Automatic reaction rate method for the determination of vicinal glycols

Sep 6, 1974 - (15) J.E. Lovelock, R. J. Maggs, and E. R. Adlard, Anal. Chem., 43, 1962. (1971). (16) . B. Singh, G. E. Kllnzing and J. Coull, AlChE, p...
0 downloads 0 Views 735KB Size
(7) Madbuli H. Noweir, Emil A. Pfitzer, and Theodore F. Hatch, Amer. lnd. Hyg. Ass. J., 33, 454 (1972). (8) Madbuli H. Noweir and Emil A. Pfitzer, Amer. Ind. Hyg. Ass. J., 35, 669 (1972). (9) R. Jeltes, E. Burghardt, and J. Breman, Brit. J. Ind. Med.. 28, 96 (1971). (10) D. Lillian, H. B. Singh, A. Appleby, and L. A. Lobban, J. Air Pollut. Conk Ass., submitted August (1974). (11) D. Lillian, H. B. Singh, L. A. Lobban, R. A. Amts, and R . L. Gumpert, 168th National Meeting, American Chemical Society, Atlantic City, N.J., September 1974. (12) L. J. Priestly, Jr., F. E. Critchfield, M. H. Ketcham, and J. D. Cavender, Anal. Chem., 37, 70 (1965). (13) J. A. Dahlberg and I. B. Kihlman, Acta Chem. Scand.. 24, 644 (1970). (14) D. Lillian and H. B. Singh, Anal. Chem., 46, 1060 (1974). (15) J. E. Lovelock, R. J. Maggs. and E. R. Adiard, Anal. Chem., 43, 1962 (1971).

(16) H. B. Singh, G. E. Klinzing and J. Coull, AlChE, presented at the 66th annual meeting, Philadelphia,Pa., 1973. (17) A. E. O’Keeffe and G. C. Ortman, Anal. Chem., 38, 760 (1966). (18) Madbuli H. Noweir, Emil A. Pfitzer, and Theodore F. Hatch, Amer. lnd. Hyg. Ass. J., 34, 110 (1973).

RECEIVEDfor review September 6, 1974. Accepted December 6, 1974. This project has been financed in part with Federal Funds from the Environmental Protection Agency under Grant No EPA: 800833. Paper of the Journal Series, Agricultural Experiment Station, Cook College, Rutgers University-The State University of New Jersey, New Brunswick, N.J. 08903.

Automatic Reaction Rate Method for the Determination of Vicinal Glycols with a Perchlorate Ion Selective Electrode C. H. Efstathiou and T. P. Hadjiioannou‘ Laboratory of Analytical Chemistry, University of Athens, Athens, Greece

An automatlc potentlometric reaction rate method is described and shown to be simple, rapid and accurate for the determination of ethylene-, propylene-, and butylene glycol. The vicinal glycol is oxidized with periodate, and the reaction rate is followed with a perchlorate ion selective electrode. The time required for the reactlon to consume a fixed amount of perlodate, and therefore for the potential to increase by a preselected amount (25.0 mV), is measured automatically and related directly to the vicinal glycol concentration. Ethylene-, propylene-, and butylene glycol concentrations in the range 1.4 X 10-3-7.0 X 10-3M in a total volume of 28 ml were determined with relative errors of about 0.7% and measurement times of only about 15 to 150 seconds. Also, a direct potentiometric method is described for the determination of ethylene glycol. The ethylene glycol reacts with a known excess of periodate, the unreacted perlodate is determined with a perchlorate ion selective electrode, and the amount of ethylene glycol is calculated from the amount of periodate consumed. The relative error for the determination of 6-9 mg of ethylene glycol was about 0.3%.

crease by a preselected amount (25.0 mV) is measured automatically and related directly to the vicinal glycol concentration. Commercial equipment and an auxiliary relay system are easily combined and provide automatic results shortly after the start of the reaction. Ethylene glycol, 1,2propylene glycol, and 2,3-butylene glycol concentrations in the range 3 to 18 mg per 28 ml of solution were determined with relative errors of about 0.7% with measurement times of only about 15 t o 150 seconds. Although the measurement method can be presented in a general way, details are presented here for the vicinal glycol method for which experimental data were obtained. The development of equations for other applications can be done in an analogous way. A direct potentiometric method is also presented for the determination of ethylene glycol. The ethylene glycol reacts with a known excess of periodate, the unreacted periodate is determined with a perchlorate ion selective electrode, and the amount of ethylene glycol is calculated from the amount of periodate consumed. The relative error for t h e determination of 6 t o 9 mg of ethylene glycol was about 0.3%.

