Anal. Chem. 2000, 72, 1819-1822
Improving the Recovery of Ionic Solutes from Aqueous Media by Modified Thermospray Liquid-Liquid Extraction E. S. Farrell and G. E. Pacey*
Miami University, Oxford, Ohio 45056
Our previously reported procedure for the extraction of charged compounds from aqueous samples by thermospray liquid-liquid extraction (TSLLE) was essentially a one-step extraction involving large sample volumes. In this report, recirculative extraction, analysis of small sample volumes, the halide ion effect, and the effect of solvency or solvent modification on the extraction efficiency of benzoic acid (BA) by TSLLE were investigated. Compared to the one-step procedure that resulted in an extraction efficiency of only 28% for BA in n-hexane, recirculative TSLLE resulted in a BA recovery of 65% after five extraction cycles. When applied to sample volumes of 5-10 mL, TSLLE extracted BA with a precision of 2.86.1%. NaF, NaCl, and NaBr were also used to enhance analyte recovery. NaF gave the best recovery, 104%, for BA relative to the 88% obtained by batch processing. Some improvements in the extraction efficiency was observed when solvent modifiers such as methanol, ethanol, and 2-propanol were used. In previous reports,1,2 we pointed out that the ability of the thermospray liquid-liquid extraction (TSLLE) technique to disrupt the hydration shell of charged compounds may enhance their extraction from the aqueous phase into the organic phase. There it was noted that probe temperature is the single most important factor in thermospray extraction of charged compounds from water. It was also noted that excessively high probe temperatures may result in the redistribution of some analytes into the aqueous phase. While it is true that lowering the probe temperature may alleviate the problem, an indiscriminate choice in this regard may also lead to a decrease in surface area with an overall loss in extraction efficiency. Adding an inorganic salt or an organic modifier to the aqueous solution containing the analyte may reduce the solubility of these species in the water and thus help to lower dependence or reliance on the need for high-temperature adjustments. This paper discusses the use of solvent modifiers and “salting-out” agents for thermospray liquid-liquid extraction of charged compounds from water. It also discusses how certain modifiers can be used as indicators for the TSLLE process. Reproducible extraction of small sample volumes by TSLLE is also examined. (1) Farrell, E. S.; Pacey, G. E. Anal. Chem. 1996, 68, 93-99. (2) Farrell, E. S.; Pacey, G. E. Anal. Chem. 1997,69, 3339-3345. 10.1021/ac991105r CCC: $19.00 Published on Web 03/23/2000
© 2000 American Chemical Society
EXPERIMENTAL SECTION Reagents. HPLC grade methylene chloride, chloroform, carbon tetrachloride, hexane, ethanol, 2-propanol, methanol, and hydrochloric acid and reagent grade glacial acetic acid, sodium chloride, sodium fluoride, potassium iodide, sodium bromide, and sodium hydroxide were obtained from either Fisher Scientific (Fair Lawn, NJ) or Baxter Health Care Corp. (McGraw Park, IL) and used without further purification. Reagent grade benzoic acid was purchased from Sigma Chemical Co. (St. Louis, MO). Triply distilled deionized water were used for all investigations. Apparatus and Instrumentation. The apparatus, which is described elsewhere,1 consists primarily of a water-cooled Pyrex glass chamber (300 mL), several thermospray vaporizer inserts (Vestec VT1460-4 vaporizer), a dual-stage condenser, and a vapor bypass assembly. Supporting instruments include a 120-V variable autotransformer, a Fluke 52 K/J thermocouple reader, an SSI model 300 LC pump equipped with an SSI LP-21 pulse dampener (Scientific System Inc.), a Perkin-Elmer model 250 binary LC pump, and a Neslab Coolflow 33 refrigeration cooler. An UV detection scheme was used to facilitate rapid analysis. UV/visible measurements were made on a Hewlett-Packard 8452A photodiode array spectrophotometer controlled by an IBM PC. Recirculative TSLLE is accomplished by drawing off the aqueous layer in the extraction chamber and using it in place of fresh sample in the sample thermospray probe. The modification in the extraction chamber is an exit port that can be attached at the sample thermospray port. This recirculation is an automated process since the aqueous layer is pumped using the same pump originally used for the fresh sample. Sample/Solvent Preparation Procedure. A sample solution