Anal. Chem. 1995, 67, 919-923
Selective Extraction of Mercury with Ionizable Crown Ethers in Supercritical Carbon Dioxide Shaofen Wang, Sadik Elshani, and C. M. Wai* Department of Chemistly, University of Idaho, Moscow, Idaho 83843
The solubility of a tert-butyl-substituteddibenzobistriazolo-crownether in methanol (5 mol %) modi6ed COZhas been determined to be in the order of 1 x M at 60 "Cand 200 atm. Using this proton-ionizablecrown ether as an extractant in methanol (5%)m o f i e d COz, quantitative extraction of h m sand and cellulose-basedfilter papers can be achieved at 60 "C and 200 atm if a small amount of water is present in the solid matrix. Other divalent metal ions including Cd2+, Co2+, Mn2+,Ni2+, Pb2+,and Zn2+are virtually not extractable under these conditions. Gold (Au3+)can be extracted by the macrocyclic ligand with a lower efficiencythan H@+. Extraction of Hg2+ from aqueous samples can also be achieved using methanol-modified COS containing the macrocyclic ligand. Formation of extractable metal chelates with the ligand at the pH defined by the C 0 2 - H 2 0 system appears to be an important factor determining the extraction dciencies of the metal ions by this in situ chelation-supercritical fluid extraction technique.
w+
Supercritical fluid extraction (SFE) of metal species from solid and liquid materials has been the subject of several recent One approach for extracting metal ions in supercritical COZis to convert the charged metal species into neutral metal chelates by use of a suitable ligand dissolved in the fluid phase. This in situ chelation-SFE method has been shown to exhibit high efficiencies for extracting metal ions from solid and liquid materials when a fluorinated chelating agent such as lithium bis(trifluoroethyl)dithiocarbamate (LiFDDC) or a fluorinated pdiketone is used as an extractant in supercritical COz.1,214 It is conceivable that selective extraction of metal ions in supercritical fluids may be achieved if the chelating agent possesses an ion recognition capability. Dithiocarbamate and P-diketones are not selective chelating agents because they complex with a large number of metals and non-metals. These chelating agents are also difficult to regenerate because of their instability in acid solutions. Crown ethers are a class of selective ligands that form stable complexes with metal ions based primarily on the ionic radius-cavity size compatibility concept.6 Modification of crown structure by attaching negatively (1) Laintz, K. E.; Wai, C. M.; Yonker, C. R; Smith, R D. Anal. Chem. 1992, 64,2875. (2) Wai, C. M.; Lin, Y.; Brauer, R D.; Wang, S.; Beckert, S. F. Talanta 1993, 40, 1325. (3) Liu, Y.; Lopez-Avila, V.; Alcaraz, M.; Beckert, W. F. ]. High Resolut. Chromatgr. 1993,16,106. (4) Lin, Y.; Brauer, R. D.; Laintz, K E.; Wai, C. M. Anal. Chem. 1993,65, 2549. (5) Lin, Y.; Wai, C. M.; Jean, F. M.; Brauer, R. D. Enuiron. Sci. Technol. 1994, 28. 1190. 0003-2700/95/0367-0919$9.00/0 0 1995 American Chemical Society
charged functional groups to a macrocyclic host can eliminate the need of counteranions required for the transport of charged complexes into the organic phase.7 Proton-ionizable crown ethers containing a carboxylic acid or a triazole functional group have been shown to react with metal ions, effectively forming neutral complexes extractable by organic solvent^.^-^^ The extraction requires no specific counteranions and is reversible with respect to pH. The complexed ligand may be regenerated to the hydrogen form by stripping the metal ions with an acid solution. There is no report in the literature regarding the extraction of metal ions in supercritical fluids using macrocyclic compounds as extractants. We have recently synthesized a group of macrocyclic compounds containing two triazole subcyclic units for complexation with divalent metal ions." This paper reports our results regarding selective extraction of Hgz+from solid and liquid samples by supercritical COZcontaining bistriazolocrown ethers as extractants. EXPERIMENTAL SECTION
