Improved Solvent Trapping of Volatiles in Supercritical Fluid Extraction

Anal. Chem. 2000, 72, 1268-1274

Improved Solvent Trapping of Volatiles in Supercritical Fluid Extraction by Pressurizing the Collection Vial Mark A. Stone and Larry T. Taylor*

Department of Chemistry, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061-0212

The high flow rates that result after decompression make the trapping of analytes one of the more difficult aspects of a supercritical extraction. By elevation of the pressure on the collection vial, the flow may be reduced and trapping efficiency improved, considerably. The effects of different collection vial pressures were evaluated. The best results were obtained at 25 atm. With the collection vial at 25 atm, the trapping efficiencies of different solvents were then investigated. Also considered were the effects of solvent volume, extraction flow rate, collection solvent geometry, and restrictor temperature. For a 30 min extractionswith the restrictor line at 200 °Csquantitative recoveries were obtained with analytes as volatile as chlorobenzene (∼90%) in 1 mL of solvent. Mild precooling of the collection vessel allowed quantitative recoveries to be obtained down to benzene (94.2%). Trapping of analytes from the effluent of a supercritical extractor is typically done after complete decompression of the supercritical stream. The sharp drop in density causes an increase in volumetric flow rate of the same factor. This makes the trapping of volatile and semivolatile analytes difficult. Of the two options that exist, sorbent trapping and solvent trapping, the former is more difficult, as there are more variables to contend with. In addition to selecting the sorbent, one must select the type and volume of solvent to be used for eluting the analytes from the trap. Different temperatures are often used for trapping and desorption: below ambient and above ambient, respectively. It has been shown that even the solvent used to prerinse the trap can affect results.1 All of this leads to more difficulty in the method development process and requires a supercritical fluid extraction (SFE) system with more hardware, including a pump for the elution solvent as well as both heating and cooling capability. Furthermore, sorbent trapping is less flexible than solvent trapping. It has been shown to be virtually impossible to trap analytes of varying chemical/physical properties on a single sorbent trap,1,2 whereas analysts were able to achieve good results (generally 80% or better) with solvent trapping of the same compounds in seven different solvent systems.3 The difficulty lies (1) Mulcahey, L. J.; Hedrick, J. L.; Taylor, L. T. Anal. Chem. 1991, 63, 22252232. (2) Mulcahey, L. J.; Taylor, L. T. Anal. Chem. 1992, 64, 2352-2358. (3) Thompson, P. G.; Taylor, L. T.; Richter, B. E.; Porter, N. L.; Ezzell, J. L. J. High Resolut. Chromatogr. 1993, 16, 713-716.

1268 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

largely in the need for a solvent which is chromatographically strong for the stationary phase yet which simultaneously has good solubility for the analyte. In addition, irreversible binding to the sorbent can occur in some cases. Polar and acidic compounds are particularly difficult to handle.1 Using inert solids would be expected to remove some of this complexity; however, even with inert materials, the nature of the elution solvent can be important. This has been shown with glass beads4 and polyethylene frits.1 In addition, inert materials are not very efficient trappers. Richards and Campbell reported less than quantitative recoveries of semivolatile organics from artificially spiked soil with a stainless steel trap at -30 to -50 °C. For example, dichlorobenzene isomers were only recovered in the mid-70% range.5 Similarly, it has been shown that decane is only recovered at 75% when trapped on glass beads at temperatures as low as -45 °C.4 Mulcahey et al. reported recoveries of less than 12% for acetophenone and dimethylaniline on stainless steel at 5 °C.2 All of these cases utilized 100% CO2 as the extraction fluid. Sorbent trapping becomes even more problematic when the extraction method employs solvents as modifiers, as they have a tendency to condense in the trap region and elute the analytes from the sorbent. Hence, the conclusion is that solvent trapping is far more simple and flexible than sorbent trapping. It should be mentioned, however, that sorbent trapping may offer a selectivity advantage in that certain compounds can be selectively trapped and/or eluted with a carefully designed method. Unfortunately, solvent trapping suffers from two fundamental problems: it is not as efficient as sorbent trapping for volatile analytes, and evaporation of solvent generally occurs during the extraction, requiring the analyst to continually replace the lost fluid. These issues also make it difficult to work with smaller volumes of solvent, which are desirable from the standpoint of sensitivity, cost, and environmental concerns. Table 1 summarizes the published work where solvent trapping has been achieved successfully with volatile analytes. The first entry in the table represents the only papersas far as we are awareswhere collection vial pressure was considered as a variable. The authors reported that with a collection vial pressure of (4) Ashraf-Khorassani, M.; Houck, R. K.; Levy, J. M. J. Chromatogr. Sci. 1992, 30, 361-366. (5) Richards, M.; Campbell, R. M. LC-GC 1991, 9, 358-364. 10.1021/ac990964c CCC: $19.00

