Anal. Chem. 1986, 58,3240-3242
3240
Preconcentration of Trace Elements from Aqueous Solutions by Osmosis Sir: Various techniques have been developed to enrich analyte levels in aqueous solutions when concentrations fall below the detection limit of current measurement methods ( I , 2). These techniques are often selective for only a few elements, suffer matrix variation, and are often time-consuming. A new technique has been developed to concentrate transition-metal ions from dilute aqueous solutions utilizing the principle of osmosis. T o the best of our knowledge, no description of the use of osmosis for preconcentration has appeared in the chemical literature. The well-known process of osmosis involves two liquids of differing solute concentration separated by a porous membrane that is permeable t o the solvent but impermeable to the solute. The solvent activity is higher on the side of the less concentrated solution, which creates a potential gradient across the membrane. In order t o bring the system to equilibrium, solvent will permeate the membrane and enter the more concentrated solution until both liquids are of equal chemical potential. This principle is currently being employed in our preconcentration technique. If the aqueous sample to be concentrated contains ions a t trace levels, a substantial potential gradient can then be established across the membrane by placing a concentrated aqueous salt solution opposite the sample. The resulting osmotic pressure is the driving force for water permeation. The net result is a nonselective concentration of ions in the sample by the osmotic removal of water. The intent of this communication is to briefly describe the fabrication, operation, and performance characteristics of the osmotic concentrating apparatus and to propose this system as an on-line sample preconcentrator.
EXPERIMENTAL SECTION Apparatus. The concentration cell was constructed from two plexiglass blocks (23 cm X 6.5 cm X 1.27 cm) into which four channels were cut (21 cm x 1.1cm X 0.03 cm), two in each block as shown in Figure 1. Two ports a t the top and bottom of each block allow the concentrated salt solution to be drawn through these outer channels by a Saturn positive displacement pump (Fluorocarbon) operating at approximately 300 mL/min. Strips of nylon mesh (not shown) are laid in these channels to help support the membrane and to create a turbulent flow pattern. An inner channel was formed by placing a piece of membrane (23 cm X 4 cm) on each side of a gasket made of Teflon (23 cm x 6.5 cm x 0.15 cm), which also contains two channels. Tubing made of Teflon (0.11 cm id., 0.17 cm 0.d.) was bonded into the gasket at the end of each channel providing inlet and exit ports. The sample solution was pumped through the inner channels in a direction countercurrent to the concentrated salt solution by a Gilson peristaltic pump (Cole-Parmer). The cell was fastened together by 16 bolts. The membrane on each side of the sample channel formed a physical boundary between the salt solution and sample channels with an effective area of about 91 cm2. The cell was vertically mounted on a ring stand and rotatable to facilitate filling and emptying of the sample channel. Saturated salt solution (NaC1) was contained in a 25-L constant-temperature bath housed in an insulated wood case. Inlet and outlet tubes, a thermometer, and an overhead stirrer were secured through the bath cover. Approximately 10 kg of solid NaCl was added initially. A block diagram of the entire system is illustrated in Figure 2. Analyte recovery data were obtained on a multielement inductively coupled plasma emission spectrometer (ICP) (Model 1100,Jarrell-Ash). Experiments for direct interfacing of the cell to an ICP were done on a Perkin-Elmer Model 6500 ICP system. 0003-2700/86/0358-3240$01 SO/O