Preparation of Macrocyclic Compounds. Macrocyclic compounds 1-3 (Figure 1)were prepared by the reaction of 3,s bis(chloromethyl)-l-(tetrahydro-2-pyrany1)-1H-1,2,4triazole with appropriate catechols and truns-1,2-cyclohexanediol, respectively, in tetrahydrofuran (THF) and NaH as a base followed by an acid hydrolysis. These compounds are 18membered macrocycles with six donor atoms, an appropriate size which should be suitable for interaction with divalent heavy metal ions. The synthesized bistriazolo-crowns were purified by chromatography and recrystallization. Desired compounds were isolated as white solids in 18-34% yields. The structures proposed for these new macrocyclic compounds are consistent with the data obtained from their IR and proton NMR spectra and elemental analyses. These triazole containing macrocyclic compounds are white solids insoluble in water. The melting points of crowns 1,2, and 3 are 258-260, 240-242, and 178-180 "C, respectively. A detailed description of the synthetic procedures, including spectral data, for each of these macrocyclic compounds is given elsewhere.ll Reagents and Apparatus. Mercury solutions were prepared from an 1000 ppm atomic absorption (AA) standard in 5% HN03 obtained from Johnson Matthey (Ward Hill, MA). A standard AA (6) Pedersen, C. J. Science 1988,241,536. (7) Wai, C. M. In Preconcentration Techniquesfor Trace Elements: Alfassi, Z. B., Wai, C. M., Eds.: CRC Press: Boca Raton, FL, 1991; Chapter 4, p 107. (8) Tang, J.; Wai, C. M. Anal. Chem. 1986,58,3233. (9)Walkowiak, W.; Charewicz, W. A.; Kang, S. L.; Yang, I. W.; Pugia, M. J.; Bartsch, R A. Anal. Chem. 1990,62,2018. (10) Izatt, R M.; Lind, G. C.; Bmening, R L.; Huszthy, P.; McDaniel, C. W.; Bradshaw, J. S.; Christensen, J. J. Anal. Chem. 1988,60,1694. (11) Elshani, S.; Apgar, P. M.; Wang, S. F.; Wai, C. M.]. Heterocycl. Chem. 1994, 31,1271.
Analytical Chemistty, Vol. 67, No. 5, March 1, 7995 919
M-d
H
H
$A
w dN-"
Crown 1
Crown 2
N-N'
H
w -N-N
ii
Crown 3
Figure 1. Structures of the bistriazolo-crown ethers tested in this study.
solution of Au3+(in 10%HCl) was also purchased from Johnson Matthey. Other standard AA solutions of Cd2+,Co2+,MnZf,Ni2+, Pb2+,and Zn2+were purchased from EM Science (Gibbstown, NJ). Sodium acetate and acetic acid were obtained from Aldrich Chemical Co. Ultrex HN03 was obtained from J. T. Baker Co. Deionized water was prepared by passing distilled water through a Barnstead ultrapure water purification cartridge and a 0.2-pm filter assembly (Pall, Ultipor DFA). All containers used in the experiments were acid washed, rinsed with deionized water, and dried in a class 100 clean hood. All experiments were performed with a laboratory-built supercritical fluid extraction apparatus. SFC-grade COZ or COZwith 5% methanol (Scott Speciality Gases, Plumsteadville, PA) was delivered to the system using a microprocessor-controlled highpressure pump (Haskel Inc., Burbank, CA). The pressure of the system was monitored to f 5 psi using a Setra Systems (Acton, MA) pressure transducer. The extractor for solid samples consisted of an inlet valve (Supelco, Bellefonte, PA) and an outlet valve connected to a 3.5-mL commercial extraction cell (Dionex, Sunnyvale, CA). The main body of the liquid extraction vessel was made from a stainless steel column (0.75 cm i.d. and 14 cm in length, Alltech), and the infittings were obtained from Swagelok (Seattle, WA). The vessel was modified for use with liquid samples in the same manner as reported earlier by Hendrick et a1.12 The extraction cell was placed in an oven that was temperature controlled by a thermostat. A fused-silica tubing (Dionex, 50 pm i.d. and 20 cm in length) was used as the pressure restrictor for the exit gas. The SFE system allowed static and dynamic extractions to be performed by use of