Behavior of a micro flowthrough copper ion-selective electrode system

Definition of the response time of ion-selective electrodes and potentiometric cells. Erno. Lindner , Klara. Toth , and Erno. Pungor. Analytical Chemi...
0 downloads 0 Views 473KB Size
gram should be careful not to add species to a particular model merely in an attempt to improve the statistical fit of the data. If all prior treatment techniques are inapplicable, it is possible systematically to examine the error surface with a variety of models. In this approach, the sum of squares of the residuals (SSR) is calculated for each constant over a predetermined range of values. The initial guesses would then be those constants giving the lowest SSR. This approach (although unnecessary) was used for the nickel-ethylenediamine data. It was effective in locating a suitable set of P's but was very slow and is obviously inefficient. This program has also been applied to analysis of spectrophotometric data obtained for the reaction of picoline2-aldehyde thiosemicarbazone with iron(II), cobalt, mercury(II), and silver. The results of these investigations will be published shortly.

LITERATURE CITED (1) C. W. Childs, P. S. Hailman, and D.D.Perrin, Talanta, 16, 1119 (1969). (2) F. J. C. Rossotti, H. S. Rossot!i, and R. J. Whewell, J. lnorg. Nucl. Chem., 33, 2051, (1971).

(3) J. J. Kankare, Anal. Chem., 42, 1322 (1970). (4) J. P. Chandler, "Minimum of a Function of Several Variables", Program 66.1, Quantum Chemistry Program Exchange, Indiana University, Bloomington, IN, 1966. (5) K. Nagano and D. E. Metzler, J. Am. Chem. Soc., 89, 2891 (1967). (6) A. Sabatini, A. Vacca, and P.Gans. Talanta, 21, 53 (1974). (7) P. Gans and A. Vacca, Talanta, 21, 45 (1974). (8)W. C. Davidon, U S .At. Energy Comm. Rept., ANL-5990 (1959). (9) R . Fletcher and M. J. D. Powell, Computer J., 6, 163 (1963). (10) I. G. Sayce, Talanta, 15, 1397 (1968). (11) S. Feldberg. P. Klotz, and L. Newman, lnorg. Chem., 11, 2860 (1972). (12) L. G. Silbn, Acta. Chem. Scand., 18, 1085 (1964). (13) T. Kaden and A. Zuberbuhler, Talanta, 18, 61 (1971). (14) J. R. Siefker, Anal. Chim. Acta, 52, 545 (1970). (15) V. Kuban and J. Havel, Scripta Fac. Sci. Univ. &no, Chemia 2, 1, 87 ( 197 1). (16) V. Kuban, Scripta f a c . Sci. Univ. Brno, Chemia 2,2, 81 (1972). (17) M. C. Chattopadhyaya and R. S. Singh, Anal. Chim. Acta., 70, 49 (1974). (18) F. J. C. Rossotti and H. Rossotti, "The Determination of Stability Constants", McGraw-Hill. New York, 1964. (19) D.Leggett, J. Chem. Educ., 51, 502 (1974). (20) L. G. Sillen and A. E. Martell, Ed. "Stability Constants of Metal-ion Complexes'', Chemical Society, London, 1964. (21) H. S. Rossotti, Talanta, 21, 809 (1974). (22) L. G. Sillen, Acta Chem. Scand., 16, 159 (1962).

RECEIVEDfor review November 12, 1974. Accepted February 3,1975.

Behavior of a Micro Flowthrough Copper Ion-Selective Electrode System in the Millimolar to Submicromolar Concentration Range W. J. Blaedel and D. E. Dinwiddie Department of Chemistry, University of Wisconsin, Madison, WI 53706

The behavior of a copper ion-selective electrode in a flowing system is investigated in the range 10-3-10-9M of copper Ion. Steady state potentials are independent of electrode pretreatment, and are Nernstian down to 10-8M, but below steady-state potentials are attained only after long periods of flowing solution contact with the electrode. Electrode and flowthrough cell construction is described.