© 2000 American Chemical Society Published on Web 02/04/2000

Table 1. Solvent Trapping of Volatilesa time (min) 30

a

flow (mL/min) 1.5 (e)

solvent/vol (mL)

collect vial temp (°C)

CH2Cl2/7 (2 atm pressure) CH2Cl2/15

-20 NC

30

2-2.4 (l)

various/15

5

30

0.3 (l)

hexane/15

2

15 40

0.5 (l) 0.6 (l)

hexane/5 CH2Cl2/2.5-10

2-5 5

15 10

1 (e) 1 (e)

CH2Cl2/3 CH2Cl2/2.5

NC -20 to -30b

most volatile analytes/ % recoveries benzene/102 C7/102 benzene/90 C8/95 acetophenone/80-95 dimethylaniline/79-91 C9/100 toluene/97 C9/92-93 dichlorobenzene/99.8 dimethyl phthalate/100 C10/“quantitative” C8/90-93

ref 9

3 12 11 13 6

NC ) temperature not controlled. e ) extraction flow rate. l ) liquid flow rate. b Temperature may be somewhat higher during the first 2 min.

2 atm and by maintaining the temperature at -20 °C, they were able to achieve quantitative recoveries of benzene and C7 (linear alkane) in only 7 mL of CH2Cl2. In the absence of any collection vial pressure, volatile analytes such as toluene, acetophenone, C8, and C9 typically require 15 mL of solvent and cooling of the collection vial, for a 30 min extraction (in some cases, the flow rates are slow as well). Cases where smaller volumes are used typically employ shorter extraction times and focus on less volatile analytes. Burford et al. (last line in Table 1) reported ∼90% collection of C8 in only 2.5 mL of CH2Cl2; however, the extraction was only 10 min long and the trap temperature was very cold. The authors reported that, even under these conditions, the recoveries dropped below quantitative if the restrictor temperature was raised above 100 °C.6 In the present work, we demonstrate that, with higher collection vial pressures, quantitative recoveries may be obtained for volatile analytes in smaller volumes of trapping solvent than have been used previously and with less effective solvents, even with restrictor temperatures as high as 200 °C. With this approach, there was no need to replenish the solvent or to provide for external cooling of the trap during the extraction. EXPERIMENTAL METHODS All extractions were performed with a Prepmaster extractor (Isco-Suprex, Lincoln, NE). Except where otherwise indicated, the extractions were 30 min in duration, an extraction flow rate of 1 ( 0.2 mL/min was maintained, and the collection vial pressure was set at 25 atm. Restrictor heating was only used for the data presented in Tables 6 and 7. The flow was controlled by crimping the 1/16 in. stainless steel tube that connects the extraction vessel to the bottom of the collection vessel. The end of the crimped line was then bent so that the part immersed in the trap solvent was pointed downward. The procedure used and the valve diagrams for each step are depicted in Figure 1. It is important to note that the collection vial was pressurized beforesnot durings the dynamic extraction. Extractions were conducted with a 3.5 mL stainless steel vessel (Keystone Scientific, Bellefonte, PA). A 5 mL Keystone extraction vessel, with a 7 mm internal diameter, was used as the collection (6) Burford, M. D.; Hawthorne, S. B.; Miller, D. J. J. Chromatogr. 1992, 609, 321-332.