Hitherto a small amount of work has been published which deals with the use of a perchlorate ion selective electrode, mainly for the determination of the perchlorate content of various salts, explosives, and solid propellants ( I 7 ) . Guilbault and Rohm (8) have reported an improved perchlorate electrode formed by placing the active exchanger in PVC. The perchlorate electrode response to periodate has already been reported ( I ) , but no analytical applications of this electrode property have appeared in the literature. T h e oxidative cleavage of vicinal diols is the classical and most widely used reaction of periodate. In this paper, it will be shown t h a t the perchlorate electrode response t o periodate can be used for the determination of certain vicinal glycols by a potentiometric reaction rate method. The vicinal glycol is oxidized with periodate; the rate of the reaction is followed with a perchlorate ion selective electrode and the time required for the reaction t o consume a fixed amount of periodate, and therefore for the potential t o in-

GENERAL CONSIDERATIONS

’ To whom correspondence should be addressed. 864

ANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975

Perchlorate Ion Selective Electrode as Periodate Sensor. The electrode employed in this study was an Orion Model 92-81 perchlorate ion-activity electrode. This electrode exhibits Nernst potential behavior for periodate a t concentrations of 10-1 t o 10-5M in the pH range between 4 and 6, in accordance with Equation I:

E = E’

-E In a, F

where E is the measured total potential of the system, E’ is the portion of the total potential due t o choice of reference electrodes and internal solution, R and F are constants, T is the temperature in degrees Kelvin and cy,- is the activity of the active periodate species x - . At higher pH, the linear part of the working curve (potential us. log [periodate]) and the sensitivity of the electrode decrease, and a t p H of about 8 or higher, the perchlorate electrode cannot function as a periodate monitor. In Figure 1, the perchlorate electrode potential behavior us. solution p H for a 1.8 X

intermediate between the glycol and periodate, followed by a slow decomposition of the intermediate, which is the rate determining step. The general reaction scheme is: G

+

K

IO4- ==

-

G*I04-

k

products

(5)

-

where G is a vicinal glycol, G IO4- is the intermediate, and K and k are the equilibrium and kinetic constants, respectively. Duke et al. (11, 12) have shown t h a t if glycol is in

I

large excess over periodate and the p H remains constant in t h e range 4-7, the kinetics of the reaction are expressed by Equation 6:

J \\ '

210t-

102

-

1 %

I

:: 6

7

8

9

lo

PH

Figure 1. Perchlorate electrode potential behavior in pure Na104 and NaC104 solutions and selectivity constant vs. pH ( A ) 1.8 X 10-4M Na104;(8) 1.8 X 10-4M NaC104;(C), selectivity constant

10-4M NaIO4 solution is given. For comparison, the potential for a 1.8 X 10d4M NaC104 solution is also shown. A reasonable explanation for this electrode behavior may be given by recalling t h e periodate species equilibria. Crouthamel et al. (9) have shown t h a t spectrophotometric data for periodate solutions are adequately described in the p H ranges 0-7 and temperature ranges 0-70' by the equilibria (the K values are for 25 "C)

H,IO,

=--L

H+

+

aH4106-

H,Io~-, K , =

- _-

where [GI is the concentration of t h e uncoordinated glycol and [PI is total periodate. In the p H range 4-7, practically all t h e periodate is in t h e form of monovalent periodate. Therefore we can write

+

[PI = [IO,']

[G*I04-]

(7)

For Equilibrium 5 we have [G.I04-] = flIO,-][G]

(8 1

Combining Equations 7 and 8,

[PI = [IO,-] (1 + f l G I )

(9 )