Early attempts to directly interface the output of the concentrating cell to a nebulizer caused rupture of the membrane due to the back pressure. A simple but effective interface consisting of a 1.0-mLEppendorf pipet tip eliminated back pressure problems. The pipet tip was positioned with the tapered end down, and it received the output from the concentrating cell through the top. Simultaneously, the solution was drawn out through a tube fastened to the tapered tip at the bottom. A peristaltic pump pumped the solution to the nebulizer of an ICP at approximately the same rate as the concentrating cell discharged. A dead volume of about 0.5 mL was maintained in the pipet tip. Membrane. SEPA-97 R.O. Membrane (lot 06014A L-011) made of cellulose acetate was obtained from Osmonics, Inc., Hopkins, ME. The membrane, a~ received from the manufacturer, consists of a thin layer (0.084.13 mm) of cellulose acetate bonded to a nylon fabric. With care, the fabric can be peeled away when immersed in water leaving behind the layer of cellulose acetate. The membrane is composed of a dense microporous surface layer (less than 1% of the total thickness) and a highly porous substructure. The surface layer (the side facing away from the fabric backing), contains asymmetric pores in the size range of 5 A. This side must face the sample solution for the most efficient sample permeation. Cut membranes were stored in a 4% formaldehyde solution prior to use in order to reduce microbial degradation. Reagents. Samples to be concentrated were prepared from appropriate aliquots of 1000 pg/mL Cd, Cu, Mn, Ni, and Zn atomic absorption standards (Fisher Scientific Co.). For some experiments the standards were prepared in dilute acid (pH 2), and for others the metals were chelated with a 10 M excess of EDTA relative to the total ion content. NaOH was added so that the sample pH was 9. Bulk sodium chloride (NaC1) (Fisher ScientificCo.) WBS purified by diluting the salt in water, adjusting the pH to approximately 10, and passing it through a 5-10-g column of Chelex 100 resin, 50-100 mesh (Bio-Rad). The column was periodically regenerated with 1%hydrochloric acid (HCl), and aliquots of the NaCl solution were drawn and analyzed by ICP for the five elements of interest. The NaCl solution was considered pure when analyte levels fell below the detection limits of the ICP. Nitric acid ("Os) (Taychemco) was purified by sub-boiling distillation prior to use. All reagents were made with doubly distilled deionized water. Procedure. After the concentrating cell was assembled, the saturated salt solution and water were cycled through the outer and inner channels, respectively, for approximately 15 min to assure a steady-state permeation rate. The dimensions of the sample channel were too large to segment the flow with an air bubble; therefore, it became necessary to rotate the cell so that gravity could assist in filling and emptying the inner channel. Permeation rates were measured by circulating water a t 10 mL/min from a 100-mL volumetric flask through the cell and back into the flask, measuring the volume reduction with time. Analyte recovery data were obtained by allowing 100-mL samples to make a single pass through the concentration cell, measuring the concentration of the analyte in the collected output. Sample solutions were loaded into the cell operating the sample pump at the highest speed (24 mL/min). When the inner channel was full, the pump speed was returned to normal (10 mL/min), collecting the discharged sample in a Teflon beaker. When the sample reservoir was exhausted, the sample channel was emptied and then rinsed out at the highest pump speed with 25 mL of dilute acid or base (pH 2 or pH 10 depending on whether the analyte was chelated or not). The contents of the Teflon beaker were returned to the sample reservoir (100-mL volumetric) and diluted back up to 100 mL in 2% nitric acid for analysis. Although the rinse-out step is not essential for route operation (sample carry-over for 200 pg/mL solution of cadmium (Cd) was determined to be 0.035%), it was important during these studies t o assure that any loss of the analyte was due to passage through 0 1986 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 58, NO. 14, DECEMBER 1986
0
3241
channels filled. If the cell was to be stored for more than 2 days without operation, the salt and sample channels were stored with 4% formaldehyde.
I I