I t has been shown that a t submicromolar concentrations, a copper ion-selective electrode can change the copper concentration of the solution in which it is immersed ( I , 2 ) . Therefore, for the measurement of low copper ion concentrations, a micro flowthrough electrode would seem to possess a marked advantage over the conventional batch technique. Some studies have been made of the behavior of the copper ion-selective electrode in flowing solutions. Thompson and Rechnitz ( 3 ) have described a fast flow system containing a fluoride ion-selective electrode with rapid response and Nernstian behavior down to about 10-7M. They have also designed and studied the behavior of heavy metal ion-selective electrodes in flowing systems (4).For the copper ion-selective electrode, they found Nerstian response down to 10-6M copper ion, with equilibrium potentials being achieved rapidly for freshly polished electrodes. Rechnitz and coworkers have also described the advantages of making potentiometric measurements with ion-selective electrodes in a differential mode, so that the junction potentials are compensated for, and so that external reference electrodes are not needed (5, 6). Llenado and Rechnitz (7) 1070

ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975

have devised a continuous system for glucose analysis, using a flowthrough iodide ion-selective electrode, based on the glucose oxidase reaction. Alexander and Rechnitz have described an automated method of protein determination with a flowthrough silver ion-selective electrode (8). Flowing systems designed for the implementation of standard addition techniques have been described in the literature (9, 1 0 ) . I t has recently been reported that the injection of copper and silver ions from the electrode into the solution must be recognized a t copper ion levels of 1 ppb (around 10-8M) ( 2 ) . This paper describes the construction of a gravity-fed micro flowthrough electrode system fabricated from commercially manufactured Ag2S-CuS pellets. The behavior of the system over a wide range of copper ion concentration is described.

EXPERIMENTAL F a b r i c a t i o n of the M i c r o C o p p e r I o n - S e l e c t i v e E l e c t r o d e s . Figure 1is a schematic of the electrode, which consists of an Ag2SCuS chip epoxied into the end of a threaded acrylic rod. As received, the Ag2S-CuS pellets were 716 inch in diameter and 3/32 inch thick (Orion Research, Cambridge, MA). The pellet came pressed onto a disk of silver metal, which was pried away. The pellet was cut into quarters with a jewelers saw, and then each quarter was cut into thirds, giving 12 chips per original pellet. The electrode body was an acrylic rod, 2 inches long and 0.25 inch in diameter, center bored to accept the silver electrical contact wire, and with a cavity drilled into one end to accept the AgzSCuS chip. The silver wire was cemented to the back side of the chip with conducting gold-filled epoxy (Epo-Tek H 80, Epoxy Technology Inc., Watertown, MA). The silver wire, once inserted

-E

-

Figure 2. The Flowthrough Cell

--D

( A ) Outlet

port for the bridging electrolyte solution, used to start the siphon and eliminate air bubbles, then closed: ( B ) Sample solution outlet channels; (C) Sintered glass frit bridge, epoxied in position: (0)Electrode cavity, electrode not shown: ( E ) Bridging electrolyte solution inlet channel to the center of the glass frit: (FJSample solution inlet channels: and (G) Acrylic cylinder, 2 inches in diameter by 1 inch high

Figure 1. The micro copper ion-selective electrode ( A ) Ag,S/CuS chip: (S) Conductive epoxy: (C) Insulating epoxy: ( D ) Acrylic body, 2 inches X 0.25-inch rod: and (/E)Silver electrical contact wire

into the acrylic rod, held the attached chip centered in the cavity machined for it while insulating epoxy (Epo-Tek 349, Epoxy Technology Inc., Watertown, MA) was added, and while it cured. After the insulating epoxy had cured, the acrylic barrel of the electrode was machined and threaded, and the end was faced off to give a flat exposed surface of the Ag&CuS chip. The Flowthrough Cell. Figure 2 is a schematic of the cell, fabricated from a Plexiglas cylinder, 1 inch long and 2 inches in diameter. The cell was designed and built with the capability to function either in a single channel mode, with the potential of the ionselective electrode in one flow channel being measured vs. a reference electrode, or in a dual channel mode, with the potential of the ion-selective electrodes in each flow channel being measured differentially against each other. While the advantages of the dual channel mode are considerable (61, it was not required and, therefore, was not used for the work reported in this paper. All of the following work was done in the single channel mode. The bridging cavity was drilled crosswise through the cell block so it would intersect the flow channels downstream from the electrodes when the channels were drilled later. A small cylinder of fine sintered glass long enough to extend into both channels was cut from a sintered glass disk and coated with a thin layer of very viscous epoxy (Epo-Tek 380, Epoxy Technology Inc., Watertown, MA) which was allowed to cure. This layer of cured epoxy covered the glass frit and prevented the thin insulating epoxy that was used to fix the frit in the bridging cavity from seeping into the frit before curing occurred. After the glass frit was in place and the surrounding epoxy (Epo-Tek 349, Epoxy Technology Inc., Watertown, MA) had cured, the solution channels were drilled as shown in Figure 2, with each outlet channel passing through opposite ends of the frit. The bridging solution channel was then drilled between the two outlet flow channels and through the center of the frit. Electrolyte solution then could flow slowly through the bridging frit and into the outlet solution channels to waste. All solution channels were drilled and tapped to accept Yk-inch by 28 liquid chromatography connectors (Chromatronix, Berkeley, CAI, which were used for all solution line connections to the cell. Finally, the holes for the electrodes were drilled and tapped. Care was taken to have the pellet surface extend just into the flow channel. The geometry of the solution channels was arranged to minimize horizontal or downward-directed segments, to decrease the chances of trapping any inadvertently introduced air bubbles. For the same reason, flow of solutions through the cell was from bottom to top. The angle of the flow channels through the cell provided improved mixing and turbulence a t the electrode surface, and allowed for a better butt seal between the end of the electrode and its seat, to prevent solution leakage around the electrode.