Differentiating Equation 9 and combining with Equation 6,

ffH5106

K is equal t o 5.74 and 4.32 for ethylene glycol and 1,2-propylene glycol, respectively (14 ). At sample glycol concentrations of about 0.1M or smaller (final glycol concentration of about 0.007M or smaller) and under proper conditions ( p H constant, low glycol concentration but in large excess over periodate), the second term in the denominator of Equation 10 becomes negligible and Equation 10 becomes 2.0 x 10-7 (4) From these data, it is seen t h a t the predominant periodate species in solution between p H 4-7 is the monoanionic dehydrated form IOd--, in constant equilibrium with whereas for p H about 8 or higher the predominant species is H3IOs2-. Therefore the large increase in potential a t p H of about 8 or higher (Figure 1) can be attributed t o the large decrease in t h e activity of t h e monovalent periodate, since the perchlorate electrode does not respond t o the species H3IOS2-. From Figure 1, it is also seen t h a t in the p H range 4 t o 7.5, the perchlorate electrode is slightly more selective toward periodate than perchlorate. From the potential data given in Figure 1, the selectivity constant Kclo,-,104- of the perchlorate electrode was determined at 1.8 X 10-4M solutions by t h e Separate Solution Method ( 1 0 ) . T h e values of K clod-,104- as a function of p H are included in Figure 1. In the p H range from 4 t o 6.5, the selectivity cons t a n t is equal t o about 1.2. Vicinal Glycols-Periodate Reaction Kinetics. T h e mechanism of periodate oxidation is still under active investigation. T h e kinetics of the oxidation of several vicinal glycols by periodate in aqueous solution have been studied extensively (11-13). The oxidation is known t o proceed via a rapid reversible formation of a singly-negatively charged

Rearranging Equation 11,

Substituting in Equation 1 a x - with f[IO4-], where f is a constant activity coefficient, and differentiating we have

d ln[IO,'] dt

E - --F d - RT dt

Combining Equations 1 2 and 13,

If it is assumed t h a t , for each sample, changes in potential during the measurement interval t 2 - t 1 = At result only from the specific vicinal glycol-periodate reaction, and t h a t the change of glycol concentration is negligible during the measurement interval, then the rate of change of the logarithm of periodate concentration (d In [IO,-]/dt) is constant throughout the interval and the total change AI3 in potential is given by Equation 15: ANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975

865

eo

30

TIME, (lec!

Flgure 2. Recorded curves of cell voltage vs. time for ethylene glycol-periodate reaction

Figure 3. Recorded curves of cell voltage vs. time for a 0.04M 1,2propylene glycol solution. Other conditions as under Procedure

A E =--

k'RT[G]Ai F

pH: ( A ) 4.0; (6) 5.0;(C) 6.0; ( D )7.0: (E) 8.0

(15)

Rearranging Equation 15,

[GI =

( S ) a f1

Since for all samples AE and h' are constants, Equation 16 demonstrates t h a t the glycol concentration is proportional to the reciprocal of the time required for the reaction to consume a constant amount of periodate and therefore for the potential to change by a constant amount. A knowledge of all the quantities enclosed in the parentheses of Equation 16 would permit a calculation of the glycol concentration from a measured time interval. However, it is difficult to determine all these quantities, and therefore the unknown glycol concentration is found from a working curve (reciprocal time us. glycol concentration) obtained with standards. Working curves can also be constructed as follows: if glycol is kept in large excess over periodate (8- to 40-fold excess for the data reported here), the term [GI in Equation 14 remains practically constant during a kinetic run and therefore the term 1E/At, t h a t is the slope of the potential us. time curve, is proportional to the glycol concentration, and thus a working curve (slope us. glycol concentration) can be obtained with standards. A working curve obtained from the recorded curves shown in Figure 2 was a straight line. Since the automatic measurement of time intervals is easier than the measurement of slopes, the former approach is used for the construction of the working curves. At sample glycol concentrations of about 0.1M or larger, the term K [ G ] cannot be omitted in the denominator of Equation 10 and this results in a negative deviation from linearity in the working curves. T h e upper limit of linearity is different for various glycols, because their K values are not the same. These deviations from linearity were used for the evaluation of the constants K and h a t various temperatures ( 1 4 ) . Optimal Concentration of Reagents. In any kinetic method, the choice of initial reactant concentrations is governed by the rate of the reaction, the experimental technique, and the sensitivity of the measurement system. Various initial periodate concentrations can be used. As has already been shown, Equation 16 is valid only if the periodate concentration is very small, relative to the glycol concentration. T h e periodate concentration chosen, 1.8 X 10-4M, is a compromise between this requirement and the 866

ANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975

lower limit of detection of the perchlorate electrode for periodate, which is about lO-5M. T h e response of the cell for a 1,2-propylene glycol solution a t various p H values is shown by the recorded curves in Figure 3. The reaction rate varied greatly with the p H of the reaction mixture. Similar curves were obtained with ethylene glycol and 2,3-butylene glycol solutions. p H control is essential in periodate oxidations, Any p H in the range 4-6 can be used; a p H of 4.0 was chosen because, a t this p H , the electrode signal is practically noiseless and better working curves were obtained. T h e acetate buffer was found satisfactory and no other buffers were tried. Temperature Control. T h e sensitivity and the accuracy of the kinetic determination of glycols are also affected by the temperature and the reaction time, and attempts were made to optimize these two parameters. T h e reaction rate has a large temperature coefficient and, therefore, the reaction is conducted in a thermostated cell. The equilibrium constant K (Equation 8) decreases as the temperature increases (11, 1 3 ) and, therefore, Equation 16 is valid over a wider concentration range a t higher temperatures. On the other hand, a t higher temperatures, the lifetime and the detection limit of the electrode decrease and the noise increases. As a compromise, a temperature of 34 0.1 "C was chosen. Voltage Interval. For any system, the voltage interval should be chosen large enough that the measurement error is small in comparison. Under the aforementioned experimental conditions the curve E = f ( t ) is linear for more than 60 mV. A voltage interval of 25.0 mV was chosen as a compromise between accuracy and measurement times and in order to ensure premeasurement times of a t least 10 seconds, and thus secure thorough mixing of the reagents. T h e premeasurement time is controlled by appropriate setting of a multiturn type potentiometer. Preparation of Working Curve. Working curves (straight lines plots of the reciprocals of measurement times us. glycol concentration) are rapidly obtained by using three to four standards. T o ensure reproducible results, all measurements were taken under constant stirring and illumination (diffuse light). Efficient magnetic stirring was provided to achieve steady and reproducible results, a t such a rate that no vortex or accompanying bubbles were formed in the solution. Light promotes the non-selective oxidation of certain compounds commonly produced by periodate oxidation of hy-

*

droxy compounds. For this reason, periodate oxidations are normally carried out in the dark ( 1 3 ) .

Determination of Ethylene Glycol by Direct Potentiometry. T h e method is based on t h e determination of periodate consumption. This is equal t o the difference between t h e known initial periodate concentration and t h e final one determined from a single potential measurement, multiplied by the final volume of the reaction mixture, and corresponds t o t h e periodate consumed for the oxidation of ethylene glycol. For better accuracy, t h e sample concentration range should be narrow (0.020-0.028M), and the usual precautions as in the classical iodometric procedure (15) should be taken to avoid overoxidation. Overoxidation is most noticeable when high concentrations of periodate are used, or when high temperatures are employed for the oxidation (13). If [ 1 0 4 - ] 0 is the initial periodate concentration (equal t o the periodate concentration in the blank), [IO4-], is t h e concentration of the unreacted periodate, and and E, are t h e measured potentials in the blank and in the sample. after the completion of the reaction, then

E , = E‘

- S log

[IO,-],

where S = 2.303 RTIF = constant. Combining Equations 17 and 18,

E , - E,, = S l o g w

(19)

T h e ethylene glycol concentration [GI is given by

(20) Combining Equations 19 and 20,

T h e ethylene glycol concentration can be calculated from Equation 2 1 from a single potential measurement. Since S depends not only on temperature b u t also on the age of the perchlorate electrode, all measurements are taken a t the same temperature ( 2 5 0.1 “C) and the value of S, which is equal t o the slope of the potential us. log [I04-] curve, is determined with a series of standards, with an accuracy of f0.05 mV. For better accuracy E b l should be t h e average of three determinations.