vial. The frits were removed from both ends of the collection vessel and were replaced with hollowed-out frit rings (Upchurch Scientific, Oak Harbor, WA). Additionally, the narrow opening in the end cap of the extraction vessel was drilled to 1/16 in. This allowed the restrictor line to be inserted into the bottom of the vessel. The drilling was also important to ensure that the collection vial pressure was equivalent to that set on the regulator, as the narrow opening at the top of the vessel would have created a backpressure thus elevating the pressure in the vessel. A 1/16 in. o.d., 0.03 in. i.d. line connected the collection vessel to the backpressure regulator (Process Instruments, Cross Lanes, WV). All other lines had an internal diameter of 0.01 in. For the heated runs, heating tape was wrapped around a 6 in. portion of the line that leads into the bottom of the collection vessel. A gasoline range organics mixture was obtained (Restek, Bellefonte, PA), and a chlorobenzene stock solution was prepared in-house. From these two stocks, a working solution was prepared. The solution used for the first portion of the study (evaluation of collection vial pressure) contained all analytes at 40 000 ng/mL in methanol. For the remainder of the study, a solution was prepared in CH2Cl2 at the equivalent concentration. A 5890 Series II gas chromatograph with a 5972 mass spectrometer (Hewlett-Packard, Wilmington, DE) was used to analyze the extracts. The injector temperature was 310 °C and the mass spectroscopic heated zone (detector B) was set at 280 °C. The head pressure was held at 18 psi for 1 min, ramped down to 5 psi, and kept in a constant-flow mode for the remainder of the run. The oven temperature was held at 35 °C for 4 min, ramped at 5 °C/min to 80 °C, and then ramped at 25 °C/min to a final temperature of 220 °C. A 15 m Rtx-5 column was used with a 0.25 mm i.d. and 0.25 µm film (Restek, Bellefonte, PA). The column contained an attached 10 m length of guard column. A 0.5 µL of solution was injected at a split ratio of 1:4. To normalize for losses which were expected to occur with some of the more volatile analytes, as well as for the effects the expansion volumes of the different solvents might have on quantitation, a calibration standard was prepared each day from the same stocks that were used for spiking and in the solvent to be used for collection. All volumes reported are final volumes. An additional volume of trap solvent was added at the beginning of each run such that Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1269

Figure 1. Extraction method and valve diagrams.

the desired volume of solvent resulted at the end of the extraction. It was found that most of the solvent loss occurred in step 1 (before the dynamic extraction began). Therefore, the reported volumes should represent, reasonably well, the volume at which the actual extraction occurred. Each run was begun by filling the extraction vessel with sand, spiking with 100 µL of the stock solution, and placing approximately 1 mL of solvent in the collection vessel. At the end of the run, the collection vessel was depressurized by carefully loosening the nut and ferrule at the top of the vial. Forty microliters of dichlorobenzene internal standard solution was then injected into the collection vessel. The contents of the vessel were mixed by expelling air into the liquid several times with a disposable pipet, and a portion of the solvent was pipetted into an autosampler vial for GC analysis. For the runs using 2 mL of 1270

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

collection solvent, the volumes of spiking and internal standard solutions were doubled. RESULTS AND DISCUSSION Theoretical Background. Trapping in SFE is a gas chromatographic process. We stipulate that the breakthrough time (tb) of analytes in the trap is inversely proportional to flow rate.

tb ∝ 1/flow rate (There is no equation, as far as we are aware, that gives breakthrough time as a function of flow rate for a liquid trap. However, for packed and open tubular traps, breakthrough time is inversely proportional to flow rate. We infer that the same proportionality would apply to liquid traps.) Anything that increases breakthrough time makes trapping more efficient.

