing cavity is shown in Figure 3. A steady-state amount of a test analyte, 33 ppm butanone, was injected to the drop headspace cell at 1 mL/min. Figure 3A shows the FID response without a mixing cavity, and the individual drop signatures are evident. Figure 3B shows an injection of the same sample, but with the mixing cavity added. The influence of individual drops has been damped, while preserving the underlying shape of the analyte concentration profile. As an alternative to the mixing cell, an analog or digital filter could have been used to smooth the data during collection. The ability of the drop headspace cell to linearly transfer organics to the gas phase over a wide concentration range was studied using steady-state injections of serial dilutions of 1-butanol. For this study, liquid injections were performed at 0.67 mL/min. Figure 4 shows a more than 5 orders of magnitude range of FID response to 1-butanol in water. For clarity, the plot uses logarithmic axes. The slope of this data is 1.007, which indicates that, over the concentration range investigated, detector response was linear. The FID linear dynamic range compares favorably with other GC detectors that have been coupled to LC.3 To prevent the introduction of bias, the mobile phase itself must not contribute an FID signal, so the mobile phase delivery system must be free of any organic solvents from previous applications before WRP-LC separations with flame ionization detection are performed. The headspace interface selectively transfers volatile analytes to the FID for detection. Compound limits of detection (LODs) will naturally depend on the efficiency of evaporation to the gas phase, coupled with the FID response to the various analytes. LODs at the detector were calculated from steady-state injections of organic compounds diluted in water that were delivered to the drop headspace cell at 1 mL/min. Figure 5 shows a steady-state injection of 37 ppb (w/w) toluene to the headspace cell. A 2 s smooth was applied to data for this and other compounds when LODs were calculated. Table 1 summarizes LOD findings for a select group of compounds. LODs were calculated by comparing the steady-state signal for analytes to the noise in the baseline. Toluene, for example, is more volatile and less water soluble than

Figure 4. FID response to a range of 1-butanol concentrations in water, evaporated from drops at the headspace interface, and transported using helium to the FID. Drops were formed at 0.67 mL/ min of water. Three points are plotted per concentration, although due to overlap there appear to be fewer points. A slope of 1.007 on a log/log plot indicates linear detector response over more than 5 orders in concentration.

Figure 5. FID response to vapor from a 0.63 mL volume (steady state) of 37 ppb toluene in water flowing at 1 mL/min to the drop headspace interface.

1-butanol and so is preferentially transferred to the gas phase, resulting in a larger FID signal for a given concentration in water. Because the FID is a general detector for organic species, virtually all volatile organic compounds in water can be detected with the FID using the drop headspace detector. The remainder of the data to be shown result from the coupling of an FID, via the drop headspace interface, to a WRP-LC column with UV absorbance detection in series. One advantage of performing WRP-LC is that UV absorbance detection limits are much better than when organic modifiers are used, as is the case with traditional reversed-phase LC. The absence of organic modifiers allows the UV absorbance detector to be operated in the short-wavelength UV region, where analytes generally have significantly larger molar absorptivities. Figure 6 compares different detection modes for a WRP-LC separation of benzene, toluene, and o-xylene. Figure 6A shows a separation of a 20 µL injection of 2 ppm benzene, 4 ppm toluene, and 10 ppm o-xylene (in addition to an impurity at the dead volume) with UV absorbance detection at 260 nm. Conventional LC detection is commonly performed around 260 nm, because organic mobile

Figure 6. Separation of aromatic species in water at 25 °C using WRP-LC, 20 µL injection at 1 mL/min: (A) UV absorbance detection at 260 nm of 2 ppm benzene (b), 4 ppm toluene (t), 10 ppm o-xylene (x), and an impurity (i); (B) UV absorbance detection at 200 nm of 230 ppb benzene, 460 ppb toluene, and 1.1 ppm o-xylene; and (C) FID in series with UV absorbance in (B).