*

EXPERIMENTAL Instrumentation. Electrodes. An Orion perchlorate ion selec-

tive electrode, Model 92-81, was used as the indicator electrode. When not in use, the electrode is kept in a dilute NaC104 solution (-0.01M). This electrode is refilled about once each month. A perchlorate electrode in poly(viny1 chloride) matrix constructed by us in a manner similar to that described by Guilbault and Rohm (8) was also used and functioned satisfactorily. Before measurements are started, the electrode is immersed for about 10 minutes in a stirred dilute NaI04 solution (-O.OOlM), in order to take the periodate form and thus avoid small potential drifts. A double-junction silver-silver chloride electrode, Orion Model 90-02-00, was used as the reference electrode. The outer chamber of this electrode is filled daily with a 10% NH4N03 solution. The distance between the electrodes was rigidly maintained during all measurements. Reaction Cell. A double-walled 50-ml beaker is used as reaction cell. Water is circulated continuously through the reaction cell jacket with a thermostated pump. The reaction mixtures were stirred with the aid of a magnetic stirrer. Measurement and Contol System. The measurement and control system consists of the following units: a) An Analog-Digital Designer Heath/Malmstadt-Enke 801A with two operational amplifier circuit cards EU-900-NA and one

“r‘

$.+,-I b rN Co +O U N T E R

;; GENERATOR

-

GJ

Von-Su’pPlY

1

+ 5v

0.05yF

Figure 4. Simplified schematic diagram of the double-switching network and the counting system

relay card EU-800-JD. The pulse generator of the designer is adjusted to give 100 pulses per second in order to be used for the time counting circuit. b) A Universal Digital Instrument Heath/Malmstadt-Enke EU805A. This unit is used as a pulse counter during kinetic runs in combination with the pulse generator and as an accurate voltmeter during circuit alignment. c) A double-switching network with two integrated circuits type Philips TAA 560 and two 12-V dc relays. These integrated circuits are used as voltage level detectors (16). If the input voltages of these detectors are lower than a “threshold voltage” of 1.34-1.36 V, depending upon the particular detector, they conduct a current sufficient for the activation of the consequent relays. If their input voltages exceed the threshold voltage, they change into a practically non-conducting stage, thus deactivating the consequent relays. A diagram of the double-switching network and the time counting system is shown in Figure 4. The general operation of the system is as follows. All contacts are shown in their “zero time” positions. The operational amplifiers OAl and OA2 act as follower and inverting multiplier, respectively. When the imput voltage (cell voltage) reaches a predetermined value E l , such that E l multiplied by the factor (-R a/R 1 ) equals the threshold voltage of the first level detector LD1, relay RL1 is deactivated and consequently relay RL3 is activated and the time counting starts. When the cell voltage reaches another predetermined value Ez, such that S(S = E l - E l ) multiplied by (-Rz/R1) equals a voltage delay preset by proper adjustment of the potentiometer R3 (Multiturn type) of the dc offset unit inserted between the output of OA2 and the input of the second level detector LD2, the threshold voltage of LD2 is reached, relay RL2 is deactivated and, consequently, relay RL3 is also deactivated and the time counting stops. Thus a time interval is determined, the reciprocal of which is proportional to the glycol concentration. In the present work LE = 25.0 mV and E 1 is usually equal to 220 mV. The follower and the level detectors have high input impedance; therefore efficient shielding is necessary to prevent pick-up of electromagnetic noise which can activate or deactivate the double-switching network prematurely. Switch S should be closed at all times, except during measurements, to prevent relay contacts from injurious and noisy oscillations. Recording System. The Heath-Schlumberger Model EU-205B Recorder System was used for recording the response curve. The recorder was connected to points A and B of the measurement and control system (Figure 4). Reagents. All solutions are prepared in double-distilled water from reagent-grade materials and are kept in Pyrex bottles. Sodium Metaperi0date.a) Stock solution I, 0.02M: 4.28 g of NaI04 (G. F. Smith Co., Columbus, Ohio) are dissolved in water and diluted to 1 1. b) Solution 11, 2.0 X 10-4M. The solution is prepared fresh daily from solution I by appropriate dilution, and it is used in the kinetic method. It is not necessary to know its exact titer. c) Solution 111, 0.03000M, for direct potentiometry: 6.418 g of NaI04 dried for 1 hr at 120 O C are dissolved and diluted to exactly 1 1. in a volumetric flask. The oxidizing capacity of a weighed sample of sodium metaperiodate corresponded to a purity of better than 99.8%, and this made unnecessary the exact iodometric estimation of the titer. The solution is kept in amber bottles in the dark and is stable for at least one month if protected from light and organic material ( 1 7 ) . Acetate Buffer pH 4.0, prepared by appropriate mixing of 0.5M solutions of NaOH and CH3COOH. ANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975