Table 2. Flow as a Function of Collection Vial Pressurea gauge pressure (atm)

decompressed flow (mL/min)

open vessel 5 15 25

397 64.1 22.1 12.4

a Calculations were based on the absolute pressures that result from the indicated gauge pressures.

Figure 3. Retention volume as a function of temperature.

Figure 2. Solubility as a function of density.

The effect of collection vial pressure on the decompressed flow rate is shown in Table 2. The decompressed flows were calculated for an extraction temperature of 100 °C, a pressure of 400 atm, and a flow rate through the extraction vessel of 1 mL/min. The temperature of the collection vessel was estimated to be 10 °C, which is reasonable, as the vessel was cool to the touch even with high restrictor line temperatures. The numbers show that a collection vial pressure of 25 atm results in a decompressed flow rate that is reduced by a factor of 32 compared with that in an unpressurized vessel. By the above equation, this would imply an equivalent increase in breakthrough time. It is important to consider whether the increased trap pressure may have an effect on the solubility of the analytes. Enhanced solubility in the extraction fluidsin the trapping regionswould make trapping less efficient. The diagram in Figure 2, adapted from Brogle,7 demonstrates the increase in solubility (measured as the mole fraction of naphthalene in solution) with an increase in density. The vertical lines represent the points where the fluid passes from the gas to the supercritical state, for the two curves. Although there are not many data points in the gaseous part of the diagram, it is clear that the curves become quite flat in this region. Hence, while solubility is a strong function of density in the supercritical region, the relationship is very weak for gases. This suggests that increasing the pressure in the collection vial will have only a marginal effect on analyte solubility, so long as the fluid remains in the gaseous state. A maximum pressure of 25 atm was chosen for this study because the fluid would remain (7) Brogle, H. Chem. Ind. 1982, (June 19), 385-390.

gaseous even at temperatures down to -5 °C. It was thought unlikely that the temperature would fall below -5 °C during the extraction. When trapping is performed with liquid solvents, there are two other factors, in addition to the flow rate effect, that contribute to improved trapping at elevated collection vial pressures. First, increased pressure causes the gas bubbles in the collection solvent to become smaller. This facilitates the mass transfer of analyte molecules into the liquid phase simply because smaller bubbles mean the analytes have less distance to travel, on average, to reach the liquid. Second, the pressure on the collection solvent is expected to shift the equilibrium such that a greater fraction of analytes associate with the liquid phase at any given time. The only disadvantage of pressurizing the collection solvent is that, because there is less of a pressure drop when the fluid decompresses, the Joule-Thompson cooling effect is smaller. In the absence of external cooling, the collection vial temperature is higher than it would be classically. This should be kept in mind, as temperature is an important variable in gas chromatography/ trapping. The diagram in Figure 3 plots gas chromatographic retention volume (Vr) against temperature (retention volume is directly proportional to retention time at constant flow). The diagram is based on data obtained empirically by Harris and Habgood8 and shows that retention volumes (or times) increase as temperature decreases and that this effect becomes very pronounced at lower temperatures. Given the “exponential” nature of the temperature curves, it may seem that temperature would be a more important variable than flow rate for volatile analytes. However, because the decompressed flow rates are so high in SFE, it was expected that the control of flow rate would, in fact, be more important. Our data support this contention. Trapping Data at Variable Pressures. In what follows, “quantitative” is defined as having a recovery of g90%. The first experiments were concerned with demonstrating that recoveries could be improved by pressurizing the collection vial. Data points were obtained with collection vial pressures of 5, 15, and 25 atm, as well as with an open collection vessel, at dynamic extraction times of 15, 30, and 60 min. Two milliliter quantities of ethanol were used for collection. Each data point is the average of three replicate extractions. The RSD’s are e4% in almost all cases, but they rise into the mid-20% range in three cases where the recoveries are quite low. In Figure 4 the data are plotted for the three most volatile analytes as well as for naphthalene. A clear (8) Harris, W. E.; Habgood, H. W. Talanta 1964, 11, 115-128.