the outlet and inlet valves. Extraction Procedures. For the solid extraction experiments, a glass tube (3 cm in length) plugged at one end with a piece of glass wool was used as a sample holder. Into the other end of the glass tube was inserted a filter paper (1 x 2 cmz) spiked with 10 pg of Hgz' or other metal ions. A 1@pLaliquot of deionized water was added to the filter paper, and the end was plugged with glass wool. The sample tube was placed in the extraction cell behind another glass tube containing a known amount (5 mg) of a bistriazolo-crown. The extraction cell was immediately positioned in the oven, which was preheated to 60 "C, and pressurized to 200 atm. Solid samples were extracted for 10 min statically followed by 15 min of dynamic extraction. After the dynamic extraction, the system was depressurized and (12) Hedrick, J. L.; Mulcahey, L. J.; Taylor, L. T. In Supercritical Fluid
Techno1og)rTheoretical and Applied Approaches to Analytical Chemist5 Bright, F. V.; McNally, M. E., Eds.; ACS Symposium Series 488; American Chemical Society: Washington, DC, 1991; pp 206-220. 920 Analytical Chemisty, Vol. 67, No. 5, March 7, 1995
the sample tube was removed from the extraction cell. The filter paper was placed in a polyethylene vial for neutron activation analysis (NAA) or acid digested for ICPMS analysis. The extraction efficiencies were calculated based on the amounts of metal ions found in the samples before and after extraction. For the extraction of metal ions from sand sample, 300 mg of sand was loaded into a glass tube followed by spiking with 10 pg of mol). The procedures for the sand each metal ion (5 x extraction experimentswere similar to those described above for the filter paper experiments. For the extraction of metal ions from liquid samples, a solution containing 5 ppm Hg2+with a total volume of 4.5 mL was placed in the liquid extraction vessel. A known amount of the ligand (crown 3) was loaded in a ligand cylinder wich was placed upstream from the liquid extraction vessel. The extraction conditions were 30 min statically followed by 30 min of dynamic extraction with 5%methanol modified COZ as the fluid phase. The flow rate of the exit gas was measured to be -700 mL/min, correspondingto -1.7 mL/min of supercritical COZ at 60 OC and 200 atm. The results given in Tables 1-3 represent the averages of triplicate extraction trials. Sample Analysis. For nondestructive NAA, filter paper samples were heat sealed in 2/5-dram polyethylene vials for neutron irradiation. Standards were also spiked on filter papers and irradiated under the same conditions as the samples. For ICPMS analysis, samples were digested with concentrated Ultrex nitric acid in glass vials on a hot plate at temperatures slightly below boiling. After 3-4 h of digestion, the cellulose tissue matrix disappeared and the clear solutions were cooled to room temperature. The digested solutions were transferred to volumetric flasks and diluted to 10 mL for ICPMS analysis. The metal chelates collected in the chloroform solution were back-extracted with 5 mL of 50% Ultrex HN03 for -0.5 h using a wrist-action mechanical shaker. After phase separation, the acid solution was taken for ICPMS analysis or for NAA. A Sciex Elan Model 250 ICPMS was used for metal analysis. Samples were introduced into the argon plasma at 1.0 mL/min using a peristaltic pump and an automatic sampler. The instrument was run in "multielement" mode averaging 10 repeats of 0.5 s/element for a total integrated count time of 5 s/element. For NAA, sealed samples were irradiated in a 1-MW Triga nuclear reactor at a steady flux of 6 x 1OI2 n cm-2 s-l for 1 h. After being cooled for several days, the activated samples and the standards were counted on an Ortec GeCi) detector. The detector had a resolution of 2.3 keV at 133hkeV radiation from 6oCoand an efficiency of 15%relative to a 3 x 3 NaI crystal. The activities of z03Hg(tllz = 46.8 days) and lg8Au(tllz = 2.7 days), at