containing 1.03 × 10-3 M benzoic acid (BA) was formed by dissolving 0.125 g of benzoic acid in 5 mL of about 0.1 M NaOH solution and then diluting to a final volume of 1 L with triply distilled deionized water. pH adjustments for BA and other analytes were made using 6 M HCl or 6 M NaOH. The chloroform/hexane solvent mixtures were prepared by adding 10, 20, 30, and 40 mL of chloroform to a series of 100-mL volumetric flasks. Five milliliters of glacial acetic acid (GA) was then added to the each flask and the volume adjusted to mark with n-hexane. In a like manner, separate halocarbon/hexane solvent mixtures, each containing 5% (v/v) GA, were prepared for methylene chloride and carbon tetrachloride. Halide ion solutions were prepared by dissolving 0.8398 g of NaF, 2.058 g of NaBr, and 1.160 g of NaCl in three separate 200-mL volumetric flask with 1.03 × Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1819
10-3 M BA solution. Alcohol/chloroform mixtures (15% v/v) containing either methanol, ethanol, or 2-propanol were prepared by diluting 30 mL of the alcohol reagent to volume with chloroform in a series of 200-mL volumetric flasks. General Extraction Procedure. Both the sample and the solvent were admitted through their respective probes at flow rates and probe temperatures as described elsewhere in the text. At the beginning of each set of extractions, triply distilled water was extracted to obtain a system blank and ascertain probe temperature equilibrium. Following this, the sample pump was briefly turned off to allow replacement of the blank with an actual sample. At the end of each extraction, an additional 5 mL of water was circulated through the sample probe to facilitate removal of residual analytes from the transfer lines. Following extraction, the temperature of the extraction chamber was elevated as necessary to break up the emulsion. Upon separation of both liquids, the organic phase was collected, diluted to 25 or 50 mL, and analyzed by UV/visible spectroscopy at a wavelength of 276 nm. Batch analyses were performed by extracting the sample three times in a 250-mL separatory funnel with 10-mL portions of the extracting solvent. The combined organic fractions were diluted to 25 or 50 mL and analyzed as described above. RESULTS AND DISCUSSION Effect of Different Solvents and Solvent Mixtures on Recovery. Often a mixture of two solvents, one having solubility parameter (d-value) higher, and the other having d-value lower than that of the solute, is a better solvent than each of the two solvents separately.3 The single solvent offers limited room for manipulating the system since it alone must satisfy all requirements pertaining to selectivity, capacity, solubility, mass transfer, phase separation, cost, etc. It is possible to achieve many of the desired properties of an ideal solvent by using binary solvent systems, where two types of solvating solvents are used to change the mutual miscibility characteristics. Binary solvent mixtures have the advantage that changing their composition will change the electrostatic solute/solvent interactions. Physical properties such as viscosity, interfacial tension, dielectric constant, and density may change as a function of solvent composition.4,5 It has been shown that the solute in a binary mixture will surround itself preferably with the component of the solvent that leads to the more negative Gibbs energy of solvation.6-8 When this happens, the ratio of the solvent components in the solvent shell may be different from that of the bulk solution and may enhance the distribution of solute in the organic phase. An examination of Figure 1 shows that a 1:1 mixture of chloroform and methylene chloride gives a better overall recovery of benzoate ion than either of the individual solvents over the pH range tested. The results were based on the extraction of 25 mL of 1.03 × 10-3 M BA at sample and solvent flow rates of 5 and 3 mL/min, respectively, (3) Barton, A. F. M. Handbook of Solubility Parameters and other Cohesion Parameters; CRC Press: Boca Raton, FL, 1983. (4) Marrison, G. H.; Freiser, H. Solvent Extraction in Analytical Chemistry; John Wiley & Sons Inc.: New York, 1966; p 9. (5) Blumberg, R. Liquid-Liquid Extraction; Academic Press: London, 1988; p 18. (6) Reichardt, C. Solvents and Solvent Effects in Organic Chemistry; VCH Publishers: New York, 1988; p 35. (7) Stace, A. J. J. Am. Chem. Soc. 1984, 106, 2306. (8) Strehlow, H.; Koepp, H. J. Phys. Chem. 1958, 62, 373.