1271

Figure 4. Recovery as a function of collection vial pressue. Table 3. Evaluation of Different Trap Solvents and Solvent Volumes: Percent Recovery (RSD)a 1 mL of solvent

2 mL of solvent

compound

pentane 0.24 cP

CH2Cl2 0.44 cP

acetone 0.32 cP

MeOH 0.59 cP

EtOH 1.36 cP

IPA 2.50 cP

pentane 0.24 cP

EtOH 1.36 cP

isooctane benzene toluene chlorobenzene ethylbenzene m-xylene trimethylbenzene naphthalene

91.6 (1.6) 95.8 (1.6) 92.2 (2.7) 94.0 (3.4) 95.0 (3.8) 95.3 (3.3) 97.6 (4.6) 98.0 (4.3)

82.6 (0.60) 102 (0.73) 92.3 (0.87) 93.7 (0.56) 94.3 (0.84) 94.9 (0.86) 96.3 (1.2) 97.9 (0.74)

79.9 (0.51) 101 (2.2) 95.4 (0.25) 92.4 (0.54) 92.4 (0.41) 91.8 (1.0) 93.0 (0.33) 96.2 (0.70)

63.4 (10) 86.6 (8.4) 86.3 (5.5) 93.5 (4.0) 92.2 (3.7) 92.0 (3.3) 95.8 (2.2) 98.4 (1.3)

69.5 (6.5) 86.3 (3.7) 87.1 (1.8) 93.0 (1.8) 92.7 (1.7) 92.5 (1.9) 97.4 (1.0) 102 (1.1)

74.4 (5.4) 85.0 (4.4) 88.0 (3.6) 93.0 (2.6) 93.0 (2.5) 93.7 (2.9) 96.2 (0.81) 97.7 (0.03)

90.9 (0.50) 97.2 (0.024) 91.8 (0.31) 93.1 (0.38) 94.7 (0) 94.7 (0.37) 97.0 (0.11) 99.4 (0.32)

78.7 (0.45) 98.1 (0.11) 92.3 (0.34) 97.1 (0.10) 96.5 (0.055) 96.2 (0.050) 98.1 (0.36) 99.9 (0.81)

a

Viscosity data correspond to a temperature of 20 °C.

increase in recovery with increasing collection vial pressure is observed. There does appear to be an effect of diminishing returns. For example, a fairly small difference was observed between values 1272


at 15 and 25 atm. This is logical, as it matches the pattern of decreasing flow rates in Table 2. This also explains why some success can be realized with trap pressures of 2 atm.9

Table 4. Evaluation of the Effect of Collection Solvent Geometry: Percent Recovery (RSD) 1 mL of ethanol

2 mL of ethanol

compound

without inserts (26.0 mm)

with inserts (50.9 mm)

without inserts (52.0 mm)

with inserts (76.7 mm)


69.5 (6.5) 86.3 (3.7) 87.1 (1.8) 93.0 (1.8) 92.7 (1.7) 92.5 (1.9) 97.4 (1.0) 102 (1.1)

72.2 (1.7) 86.4 (2.6) 87.7 (2.0) 94.5 (0.82) 94.5 (0.41) 95.3 (0.41) 97.9 (2.3) 99.9 (4.1)

78.7 (0.45) 98.1 (0.11) 92.3 (0.34) 97.1 (0.10) 96.5 (0.055) 96.2 (0.0050) 98.1 (0.36) 99.9 (0.81)

74.3 (2.0) 92.6 (1.2) 90.8 (0.25) 95.3 (0.85) 94.5 (0.92) 94.4 (0.79) 97.7 (1.2) 101 (0.82)