867

Table I . Automatic Results for Aqueous Vicinal-Glycol Sblutions

Table 11. Results for Aqueous Ethylene Glycol Solutions by Direct Potentiometry

mmol vicinal g l col in 2-m1 rampie Vicinal glycol

Taken

Ethylene glycol

1,2 -Propylene glycol

2,3 -Butylene glycol

a

Ethylene glycol in 2 - m l sample

Error,

Foundo

?/r

0.0600 0.1100 0.1500 0,1900

0.0604 0.1090 0.1498 0.1888 Av.

-0.9 -0.1 -0.6 0.6

0.0500 0.0900 0,1080 0.1300

0.0504 0.0910 0.1070 0.1320 Av.

+0.8 11.1 -0.9 +1.5 1.1

0.0440 0.0700 0.1040 0.1760

10.7

0.0441 +0.2 0.0694 -0.9 0.1054 +1.3 0.1760 .. . Av. 0.6

Average of two determinations.

Sample

E,-

Taken,

E bl,

So.

E x , mV

mVa

Calcd, mmol

mg

la lb 2a 2b 3a 3b

140.8 140.8 152.8 153.0 175.7 175.6

27.2 27.2 39.2 39.4 62.1 62.0

0.0984 0.0984 0.1178 0,1180 0.1369 0.1368

6.08 6.08 7.30 7.30 8.51 8.51

Found, mgb

Error, %

6.11 +0.5 6.11 +0.5 7.31 +0.1 7.32 +0.3 8.50 -0.1 8.49 -0.2 Av. 0.3 a Eb, = Eo = 113.6 mV? average of three determinations. Calculated from Equation 21 modified as follows: mg G = 9.310 (1 - 1 0 - ( E . r - 6 b l ' " ) (9,310 = V ~ O ' M l o - . MU'G = 5 (0.03000) (62.07); S = 58.70 mV (per decade of concentration). ~

When the temperature reaches 25 f 0.1 "C, the Ion meter (Orion Ionanalyzer Model 801/digital pH) is read. Include the blank in triplicate and four periodate standards in the range 10-4-10-3M, containing 2.00 ml of buffer per 25 ml standard solution, with each series of unknown samples.

RESULTS AND DISCUSSION

0

I

2

4

6

myco. cohcEh-ar\Tioh

8

1c

N x1o2

Figure 5. Working curves for vicinal glycols determination ( A ) 1,2-propyleneglycol; ( B )2,3-butylene glycol; ( C )ethylene glycol

Ethylene-, I,Z-Propylene-, and 2,3-Butylene Glycoi (a mixture of meso- and racemic form) stock solutions 0.5M. The solutions

are prepared by dissolving the reagent in the appropriate amount of water. Their exact titer was estimated iodometrically by the Fleury-Lange method ( 1 5 ) . The solutions are stable for at least a week if stored in the refrigerator. Any aldehyde odor of these solutions is evidence of disintegration and such solutions should be rejected. Working standards, 0.02-0.04-0.07-0.10M, are prepared from the stock solutions by appropriate dilution. Procedure. Kinetic Method. Pipet 25.00 ml of sodium metaperiodate solution (Solution 11) and 1.00 ml of buffer into the reaction cell. Press the Start button on the Universal Digital Instrument (at the events counter mode), open switch S and pipet 2.00 ml of sample or standard vicinal glycol solution into the reaction cell. The analysis is completed automatically and the number on the digital readout (time in hundredths of a second) is recorded. Press the Reset button and close switch S, empty the cell with suction, rinse the electrodes and the cell with water and dry them with suction. Repeat the procedure for each analysis. Direct Potentiometry. Pipet 5.00 ml of ethylene glycol sample (in the range 0.020-0.028M), 5.00 ml of sodium metaperiodate solution (Solution III), and 2.00 ml of buffer into a 25-m1 volumetric flask and dilute quickly to the mark with water. Keep the volumetric flask in the dark for 30 minutes and then transfer its content into the thermostated reaction cell. Immerse the electrode and a thermometer into the solution and start the magnetic stirrer. 868

ANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975

Analysis of aqueous ethylene glycol, 1,2-propylene glycol, and 2,3-butylene glycol solutions gave the results shown in Table I. T h e average error was about 0.7% and measurement times were about 15 t o 150 seconds. Ten replicate determinations were made for a 0.04M propylene glycol sample; the average measurement time was 26.84 sec (range = 0.75 sec) and t h e relative standard deviation was 0.86%. Typical working curves are shown in Figure 5 . From t h e slopes of the working curves, it is seen that the reaction rate is larger for propylene glycol than for ethylene glycol, in agreement with data in the literature (12, 18). Results for the determination of ethylene glycol by direct potentiometry are given in Table 11. T h e relative error for the determination of 6 to 9 mg of ethylene glycol was about 0.3%. It can be seen from Tables I and I1 t h a t the obtained results substantiate t h e validity of Equations 16 and 21. Any substance which reacts with periodate would interfere and should be eliminated prior to t h e measurements. T h e kinetic method is fast and accurate and it eliminates t h e usual precautions which should be taken with t h e iodometric method (15) or t h e proposed method for t h e determination of ethylene glycol by direct potentiometry. T h e direct potentiometric method is more accurate than t h e kinetic method b u t it requires t h a t t h e ethylene glycol concentration fall within a rather narrow range. Although the application reported here deals with t h e determination of certain vicinal glycols, t h e scope of t h e method is intended to be more general. T h e rapid response of t h e perchlorate electrode to changes in periodate concentration makes it a valuable sensor for following the rate of periodate redox reactions. Work in progress indicates that various 1,2-amino alcohols can be determined in a n analogous way. Also various elements can be determined in the p p b range on t h e basis of their catalytic action on periodate oxidation reactions (19). T h e perchlorate electrode has also been successfully used in catalytic titrations involving periodate in the indicator reaction (20 ).

ACKNOWLEDGMENT T h e authors are grateful t o E. McNelis for stimulating discussions and to M. A. Koupparis for technical assistance.

LITERATURE CITED (1) R. J. Baczuk and R. J. Dubois. Anal. Chem., 40, 685 (1968). (2) T. M. Hseu and G. A. Rechnitz, Anal. Lett., 1, 629 (1968). (3) 40. - , C- . .I-. Contznn - - - .- - - and -. - H.. Frniser. . .- -. , Anal. - . Chem.. - . - , 2071 - - (1968). (4) M. J. Smith and S. E. Manahan, Anal. Chim. Acta, 48, 315 i1969). (5) R. F. Hirsch and J. D. Portock, Anal. Lett., 2, 295 (1969). (6) N. lshibashi and H. Kohara, AnalLett., 4, 785 (1971). (7) W. Selig and G. L. Crossman, Informal Report UCID-15623, Lawrence Radiation Lab., Livermore, Calif. (8) T.L. Rohm and G. G.Guilbault, Anal. Chem., 46, 590 (1974). (9) C. E. Crouthamel, A. M. Hayes, and D. S.Martin, J. Amer. Chem. Soc., 73, 82 (1951). (10) G. J. Moody and J. ID. R Thomas, "Selective Ion Sensitive Electrodes," Merrow Publishing C.O.,Watfort Herds, England, 1971, p 10. ( 1 1) F. R. Duke, J. Amer Chem. SOC.,69, 3054 (1947). (12) F. R. Duke and V. C. Bulgrin, J. Amer. Chem. SOC.,76, 3803 (1954). (13) G. Dryhurst. "Periodate Oxidation of Diol and Other Functional Groups," Pergamon Press, London, 1966, p 121. ~