Figure 3. Equipment configuration and flow train (Only one sample channel shown for clarity) ( A ) Flowthrough

cell: ( B ) Flow rate adjusting stopcock: (C) Rotameter for measuring flow rate: (0)Moveable Teflon inlet siphon tube for sample, standard, and wash solutions: (0Teflon inlet siphon tube for bridging solution electrolyte: (0Copper ion-selective electrode: (G) Reference electrode: (H) Sample, standard, and wash solution containers: (0Container for bridging solution: (4 Waste container: and ( K ) Shielded cabinet

Solution Flow System. Figure 3 is diagram of solution flow paths and control components. Flow rate was maintained by gravity, controlled by a Teflon stopcock, and measured with a conventional rotameter-type flow indicator. Both the stopcock and rotameter were located downstream from the electrodes. The silver-silver chloride reference electrode (Model 90-01, Orion Research Inc., Cambridge, MA) was immersed in the beaker of electrolyte solution that was connected via a Teflon tube siphon to the cell inlet leading to the center of the glass frit bridge. Switching of sample and standardizing solutions was done manually, by transfer of the siphon inlet tube from one beaker of solution to another, with careful wiping off of the tube between beakers. The apparatus was mounted on a board with a shelf to hold the solution beakers. The mounting board comprised the back of a shallow (4 inches deep) open-front cabinet. All sides of the cabinet were electrically shielded with grounded aluminum wire screening. A wire screen curtain could be dropped to cover the open face of the cabinet when deemed desirable. Complete shielding (wire screen curtain down) was required only for dual channel mode operation, or for single channel mode runs with sample solutions containing no supporting electrolyte. Without the shielding, electrical noise pulses up to 30 mV were observed, and these were reduced to about 0.5 mV by the shielding. Procedure for the Measurement of Potentials. Potentials were measured using a pHlspecific ion/mV meter (Leeds and Northrup, Model 7410, North Wales, PA) whose output was connected to a Sargent Model SR recorder. Since the recorder had a restricted range of chart speeds, runs over long periods of time were recorded for a minute or so a t selected times, with the chart stopped a t all times in between. Sample solutions were put in plastic beakers (Tri-Pour, T. M. ANALYTICAL CHEMISTRY, VOL. 4 7 , NO. 7, JUNE 1975

1071

T a b l e I. Interferences in t h e S u p e r m i c r o m o l a r Concentration R e g i o n Error

Interference

In potential,

In concentra-

mVe

tion, %

'

In pCu

0.0 0 0 None CaC12 0.0 0 0 Z nC l2 0.0 0 0 17 .07 FeC13 +2 AlC13 +1 8 .03 a The tabled potentials are the potentials of 10-6M Cu(NO& in 0.01M KC1, with the interferent added to lO-5M. The standard deviation of the potential reading is 0.5 mV, which corresponds to an error of 0.02 in pCu, or to a relative error of 4% in concentration. * The error in concentration is that in the Cu(I1) concentration that would be calculated from the measured potential and the Nernst equation,

-

-+ -

10-9

-100

I

I

I

2

I : , I

3

'

8

I

I 8 I I

12 16 24 TIME, HOURS "