It may be observed that some of the recovery values in Figure 4 are slightly elevated compared to corresponding values in subsequent parts of the study. This is believed to be due to the fact that a methanol stock solution was used for spiking with these runs. The quantity of methanol injected may have been such that its solubility in CO2 was exceeded. This would have impeded flow through the system in the early part of the extraction, slowing the transfer of the analytes to the collection vessel and resulting in slightly elevated recoveries. In all subsequent work, a methylene chloride stock was used and two replicate extractions were run. Effect of Trap Solvents and Solvent Volumes. The next set of analyses were focused on evaluating the effects of different trap solvents and solvent volumes. The data are presented in Table 3. The viscosities of each solvent are also indicated in the table in units of centipoises (cP). All viscosities correspond to a temperature of 20 °C. These, and all subsequent data, are in the form of percent recovery, with RSDs in parentheses. As one would anticipate, pentane and methylene chloride showed better recoveries, in almost all cases, than acetone or alcohol solvents. With pentane, even isooctane was quantitatively trapped in 1 mL of solvent. However, the difference in recoveries obtained with different solvents was less pronounced than expected. That recoveries should increase with larger volumes of trap solvent is intuitive and has been demonstrated in the literature. For example, the recovery of naphthalene was observed in 5 and 15 mL of acetone and found to increase from 70 to 86%.10 Similarly, it was shown that, for benzene, recovery in 8 vs 15 mL of CH2Cl2 increased from 60 to 90%.9 There are only two studies where successful trapping of analytes as volatile as C8 or benzene have been reported. One required either 7 mL of solvent with 2 atm of pressure and cooling to - 20 °C or 15 mL of solvent without

pressure or cooling.9 The second study used only 2.5 mL of CH2Cl2; however, the extraction was only 10 min long and the temperature was in the -20 to -30 °C range.6 The fact that quantitative recoveries were obtained, in the present work, for volatile, nonpolar analytes, in 1 or 2 mL of solventsincluding less efficient alcohol solventssis a testament to the improvement in trapping efficiency gained by virtue of elevating the collection vial pressure. It is also interesting to note that there appears to be no correlation between trapping efficiency and viscosity. This is particularly noteworthy with the alcohol solvents, as there are large differences in viscosity but fairly small differences in recovery. Effect of Collection Solvent Geometry. The effect of collection solvent geometry was evaluated by comparing the results obtained with a given volume of solvent vs the equivalent volume with six glass inserts, 2 mm i.d. by 5 cm length, added to the collection vial. The data are presented in Table 4 along with the calculated height of the solvent layer for each case. There was virtually no difference observed with either 1 or 2 mL of trap solvent. In the 2 mL data, the benzene result seems notably higher without inserts; however, this is not of concern, as some erratic behavior was observed for benzene throughout the study. This is believed to result from some type of azeotrope effect. Previous work with 2.5 mL of an unspecified solvent demonstrated a difference in recovery for carvone (82.6 vs 93.3%) when the height of the layer was 8 vs 22 mm.11 Although this difference is fairly small, it is significant, considering their extraction time was only 10 min. The fact that we observed no effect with more volatile analytes, in only 1 mL of ethanol, is further evidence that improvement in trapping efficiency may be obtained by this approach. Effect of Flow Rate. Data demonstrating the effect of flow rate are presented in Table 5. With 2 mL of collection solvent, the only observable effect was the somewhat elevated isooctane result at a flow of 0.4 mL/min. When only 1 mL of solvent was used for collection, a discernible trend was observed for isooctane, benzene, and toluene. A very weak trend can be seen with a few other analytes but the differences are quite small. Although few studies have focused on the effect of flow rates on the solvent trapping of volatiles, substantial effects have been reported. Langenfeld et al.11 showed a drop in recovery of R-pinene when the flow was raised from 0.3 to 1.2 mL/min (94.5 to 76.6%). They used a 30 min extraction time and 5 mL of solvent. AshrafKhorassani et al. stated that decreasing the flow from 2 to 0.9 mL/minswhile doubling the length of the extraction to 1 hsled