phase modifiers absorb strongly at lower wavelengths, obscuring analyte signals. Analyte sensitivity is greatly improved at 200 nm, and the use of a water mobile phase allows for detection at this wavelength. Figure 6B shows the separation of a 20 µL injection of 230 ppb benzene, 460 ppb toluene, and 1.1 ppm o-xylene with absorbance detection at 200 nm. Note that, although this mixture is about 9 times more dilute than the mixture separated at 260 nm, the compounds are more clearly discerned by the UV absorbance detector due to the higher molar absorptivities for analytes at 200 nm. Flame ionization detection is competitive with UV absorbance detection for volatile aromatic compounds, despite the incomplete transfer of these compounds to the gas phase. Following UV absorbance detection at 200 nm, the eluent stream connects to the drop headspace interface, where volatile compounds are extracted into a helium stream leading to an FID. Figure 6C shows the FID signal of the same separation that was first detected using UV absorbance at 200 nm in Figure 6B. Inspection of Figure 6B,C reveals an improvement in detection limit using the FID for these compounds. The FID shows a clear advantage over UV absorbance for detection of compounds with no UV-active functional groups. Figure 7 shows the UV absorbance and FID detection of a separation of a 20 µL injection of 11 ppm 1-butanol, 3 ppm 1,1,2trichloroethane (TCE), and 3.5 ppm chlorobenzene. The UV absorbance detector at 200 nm has an appreciable response only to chlorobenzene, due to its aromatic group. The butanol absorbance response is hidden in the injection disturbance, and TCE’s absorbance signal is barely above the noise level. In contrast to the 200 nm UV absorbance response, the FID shows a significant response for all these compounds. Comparison of the UV absorbance and FID baselines shows that the FID baseline is less influenced by low-frequency noise, so analysis of samples near the detection limit is easier with an FID. As the drop headspace interface transfers only volatile organics to the gas phase, selective LC detection is useful when the LC Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3469

Figure 7. WRP-LC separation at 25 °C with 20 µL injection at 1 mL/min of (A) UV absorbance detection at 200 nm of 11 ppm 1-butanol (bu), 3 ppm 1,1,2-trichloroethane (tce), and 3.5 ppm chlorobenzene (cb) and (B) FID in series with UV absorbance in (A).

Figure 8. WRP-LC separation at 25 °C with 20 µL injection at 1 mL/min of (A) UV absorbance detection at 200 nm of 7 ppm 1-butanol (bu), 2 ppm 1,1,2-trichloroethane (tce), 2 ppm chlorobenzene (cb), and 0.13% benzoate anion in water and (B) FID in series with UV absorbance in (A).

stream contains, for example, coeluting nonvolatile species. Figure 8 illustrates the advantages of selective FID detection for LC, for an analysis of a 20 µL injection of 7 ppm 1-butanol, 2 ppm TCE, and 2 ppm chlorobenzene, in a 0.13% benzoic acid solution. As the solution is unbuffered, benzoic acid exists in the benzoate (24) Smith, R. M.; Burgess, R. H. Anal. Commun. 1996, 33, 327-329. (25) Miller, D. J.; Hawthorne, S. B. Anal. Chem. 1997, 69, 623-627.

3470 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

anion form. Benzoate coelutes with butanol and TCE, and chlorobenzene elutes in benzoate’s tail. Because the benzoate anion is not volatile, the FID channel easily picks out all the components but benzoate. The example in Figure 8 points the way to the analysis of more complex samples with this dualdetector approach, because the complementary selectivities of the FID and UV detectors can distinguish between compounds that have not been resolved by the chromatography. The investigation into the use of nonporous packing materials and various stationary phases for preparing WRP-LC columns is very much underway,1,2 and this article highlights some of the advantages of using these types of columns. The separations shown in Figures 6-8 show peaks whose efficiency decreases markedly with increasing retention. We believe this is due to the nonuniform stationary phase coverage on the packing material. The retention order in previous work was found to correlate to the analyte octanol-water partition coefficient and is consistent with a reversed-phase mechanism.1 The retention order in Figures 6-8 is consistent with these results. The WRP-LC columns have been used to separate hydrophobic species up to naphthalene at room temperature,2 and this range can be extended by heating the water mobile phase.24,25 These separations represent some of the early work in preparing WRP-LC columns and do not necessarily reflect the achievable limits in separation efficiency. CONCLUSIONS The drop headspace cell is intended to be a simple interface facilitating flame ionization detection of all volatile organic species separated with conventional bore water-only liquid chromatography. Used in conjunction with UV absorbance detection, a chromatographer can now obtain more information from a single separation. This interface can be just as easily used with selective GC detectors, such as the thermionic detector. Direct introduction of water-only LC eluent to the FID will allow for detection of both volatile and nonvolatile species, although smaller bore columns will most likely be required. ACKNOWLEDGMENT We thank the Center for Process Analytical Chemistry (CPAC), a National Science Foundation University/Industry Cooperative Research Center at the University of Washington, for financial support. Received for review January 28, 1997. Accepted June 6, 1997.X AC9701142 X

Abstract published in Advance ACS Abstracts, July 15, 1997.