279.2- and 411.8keV y peaks, respectively, were used for quantification of these elements. The details of y spectrometry as well as neutron irradiation are given elsewhere.'3 Solubility Measurement. A weighed amount of a bistriazolccrown ether was placed in a glass tube. The sample tube was plugged with glass wools at both ends and inserted into an extraction cell of known volume (3.5 mL). The extraction cell was installed in the supercritical fluid extractor at 60 "C and 200 atm. After 30 min of static extraction, the system was depressurized by opening the outlet valve. The analyte in the exit gas was collected in two glass vials connected in a series each containing 4 mL of chloroform. After depressurizing the system, the sample tube was removed from the extraction cell and the empty cell was reinstalled into the oven. The system was flushed with C02 for 20 min, and the analyte in the exit gas was also collected in a chloroform solution. The total amount of the bistriazolo-crown collected in the chloroform solution was determined by UV-visible spectroscopy. The solubility of the bistriazolo-crown in supercritical COZ was calculated from the total amount of the crown found in the chloroform solution divided by the volume of the extraction cell. RESULTS AND DISCUSSION
The structures of the bistriazolo-crownethers included in this study are given in Figure 1. The bistriazolo-crowns are sparsely soluble in neat C02. For example, at 60 "C and 200 atm, the solubilities of bistriazolo-crowns 1,2, and 3 in neat COZare 1.3 x 1.0 x W5,and 4.3 x M, respectively. Bistriazolocrown 3, with a tert-butyl substitution in each of the benzene rings, is more soluble in neat C02 than the other two crown ethers. However, in methanol-modified COZ, the solubilities of the bistriazolo-crowns are significantly enhanced. In 5 mol % methanol modified COz, the solubilities of bistriazolo-crowns 1,2, and 3 and 1.3 x loy3 M, are increased to 2.1 x 2.0 x respectively, at 60 "C and 200 atm. Bistriazolo-crown 3 has a solubility in 5% methanol modified C02 comparable to that of LiFDDC.I4 The molecular mass of bistriazolo-crown 3 is 518 g/mol. On the basis of the solubility data, -0.67 mg of this macrocyclic compound can be dissolved in each milliliter of the methanol-modified COSat 60 "C and 200 atm. The results of extracting Hg2+from the cellulose-based filter papers by supercritical COZcontaining bistriazolo-crowns 1-3, at 60 O C and 200 atm, are given in Table 1. In each of these experiments, 10 pg of Hg2+ (-5 x mol) was spiked onto a Whatman No. 42 filter paper (1 x 2 cm2). The amount of each ligand was fixed at 5 mg, which corresponds to about 1.19 x 1.23 x and 0.96 x mol of the macrocyclic compounds 1 , 2 , and 3, respectively. Only a portion of the ligands present in the system was dissolved in the fluid phase during the static extraction. For example, in the case of bistriazolo-crown 3, about half of the ligand placed in the extraction cell (3.5-mL volume) was dissolved in the methanol (5%) modified fluid phase during the static extraction step. The rest of the ligand was presumbly consumed during the dynamic extraction step. In all cases, the ligand was in large excess relative to the Hgz+ present in the sample. The extraction times for these SFE experiments were also fixed (10-min static followed by 15-min dynamic extraction). (13) Mok, W. M.; Wai, C.M.Anal. Chen. 1987,59,233. (14) Laintz, K E.; Wai, C.M.; Yonker, C.R; Smith, R D.J. Supercn't. Fluid 1991,4, 194.
Table 1. Extractlon of Hg2+and A d + from Filter Paper with Supercritical COSContaining Blstriazolo-crowns
ligand"
fluid phase
crown 1
C02
crown2
C02 5%MeOH C02 5%MeOH COZ
crown3
COz COz C02
coz
coz
coz COz COz
+ +
+ 5%MeOH + 5%MeOH
+ 5%MeOH + 5%MeOH
matrix conditionb dry wet dry wet dry wet dry wet dry wet dry wet
% extractionC Hg Au
24f2 42f3 64&3 81f2 19f2 26f2 5153 8152