Table 5. Evaluation of the Effect of Extraction Flow Rate: Percent Recovery (RSD) 1 mL of ethanol

2 mL of ethanol

compound

0.4 mL/min

1mL/min

1.5 mL/min

0.4 mL/min

1 mL/min

1.5 mL/min


77.8 (6.2) 92.0 (4.0) 89.6 (2.9) 92.0 (1.7) 93.4 (1.7) 93.1 (1.9) 95.3 (1.1) 98.9 (0.53)

69.5 (6.5) 86.3 (3.7) 87.1 (1.8) 93.0 (1.8) 92.7 (1.7) 92.5 (1.9) 97.4 (1.0) 102 (1.1)

64.7 (1.7) 86.1 (1.4) 83.4 (1.9) 91.7 (2.4) 90.9 (2.5) 91.0 (2.4) 95.0 (2.2) 99.8 (1.3)

85.6 (3.2) 92.4 (3.1) 93.4 (2.4) 94.8 (1.6) 96.6 (1.5) 96.7 (1.3) 99.6 (0.81) 103 (0.94)

78.7 (0.45) 98.1 (0.11) 92.3 (0.34) 97.1 (0.10) 96.5 (0.055) 96.2 (0.050) 98.1 (0.36) 99.9 (0.81)

76.0 (0.70) 99.5 (2.7) 89.8 (0.76) 95.4 (0.55) 95.6 (1.2) 94.8 (1.7) 97.9 (1.6) 102 (0.91)


1273

Table 6. Evaluation of the Effect of Restrictor Line Heating with Pentane as the Trap Solvent: Percent Recovery (RSD) 1 mL of pentane compound

100 °C

150 °C

200 °C

2 mL of pentane 200 °C


84.4 (0.74) 93.6 (0.21) 87.6 (0.72) 90.7 (0.54) 91.6 (0.70) 91.2 (0.57) 93.6 (0.80) 95.4 (0.58)

85.5 (1.1) 93.3 (1.6) 88.8 (0.81) 91.7 (0.96) 92.9 (0.98) 92.5 (1.2) 95.7 (0.49) 97.2 (0.76)

83.3 (3.7) 87.0 (7.4) 85.8 (3.6) 89.2 (3.1) 91.1 (2.6) 91.0 (2.2) 94.6 (1.3) 97.0 (0.70)

92.8 (0.019) 100 (0.72) 93.4 (0.011) 95.5 (0.033) 95.7 (0.035) 95.4 (0.10) 96.0 (0.12) 98.9 (0.35)

Table 7. Evaluation of the Effect of Restrictor Line Heating with Ethanol as the Trap Solvent: Percent Recovery (RSD) 1 mL of 1 mL of ethanol compound

100 °C

150 °C

200 °C

2 mL of ethanol 200 °C


62.5 (0.99) 82.0 (1.4) 82.1 (1.6) 89.1 (2.8) 87.5 (2.9) 87.7 (2.6) 90.7 (3.4)

59.3 (4.7) 78.8 (3.9) 80.8 (3.0) 89.1 (2.2) 88.0 (2.1) 88.3 (1.8) 92.4 (1.2)

64.9 (0.52) 85.9 (0.062) 86.1 (1.2) 91.4 (0.82) 90.8 (0.98) 90.6 (0.73) 93.1 (1.6)

72.6 (2.6) 92.6 (2.5) 90.9 (0.78) 96.4 (0.77) 96.6 (1.2) 96.2 (1.2) 100 (0.67)

93.2 (4.7) 96.0 (1.1) 95.0 (1.4)