I

I

I

36

48

60

'

g

I

72 84

Figure 5. Electrode response in the submicromolar concentration region

Sherwood Medical Industries, Inc.) which were segregated for use at each decade of concentration. Switching between samples, washes, and standard solutions was accomplished by stopping the solution flow at the stopcock, removing the Teflon siphon tube from the beaker of solution that had been flowing, wiping with a paper tissue, inserting the tube into the beaker of solution to be passed through the cell, and then turning the stopcock on again to achieve the proper flow rate. This procedure avoided drawing air bubbles into the lines, which sometimes proved a nuisance to dislodge. Most measurements were made at flow rates around a milliliter per minute. At steady-state, the potential was independent of flow rate over a moderate range of flow rates, 0-3 ml/min. Calibration was performed by setting the meter to zero millivolts with lO-5M Cu(N03)pO.OlM KC1 passing through the apparatus, unless specifically stated otherwise. Calibration was followed by the sample, also in 0.01M KC1, or by 0.025M H2S04 which was used to clean the system just before the sample was run. Before the start of a new series of experiments, the electrode was usually given a fresh surface by a light hand polishing on wet U1tralap paper ( 3 Wm, Pfizer, Minerals, Pigments and Metals Division, New York, NY) and then polished to a mirror surface with 0.3 pm alumina (Fisher Scientific Co., Fairtown, NJ). The effect of such resurfacing was to maintain an acceptable speed of approach to steady-state. Without such resurfacing, the electrode response rate slowly deteriorated over long periods of use or idleness (weeks) t o a marked sluggishness. During use of the equipment in the single channel mode, 0.01M KC1 was passed slowly through the unused channel. Generally, it is desirable to keep the bulk composition and concentration of the electrolyte solution in the two sample channels and in the refer1072

ence electrolyte channel as much the same as possible. Otherwise, small drifts and fluctuations of potential occurred, which were ascribed to uncontrolled changes in composition and/or concentrations at the bridging glass frit boundaries. Reagents. All solutions were 0.01M in KCI, prepared as described previously ( I ) from reagent grade chemicals and triple distilled water, by serial dilution of a master 0.100M KCI solution in sets of volumetric flasks (glass for 10d5M Cu(I1) or higher, and Nalgene plastic for 10-6M Cu(I1) or lower), reserved for each decade level of Concentration. It must be emphasized that the concentrations given in this paper are nominal ones, representing the concentration of copper added in preparing the solutions, and do not include reagent or water blanks. Supporting electrolyte was 0.01M KCl, except where mentioned otherwise. The reference electrode filling solution was saturated with respect to KCl and AgCI, and the bridging electrolyte solution was 0.01M KCl.

ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975

RESULTS A N D DISCUSSION

S y s t e m Response to Supermicromolar Concentrations. T i m e of response t o steady-state potentials in t h e supermicromolar region was found t o be similar t o that reported elsewhere for flowing systems ( 3 ) .Steady-state potentials plotted vs. t h e log of t h e copper ion concentration (Figure 4) show a response very close t o Nernstian over t h e range 10-3-10-6M. T h e figure also shows that t h e working curves for two different supporting electrolytes a t different concentrations are almost identical, providing that t h e calibrating solution has t h e s a m e supporting electrolyte at t h e s a m e concentration as t h e sample. If a single calibrating solution is employed, however, different supporting electrolytes a t different concentrations will n o t generally produce t h e s a m e working curves. System stability a n d reproducibility in the supermicromolar region were very good. Once calibrated with the 10-5M standardizing solution, t h e electrode could be cleaned repeatedly with acid a n d even removed for polishing normally without altering t h e calibration point b y more t h a n 0.5 mV. S u c h stability could b e maintained for days with proper care. A brief interference s t u d y was m a d e of t h e error caused in t h e determination of the concentration of a 10-6M Cu(I1) solution in t h e presence of a 10-fold higher concentration of a few other cations. T h e results of this s t u d y ( T a b l e I) indicate relative errors below 0.1 pCu, or below 20% in concentration for all of t h e interferences studied. System Response t o Submicromolar Concentrations. Figure 5 is a series of curves showing t h e time response of t h e electrode system with submicromolar solutions of Cu(I1) in 0.01M KCl. T h e system was initially brought t o steady s t a t e in 10-6M Cu(II), a n d this level was t a k e n as zero millivolts. T h e n t h e other solutions were r u n sequentially through t h e cell in t h e order of decreasing concentra~

0

>

E

>

E

-1001 10-9

I

I.