1820 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000
Figure 1. Comparison of BA recovery in different organic solvents using TSLLE.
and sample and solvent temperatures of 160 and 142.8 °C, respectively. Since solvency increases as the solubility parameter (d-value) of the solvent approaches that of the solute, the extraction efficiency should increase with increasing d-value. Except for chloroform and methylene chloride, where differences in recovery were small, extraction efficiency seems to correlate with both the d-value and the dielectric constant for each of the solvents tested. For the extraction of acidic or basic solutes, the ionization constant of the acid or base is affected, not only by the acidity or basicity of the solvent but also by a change in dielectric constant and solvation capability of the solvent. In turn, the proportions for each component in the solvent mixtures affected the extraction, but as a general rule, the primary solvent determines the nature of the solute distribution. To underscore the above observation, Reichardt noted that whereas nitric and carboxylic acids are respectively completely and partially ionized in water, the former is only partially ionized in methanol, while the latter is completely ionized in liquid ammonia.9 Such a solvent mixture effect is where 25-mL aliquots of 1.03 × 10-3 M BA (pH 8.04) were subjected to TSLLE with a solvent mixture containing 5% GA and 10-40% chloroform in hexane. In the case of the 10, 20, 30, and 40% chloroform/hexane mixtures, the increase in recovery was 17, 26, 30, and 39%, respectively. Each extraction was performed with temperature and flow rate of the solvent probe set at 154 °C and 3 mL/min, respectively. The temperature and flow rate of the sample probe were fixed at 177 °C and 5 mL/min, respectively. In agreement with the foregoing discussion, the recovery of BA increased linearly with increasing levels of chloroform for the range tested. Circulative TSLLE. Many of the solvents used in conventional liquid-liquid extraction techniques are, in some way, linked to a (9) Reichardt, C. Solvents and Solvent Effects in Organic Chemistry; VCH Publishers: New York, 1988; pp 64-65.
Table 1. Effects of Alcohol Percentage and Type on Extraction Efficiency
Figure 2. TSLLE of BA in hexane as a function of sample recirculation at varying flow rates.
number of health and environmental concerns.10,11 Solvents such as carbon tetrachloride and methyl chloroform will be phased out by the year 2000.12 As a means of reducing the risk of exposure to the undesirable chemicals, research efforts have focused on minimizing the quantity of solvent used, eliminating their use altogether, or exploring less toxic alternatives. Although linear alkanes lack the solvency of chloroform and methylene chloride, nonpolar liquids such as octane and hexane are less toxic and pertinent to LLE of organic compounds. To counteract losses in extraction efficiency that may result from the selection of n-alkane solvents, the sample can be subjected to repetitive extraction to achieve the desired level of recovery. This was the idea behind the extraction of a 1.03 × 10-3 M BA solution, which was subjected to thermospray liquid-liquid recirculative extraction with a 5% GA/hexane mixture. Compared to the inefficient single-cycle process depicted in Figure 1 for hexane, Figure 2 shows that analyte recovery in hexane can parallel the single-cycle performance for chloroform and methylene chloride, providing the raffinate is reextracted a sufficient number of times at high solvent flow rates. The improved recovery at higher flow rate is not surprising since the admission of larger volumes of solvent exposes the benzoate ion to the neutralizing effect of greater amounts of GA. When the selected solvent probe temperature no longer supplies the latent heat of vaporization at a given flow rate, solvent nebulization will diminish and a decrease in analyte recovery will be observed. This is certainly the case in Figure 2 for the 4 and 5 mL/min cycles, where recovery appears to be flattening off at higher flow rates. For this series of extractions, the solvent probe temperature was set at 145 °C. All other experimental conditions are identical to those reported for Figure 1. (10) Haggin, J. Chem. Eng. News 1995, 73 (July 10), 25-26. (11) Hughes, D. E.; Gunton, K. E. Anal. Chem. 1995, 67, 1191-1196. (12) Zurer, P. S. Chem. Eng. News 1995, 73 (Dec 4), 26-27.