R E S U L T S AND DISCUSSION Sodium chloride was chosen as the concentrated salt for several reasons: it is relatively inexpensive, solid NaCl rapidly equilibrates in water without a large evolution of heat, the solid salt in a saturated solution does not cake, and the solubility changes only slightly with large changes in temperature (35 g/100 mL at 0 "C and 39 g/100 mL at 100 "C). Other salts offer greater solubilities (Le., lithium bromide (LiBr) 16.7 M v9. NaC15.7 M at 0 "C) and therefore greater permeation rates, but they do not seem to offer a net advantage over NaCl at this time. The stahility of the permeation rate is dependent on the concentration of the salt solution, the flow rate of the salt solution through the cell, and the temperature. The reproducibility of permeation rates is then a reflection of the ability to fix these three parameters. The concentration process inevitably causes the salt solution to be continuously diluted by the permeation of water from the sample. Saturating the salt solution reservoir with excess NaCl maintains a constant salt concentration in the reservoir. However, localized dilution can occur in the salt channels of the cell. High salt solution flow rates along with turbulent flow caused by the nylon mesh spacers maintain a nearly constant salt concentration within the salt channels. The salt solution reservoir, as described in the Experimental Section, is insulated to minimize the effects of fluctuations in the r w m temperature. With these three parameters fixed, the stability of the permeation rate with time is illustrated in Table I. At room temperature, permeation rates remain relatively constant from sample to sample for the same membrane cut and vary only slightly for different membrane cuts. This constancy indicates that the porosity of the membrane is uniform over the entire sheet purchased. When the temperature of the saturated salt solution is raised to 79 "C, permeation rates more than double as illustrated in Table 11. The advantage of operating the system at elevated temperatures is obvious. However, for the sake of convenience, the remainder of the data reported here were obtained at room temperature. Complexing the analyte ions with EDTA prior to osmostic Concentration provides dual mechanisms of anion repulsion (cellulose acetate membranes behave as net proton acceptors) (3) and size exclusion, which helps to prevent loss of the analyte by permeation through the membrane. A comparison of chelated and nonchelated analyte recoveries is shown in Table III. Chelation of the analyte is not absolutely necessary as seen in the recoveries of manganese (Mn) and nickel (Ni); however, recoveries from element to element vary only about 1%when complexed with EDTA, which is important if the technique is to be nonselective. The reproducibility for n o n c h e k d ion recoveries has been relatively poor (9.%4%), which leads us to believe that there are variables which have
I
I
I
Figure 1. Osmotic cor
Table I. Stability of membrane cut
,ne r e r m e a ~ ~ n an bem
permeation rate, mL/min after 2.0 min after 5.0 min 5.08 f 0.02 5.38 f 0.02 5.60 + 0.01 5.73 f 0.01
1 2 3 4
temp, "C
*
5.02 0.02 5.39 f 0.04 5.66 f 0.04 5.73 f 0.01
22 25 26.5 28
Table 11. Effect of Membrane and Temperature on the Permeation Rate membrane
cut
permeation rate, mL/min
temp, 'C
5.08 f 0.02 5.21 f 0.04 5.08 0.08 5.01 + 0.05 12.6 f 0.1
22 22 22 22 79
the membrane and not due to adsorption. Analyte enrichment data for the interfacing of the concentrating cell output to the nebulizer of an ICP were obtained in a similar fashion. Sample solutions were loaded into the concentrating cell at the highest pump speed. When the inner channel was full the pump speed was reduced tn 6.3mL, min. To a w e a steadystate permeation rate, the sample was allowed to concentrate for approximately 5 min prior to directing the sample output to the interface. About 0.5 ml. of the disrharged sample was allowed to collect in the interface (pipet tip) prior to s w i n g the n e h u l i r feed pump. Emission intensity wa9 recorded un a strip chart recorder. When the concentrating work was completed, the cell was flushed free of the salt solution with water and stored with the ~~~
~~~
Table 111. Effect of Cbelation, Concentration, and Membrane on Analyte Recoveries membrane C"t
1 1 1
1 2 3
andyte
concn, pg/L
EDTA chelation
200 200 20 2P 200 200
no yes
'Concentration in mg/L.
yes yes
yes yes
Cd
cu
90 f 4 94.2 + 0.6 93f1 95.3 f 0.2 94.9 f 0.3 94.4 f 0.6
92 f 4 93.3 0.1 90*2 94.8 f 0.3 94.4 f 0.4 93.4 f 0.5
70 analyte recoveries Mn Ni
95 f 2 94.0 + 0.4 96 f 1 95.6 f 0.4 94.7 f 0.3 93.9 + 0.1
95 1 93.0 + 0.4 97 1 95.0 f 0.2 95.0 0.9 94.2 + 0.6
*
Zn
a"
91 + 4 94.1 f 0.3 95 f 1 94.4 + 0.2 94.4 + 0.2 94.0 f 0.9
93 f 2 93.7 + 0.5 94 + 2 95.0 f 0.4 94.7 f 0.2 94.0 0.3
Anal. Chem. 1986, 58,3242-3244
3242
SAMPLE SOLUTION
-
I
I
OSMOTIC CONCENTRATION CELL
t '
SOLUTION SALT
I
RESERVOIR
,GT, j INTERFACE
SAMPLE
~
I
~
RESERVOIR
3.
NEBULIZER DELIVERY PUMP
~
ICP
1
Figure 2. Schematic diagram of osmotic concentrating assembly.