10-8 10-7 NOMINAL COPPER MOLARITY

106

Flgure 6. Steady-state response potentials in the submicromolar concentration region The points shown represent replicate determinations; lO-'M, 6 replicates, range 1 mV: lO-'M, 5 replicates, range 1 mV; and 10-gM,2 replicates, range 3 mV

'

-100

L '

0

TIME,

HOURS

;I

24

36 418 6d

Flgure 7. Electrode response in the submicromolar concentration region after cleaning to -60 mV in 0.025MH2S04

tion until each came to steady-state. After steady-state had been reached in the 10-9M solution, the sample solution sequence was reversed, so that each succeeding solution was more concentrated than the previous one. The response time to steady-state for the series of runs going in the order of increasing concentration were similar to those shown in Figure 5. Within about 1 mV, the same steadystate potentials were reached, but the data are not shown in Figure 5. I t is apparent from Figure 5 that even under the best of conditions, very long times are required to reach steady-state a t low concentrations of copper ion. Figure 6 is a semi-log plot of steady-state potential vs. nominal copper ion concentration, with the data points representing replicate sets of determinations like those in Figure 5. Fewer data points were taken for the more dilute solutions because of the long times required. The potential readings for 10-6-10-8M solutions lie on a straight line with a slope of 29.3 mvldecade, which is very close to Nernstian response (29.6 mV/decade). The potential of 10-9M solutions is about 7 mV above Nernstian, and may be accounted for by postulating a blank of 0.7 X 10-9M Cu(I1) in the water and supporting electrolyte used to prepare the solution. This postulated blank was not verified by independent means. Figure 7 represents the achievement of equilibrium potentials for samples run like those in Figure 5, except that the electrode was acid-cleaned after each sample. Within experimental error, the same equilibrium potentials were reached as in Figure 5, with very long times needed to achieve steady-state. No explanation is proposed for the maxima in the curves of Figure 7 .

CONCLUSIONS For highly different modes of pretreatment, ranging from exposure to high copper ion concentrations, low copper ion concentrations, and acid cleaning solutions, a steady-state potential is approached on immersing the electrode in a solution of copper ion that appears to be Nernstian within 1 mV down to copper ion concentrations of 10-8M. The deviation from Nernstian potential of the nominal 10-9M solution corresponds t o a blank copper concentration of about 1 x 10-9M.

Unless they can be reduced, the long times required to reach steady-state potentials a t low copper ion concentrations negate the analytical usefulness of the steady-state potential as a measure of the copper ion concentration below lo-". It is apparent from Figure 5 that the rate of approach to the steady-state potential is related to the copper ion concentration of the solution in which the electrode is immersed. But Figure 7 compared with Figure 5 shows that the rate of approach to steady-state potential also depends on the pretreatment of the electrode, which .confirms earlier observations ( I ) . The analytical usefulness of this rate phonomenon is being investigated in the submicromolar concentration range.

ACKNOWLEDGMENT We acknowledge the cooperation of D. R. Keeney of the University of Wisconsin Soils Department for the extended use of his electrodes. Appreciation is also expressed to Orion Research, Inc., for donating two copper ion-selective pellets for our use.

LITERATURE CITED ( 1 ) W. J. Blaedel and D. E. Dinwiddie. Anal. Chem., 46, 873 (1974). (2)R. Jasinski, I. Trachtenberg, and D. Andrychuk. Anal. Chem., 46, 364 (1974). (3) H. I. Thompson and G. A. Rechnitz, Anal. Chem., 44, 300 (1972). (4)H. Thompson and G. A. Rechnltz, Chem. Instrum., 4(4), 239 (1972). (5) M. J. D. Brand and G. A. Rechnltz, Anal. Chem., 42, 616 (1970). (6) R. Wawro and G. A. Rechnitz, Anal. Chem., 46, 806 (1974). (7)R. A. Llenado and G. A. Rechnltz, Anal. Chem., 45, 2165 (1973). (8)P. W. Alexander and G. A. Rechnitz, Anal. Chem., 46, 860 (1974). (9)J. M. Riseman, Wafer Qual. Instrum., 1, 89 (1972). (10)J. P. Elder, 25th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, March 7, 1974.

RECEIVEDfor review November 4, 1974. Accepted February 18, 1975. This research was presented in part a t the Tenth Midwest Regional A.C.S. Conference, Iowa City, IA, November 8, 1974. This research was supported in part by an Office of Water Resources Research Grant, No. A-053WIS. DED is a member of the Air Force Institute of Technology Education Program.

ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975

1073