alcohol type/percent
percent extraction
methanol/10 methanol/20 methanol/40 methanol/50 methanol 15/chloroform ethanol 15/chloroform 2-propanol 15/chloroform
5.5 5.7 32 63 38 25 17
Role of Alcohol Modifiers in TSLLE of Anionic Species. Alcohol modifiers may serve several useful roles in LLE. Since they may readily distribute between both phases, alcohols can be used to interact with both the extractant and the solute. As protic solvents, they are easily integrated into the solvent shell of anionic solutes and, in so doing, change the microenvironment of the solutes to increase distribution in the organic phase. Thus, it would be reasonable to expect a correlation between the strength or concentration of these protic additives and the extraction efficiency of anionic species. Consistent with this expectation, Table 1 shows that an increase in the methanol content of the extractant led to proportionate increases in extraction efficiency. The result was based on the extraction of 1.03 × 10-3 M BA, which was processed at a flow rate of 5 mL/ min and a temperature of 165 °C. The solvent probe flow rate and temperature were set at 3 mL/min and 142 °C, respectively. In TSLLE, alcohol additives may serve a secondary, but more important, role. When droplets are formed from bulk solutions in a thermospray extractor, there is always the risk that the flow rate and the temperature of the probes can be adjusted such that a dry gas may result. For the extraction of salts or thermally sensitive solutes, these conditions tend to promote analyte decomposition, intraprobe salt deposition, and eventual clogging of the probe capillary. To monitor this behavior, we capitalized on the observation by Brauman and Blair that the acidity of aliphatic alcohols in the gas phase is the opposite of that in solution.13,14 The reversal of the relative acidities of methanol, ethanol, and propanol was studied using two sets of temperature operating conditions. The first experiment was representative of the typical operating temperatures where the sample and solvent probes were set at 165 and 152 °C, respectively. The second experiment was performed with the sample probe temperature at 270 °C and the solvent probe temperature elevated to 307 °C. Under these conditions, 25-mL portions of 1.03 × 10-3 M BA was extracted with a 15% (v/v) alcohol/chloroform mixture containing either methanol, ethanol, or 2-propanol. In both cases, the sample and solvent flow rates were held constant at 5 and 3 mL/min, respectively. If gas-phase conditions are approached, the expected order of acidity would be CH3OH < CH3CH2OH < CH3CHOHCH3. Table 1 shows that the reversal of acidity for the alcohols tested was not observed, despite a decrease in the recovery by the more acidic methanol relative to ethanol and 2-propanol. The observed order indicates solution-based interactions with the more acidic alcohol giving the higher percent recovery. These experimental observations agree with the idea that solvation of the departing (13) Brauman, J. I.; Blair, L. K. J. Am. Chem. Soc. 1968, 90, 6561. (14) Brauman, J. I.; Blair, L. K. J. Am. Chem. Soc. 1970, 92, 5986.
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000
1821
anion in protic solvents takes place preferably through hydrogen bonding; hence the more acidic the additive, the greater will be the electrophilic pull on the departing anion. It should be pointed out that the solvating abilities of isolated solvent molecules for some solutes are often very different from their performance as bulk liquids,15 even when true gas-phase conditions are not approached. This type of pseudo gas phase or differential solvation condition was credited for the reversal of the acidity order among substituted phenols and halo-substituted carboxylic acids.16-18 Care must therefore be taken in selecting test compounds or interpreting data that suggest the approach of the gas-phase condition during TSLLE. Effect of Halide Ions on TSLLE Recovery. Water is a highly polar solvent with a high dielectric constant. It permits charge separation of anions and cations by surrounding them with a solvent sheath so that they experience less electrostatic attraction. Much of the solvent action of water is attributable to its hydrogenbonding capability. Adding an electrolyte to the solution can reduce the competitive strength of water for ions. High electrolyte concentration helps extraction by lowering the dielectric constant, by reducing the water activity, and by mass action effect (that is, replacement of water by a suitable anion or cation from within the electrolyte). The term salting-out is applied to those electrolytes whose addition to the aqueous phase increases the distribution ratio in favor of improved solute extraction. In the presence of dissociated salts, a fraction of the aqueous molecules undergoes solvation interactions with the dissolved ions. The water forms a solvent shell around these ions and is unavailable as a “free solvent” for interaction with the targeted analyte. A net increase in extraction efficiency should therefore be observed with corresponding addition of the electrolyte. The ability of aqueous