I

I

~~~

(14) C. H. Efstathiou, T. P. Hadjiioannou, and E. J. McNelis, unpublished work, Laboratory of Analytical Chemistry, University of Athens, Greece, 1974. (15) P. F. Fleury and J. Lange, J. Pharm. Chim., 17, 107, 196 (1933) (161 PhiliDs Pocketbook. 1971. D 8220. (17) D. j. B. Galliford, R. H. Nuttall, and J. M. Ottaway, Talanfa, 19, 871 (1972). (18) G. J. Buist, C. A. Bunton, and J. H. Miles, J. Chem. SOC.4567 (1957). (19) T. P. Hadjiioannouand C. H. Efstathiou, unpublished work, Laboratory of Analytical Chemistry, University of Athens, Greece, 1974. (20) T. P. Hadjiioannou. M. A. Koupparis, and C. H. Efstathiou, unpublished work, Laboratory of Analytical Chemistry, University of Athens, Greece, 1974.

RECEIVED for review July 29, 1974. Accepted November 7 ,

1974. This research was supported in part by a research grant from the Greek National Institute of Research.

Interactive-Experimentation Employing Ion-Selective Electrodes Jack W. Frazer, Arthur M. Kray, Walter Selig, and Robert Lim Lawrence Livermore Laboratory, University of California, Livermore, CA 94550

An automated titration system has been assembled which consists of ion selective electrodes (ISEs), automatic buret, Interface hardware, minicomputer, cathode-ray tube (CRT) display, light-pen, teletypewriter, and a special function panel. Each of the various parts of the system was designed to perform the task for which it is best suited, with provisions for powerful scientist-computer interactive techniques. The system software changes which are often required for methods development and semiroutine determinations are made in a high-level language FOCAL. Titrations using ISEs as detectcrs and CRT display of the titration curve, Gran plot, second derivative and error function, together with interactive data-reduction techniques, are discussed. Least-square regression fitting was used for extrapolation of the equivalence points.

Titration systems employing ISEs as detectors have become very popular judging from the prolific literature on the subject. Anion and cation selective electrodes measure the activity rather than the concentration of the species of interest. T h e equivalence point for the titrations is usually determined by finding the maximum slope of the potentialvolume titration curve. T h e method of maximum slope can only be used to evaluate the equivalence point if a well-defined titration break is displayed. In addition, the method is subject to considerable systematic errors ( 1 ) . T h e electrode potentials art' usually unstable and subject to drifting near the end point because the level of ions being sensed is very low and the establishment of equilibrium conditions may be slow. Also, in low level titrations, the solubility of the precipitate formed or the dissociation of the complex formed may become significant near the end point. These problems may give intrinsic end-point errors (2, 3). Gran ( 4 ) has shown that a linear presentation of the titration offers a number of advantages. T h e method presented here is similar and involves antilogarithmic conversion of the titration data followed by a straight-line regression analysis of selected sections of the plot to determine the equivalence point. One of the advantages is t h a t information from a large portion of the titration curve can be exploited, not jus1 the small area near the equivalence

point as in conventional titrations. Gran's method has not been widely used because the calculational time required usually offsets the advantages. A volume-corrected graph paper for obtaining linear plots is available from Orion ( 5 ) ; however, the required point-by-point plotting method is very tedious. Contrary to the opinion expressed in Orion Newsletters, ( 5 ) , Eriksson (6) pointed out t h a t greatest precision and accuracy is obtained with the largest number of data points. Furthermore, for dilute solutions, points deviating from the straight line must be rejected or a systematic error will occur. T h e use of an on-line computer to process and display the data obviously affords a considerable gain in time, accuracy and precision when using Gran's method. Leastsquare curve fitting for a large number of data points is extremely tedious and time-consuming without the aid of a computer. In addition, incorporation of interactive graphics into the computerized system allows the chemist to rapidly apply advanced data reduction techniques. On-line computers have been used for automation of titrations (7-9). T o date. the greatest use of these computers has been for data acquisition, to provide rapid on-line calculations and generate reports. Now, however, there is an increasing trend toward use of the computer for control of instruments and experimental equipment. In this paper, an automated titration system employing ISEs as detectors is described. Because of the non-routine nature of our samples, we chose to design a system to provide easy-to-use hardware and to permit powerful scientistcomputer interactive techniques. T h e various parts of the integrated automated system will each perform the task for which they are best suited. For example, the computer will perform all the routine operations which can be predefined: i.e., mathematical calculations, optimization, data acquisition, data correlation, control, and report generation. T h e tasks can also include higher level logic such as t h a t contained in pattern recognition and learning-machine techniques-but note that all these are predefinable algorithms. In short, the computer will perform routine operations and functions that require a series of predefined steps to be performed in a rapid sequence. T h e scientist is thus free to perform those functions for which he is most suited: observation of unusual events and creative thinking. T h e ANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975

869