Table IV. Enrichment Factors for Osmotic Preconcentration"
LITERATURE CITED
concn, pg/L % enrichment element* initial measd calcd disagreement factor Mn
100
Zn Cd
100
91.6
839 826 722
848 848 179
1.1 2.6
7.3
the system the flexibility needed to interface it with a nebulizer of an ICP. The present interface was designed to eliminate the buildup of pressure in the sample channel. The delivery of the sample from the interface to the nebulizer was set a t 0.7 mL/min. The input of the sample into the concentrating cell was adjusted so that the sample output matched the flow rate to the nebulizer. Enrichment factors for the preconcentrator are illustrated in Table N. "he concentration factor for water removal by osmosis was 9.0; however, the enrichment factors were slightly lower due to the incomplete analyte recovery (93.7-94.7%). Enrichment factors for Mn and zinc (Zn) are in good agreement with the predicted values while Cd shows marginal agreement. One limitation of the technique a t this time is the back permeation of NaCl from the saturated salt solution. Even though rejection of NaCl by the membrane was greater than 99.98%, the sample solution after concentration contained about 3600 pg/mL Na. This salt concentration can cause problems in the measurement of analyte emission intensity. For example, we noted a deterioration in precision after prolonged nebulization into the ICP unit, presumably from the nebulizer fouling. Preliminary tests on other salts such as sodium hydrogen phosphate, sodium citrate, sodium tartrate, and magnesium chloride have shown a reduced back permeation of the salt with only a modest loss in the sample concentration rate. These investigations are continuing. Registry No. Cd, 7440-43-9; Cu, 7440-50-8; Mn, 7439-96-5; Ni, 7440-02-0; Zn, 7440-66-6.
8.39
8.26 7.88
"Sample input = 6.30 mL/min, sample output = 0.70 mL/min, concentration rate = 5.60 mL/min, concentration Factor = 9.0. *Sampleschelated with EDTA. not yet been identified. Analyte recoveries (chelated) for three different cuts of membrane also show good agreement having only a 2% range a t the 200 pg/L level. This agreement indicates that the pore size of the membrane is consistent throughout the entire sheet that was purchased. The percent recoveries for analyte concentrations (chelated) ranging from 20 gg/L to 20 pg/mL are in good agreement, indicating that analyte recovery is independent of the sample concentration. The ability to vary the rate of sample output from the concentrating cell by varying the rate of sample input gives
(1) Minczewski, J.; Chwastowska, J.; Dybczynski, C. Separation and R e concentration Methods in Inorganic Trace Analysis, Williamson: New York, 1982. (2) Mizuike, A. Enrichment Techniques for Inorganic Trace Analysis ; Springer-Verlag: Berlin, Heildelberg, New York, 1983. (3) Matsuura, T.; Pageau, L.; Souirajan, S. J . Appi. Polym. Sci. 1975, 19. 179-198.
R. J. Stec S. R. Koirtyohann*
Department of Chemistry University of Missouri Columbia, Missouri 65211
H. E. Taylor U S . Geological Survey Box 25046, MS 407 Denver Federal Center Denver, Colorado 80225 RECEIVED for review March 4,1986. Resubmitted August 20, 1986. Accepted August 20, 1986. Use of trade names does not imply endorsement by U.S. Geological Survey.
Capillary Gas Chromatograph Determination of Aniline Derivatives by Supersonic Jet Resonance Multiphoton Ionization Mass Spectrometry Sir: Supersonic jet expansion cools a sample molecule to several kelvin and it greatly simplifies the spectrum. Therefore, this spectrometric technique is advantageous for identification of molecular species. This method has already been used for analytical purposes (1,2). Polycyclic aromatic hydrocarbons such as benzo[a]pyrene are determined by fluorometry with the excitation sources of dye lasers (3-5) and xenon arc lamps (6-9). Multiphoton ionization mass spectrometry is also used for determinations of aromatic molecules
(10-16). More recently, a supercritical fluid sample introduction technique has been developed for measurements of nonvolatile substances (17, 18). Since the real samples, such as airborne particulates, contain more than 100 organic species, a sample separation procedure is necessary before the determination of these compounds. Gas chromatography is successfully used for this purpose because of good separation resolution and high sensitivity. Hayes et al. demonstrated the determination of naphthalene
0003-2700/86/0358-3242$01.50/00 1986 American Chemical Society