halide ions to interact with water and enhance the extraction of BA was investigated. Three solutions containing BA at a concentration of 1.03 × 10-3 M and F-, Cl-, or Br- at a concentration of 0.1 M were adjusted to a pH of 8.06 and then spray extracted with a 5% GA/chloroform mixture. The solvent and sample probe temperatures were respectively set at 142 and 192 °C, with corresponding flow rates of 3 and 5 mL/ min. As a comparison, the samples were also extracted in a separatory funnel with three 10-mL portions of the solvent. The combined organic phases were diluted to a final volume of 50 mL with chloroform and analyzed as described before. The relationship of BA recovery to the type of halide ions used was in the order of F- > Cl- > Br-. This agrees with the observation that the smaller an ion, the greater will be its charge density, and the more effective would be its solvation.19 This fact is also apparent from the corresponding Gibbs energies of hydration for F-, Cl-, and Br- that are listed in the literature as negative 472, 347, and 321 kJ/mol, respectively. These energies can be as high as bond energies that are generally between 209 and 628 kJ/mol.20 While
it is apparent that F- is the most effective salting agent for the TSLLE conditions selected, no significant difference appears to exist between the recoveries with Cl- and Br-, which is perhaps a reflection of the similarity in hydration energy of both species. Quite interestingly, there is a very close agreement among the values obtained when both the percent recovery and the hydration energy of Cl- and Br- are expressed as a fraction of the percent recovery and the hydration energy of F-, respectively. Halide ions may also help to reduce solvent viscosity during TSLLE. Aqueous solutions containing large, singly charged, spherical ions such as potassium iodide, are known to exhibit greater fluidity than pure water itself.21 Similar effects were reported for ethylene glycol and glycerol solutions.22 The observed pressure for a 1.0 M KI solution pumped at a flow rate of 5 mL/ min and a temperature of 190 °C was reduced by about 100 psi to 1400 psi. The greater mobility of the water molecules between the relatively immobilized solvent shell of the ion and the ordered bulk solvent accounts for the enhanced fluidity in water. The exchange frequency of water molecules in this intermediate region of the ion is greater than in the region of pure water.23 It should be pointed out tht, under TSLLE conditions and at the concentration level described in this paper, the use of KI led to spectral interference of BA and discoloration of the sample solution. Small-Volume Extraction. Despite attempts to minimize sample loss, liquid-liquid extraction of small quantities of liquids may often result in loss of analyte. Since the volume of the extraction chamber in a TSLLE is 300 mL, it was important to establish the effectiveness of such an apparatus for the concentration of small volumes of nonvolatile samples. For this purpose, 5and 10-mL aliquots of a 1.03 × 10-3 M BA solution adjusted to pH 8.06 were spray extracted at 5 mL/min and a temperature of 165 °C. A 5% GA/hexane mixture introduced at a flow rate of 3 mL/min and a temperature 156 °C served as extracting solvent. Under these conditions, the single-cycle extraction efficiency of BA in hexane using a sample volume of 50 mL was shown to be 16%. Recoveries of 99.3% of the expected value for the 5-mL samples (% RSD ) 6.1, n ) 4) and 97% for the 10-mL aliquots (% RSD ) 2.8, n ) 4) demonstrate acceptable performance for small volume extraction under the stipulated conditions. Conclusion. From the foregoing information we may conclude that liquid-liquid extraction techniques such as salting-out and solvent modification can be used in conjunction with thermospray liquid-liquid extraction to enhance the extraction of charged species from water. For these solutes, sample volumes as small as 5 mL can also be extracted reproducibly. Despite the inefficient performance of less toxic solvents such as n-hexane, recirculative extraction can be employed to improve extraction efficiency within a reasonable time frame.
(15) Reichardt, C. Solvents and Solvent Effects in Organic Chemistry; VCH Publishers: New York, 1988; p 88. (16) Mashima, M.; McIver, R. T.; Taft, R. W.; Bordwell, F. G.; Olmstead, W. N. J. Am. Chem. Soc. 1984, 106, 2717. (17) Cadwell, G.; McMahon, T. B.; Kebarle, P.; Bartmess, J. E.; Kiplinger, J. P. J. Am. Chem. Soc. 1985, 107, 80. (18) Yamdagni, R.; Kebarle, P. J. Am. Chem Soc. 1973, 95, 4050. (19) Amis, E. S.; Hinton, J. F. Solvent Effects on Chemical Phenomena; Academic Press: New York, London, 1973; Vol. 1.
AC991105R
1822
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000
Received for review September 23, 1999. Accepted October 12, 1999.
(20) Marcus, Y. Pure Appl. Chem. 1982, 54, 2327. (21) Samoilov, O. Y. Structure of Aqueous Electrolyte Solutions and the Hydration of Ions; Plenum Press: New York, London, 1965. (22) Engel, G. E.; Hertz, H. G. Phys. Chem. 1968, 72, 808. (23) Reichardt, C. Solvents and Solvent Effects in Organic Chemistry; VCH Publishers: New York, 1988; p 34.