101 (0.97)

ethanol with 15 min precool 200 °C 68.2 (1.2) 94.2 (1.4) 88.9 (0.65) 95.2 (1.9) 94.9 (1.6) 94.6 (2.3) 97.7 (2.6) 100 (5.1)

to a drastic increase in the recovery of naphthalene (from 90%) in 2 mL of CH2Cl2.4 Effect of Restrictor Line Heating. Finally, the effect of restrictor line heating was evaluated (Tables 6 and 7). It was found that heating of the restrictor did cause a small decrease in the recoveries of the volatile analytes in 1 mL of solvent. The effect was less noticeable when collection was performed in 2 mL, particularly for pentane. Surprisingly, there was very little difference in the results with line temperatures between 100 and 200 °C. It is interesting to note that, even with classical solvent trapping, the effect of restrictor heating is not as drastic as one might expect. It has been demonstrated that, at temperatures of e200 °C, recoveries are not affected for naphthalene (in 15 mL of acetone for 45 min),10 C9 (in 5 mL of hexane for 30 min),12 or C8 (in 2.5 mL of CH2Cl2 for 10 min).6 Although we did not have cryocooling capability on our system, it was possible to get a preliminary idea of the gain that could be (9) Yang, Y.; Hawthorne, S. B.; Miller, D. J. J. Chromatogr. 1995 , 699, 265276. (10) Reindl, S.; Hofler, F. Anal. Chem. 1994, 66, 1808-1816. (11) Langenfeld, J. J.; Burford, M. D.; Hawthorne, S. B.; Miller, D. J. J. Chromatogr. 1992, 594, 297-307. (12) Porter, N. L.; Rynaski, A. F.; Campbell, E. R.; Saunders, M.; Richter, B. E.; Swanson, J. T.; Nielsen, R. B.; Murphy, B. J. J. Chromatogr. Sci. 1992, 30, 367-373.

1274


realized, by allowing the system to remain in the trap-pressurization stage for 15 min prior to the extraction. In this stage, extraction fluid flows though the collection vialsbut not the extraction vesselsand some cooling of the collection solvent occurs. The result was a noticeable increase in recoveries such that benzene was quantitative in 1 mL of ethanol with the restrictor at 200 °C (Table 7). CONCLUSION It has been shown that solvent trapping can be improved considerably by maintaining substantial pressures on the collection vessel. For a 30 min extraction, at a 1 mL/min flow rate, with the restrictor line at 200 °C, recoveries were quantitative down to chlorobenzene in only 1 mL of ethanol, with benzene and toluene recoveries just below quantitative. Keeping the system in the trap pressurization stage for 15 min, which allowed cooling of the collection solvent prior to extraction, extended the range of analytes recovered quantitatively down to benzene. With pentane as the collection solvent, or with 2 mL of ethanol, similar results were obtained without precooling. The Joule-Thompson effect was less pronounced in this study, as there was less of a pressure change than would occur with classical solvent trapping. As a result, the collection vial temperatures were somewhat higher than would normally be the case. Given the importance of trap temperature, it is encouraging that good results were obtained despite this fact. If a system were designed with the capability to cryocool the collection vials, it is predicted that quantitative recoveries would be achieved with even smaller volumes and/or with more volatile analytes. However, it seems that, for the vast majority of analytes, quantitative recoveries could be obtained without any external cooling mechanism. The effects of different solvents, solvent volumes, restrictor line temperatures, and flow rates on volatile compound recoveries were generally less than have been observed with unpressurized trapping. No effect was observed for collection solvent geometry. By making solvent trapping more efficient and less sensitive to the above-mentioned variables, we expected that the need for sorbent trapping would be greatly reduced. In addition, when trapping was performed at these pressures, very little evaporation of the collection solvent was observed. These factors have the potential to greatly simplify supercritical fluid extractions and make the technique far more user friendly. ACKNOWLEDGMENT We thank Keystone Scientific for donation of the high-pressure vessels used in this study. In addition, we thank Isco-Suprex Inc. for the supercritical extractor as well as Air Products and Chemicals Inc. for providing the carbon dioxide. Received for review August 24, 1999. Accepted December 15, 1999. AC990964C

Improved Solvent Trapping of Volatiles in Supercritical Fluid Extraction

Recommend Documents