LITERATURE CITED
(1) Blaedel,
W.J., Knight, H . T., .4NAL. CHEM.26,741 (1954). (2) Diehl, H., Buchanan, E. B., Smith, G. F., Ibicl., 32, 1117 (1960). (3) Diel4 H., Smith, G. F., “The Iron Reagents,” G. Frederick Smith Chemical Co., Columbus, Ohio, 1960. (4) Feigl, F., “Spot Tests, I, Inorganic Applications,” 4th ed., pp. 153-8, Elsevier, New York, 1954. ( 5 ) Kolthoff, I. M., Belcher, R.: “Volumetric Analyses,’’ Vol. 111, Interscience, New York, 1957.
( 6 ) Iioros, E., Barcza, L., Chemist Analyst 48,69 (1959). (7) Laitinen, 11. A., Hall, L. C., ANAL.
CHEX29,1390 (1957).
(8) Lucchc.si, C. .4,> Hirn, C. F., Ibid., 32, 1191 (1960). (9) Lundell, G. F. F., Bright, A. A., Hoffman, I. F., “Applied Inorganic .4nalyses,” 2nd ed., Chap. 21, John Wileg BE. Sons, New York, 1953. (10) Morrison, G. H., Freiser, H., “Solvent Extractions in Analytical Chemistry,”
John \%ley K: Sons, New York, 1957. (11) Reilley, C. N., Schmid, R. W., -\SAL. CHEM. 30,947 ( 1 958).
(12) Ringborn, >i., J . Chem. Ecluc. 35, 282 (1958). (13) Ringborn, A., “Treatise on Analytical Chemistry, Theory and Practice,” Part I, Vol. I, I. M. Kolthoff, P. J. Elving, Eds., Chap. 14, Interscience, New York, 1959. (14).Sandell, E. B., “Colorimetric Determlnation of Traces of Metals, 3rd ed., Chap. 22, Interscience, Ken- York, 1959. RECEIVEDfor review l l n y 4, 1961. Accepted July 13, 1961. Southeastern Regional Meeting, ACS, Birmingham, Ala., Kovember 1960.
Chronopotentiometry of Iron(ll) and Iron(III) Adsorbed on Platinum Electrodes FRED C. ANSON California lnsfifufe of Technology, Pasadena, Calif.
b In sulfuric and perchloric acid sobtions Fe(ll) and Fe(lll) are adsorbed on the surface of platinum electrodes. The adsorption is strong enough so that the electrodes can b e washed free of any unadsorbed iron without removing the adsorbed iron. Chronopotentiograms recorded with such electrodes demonstrate the presence of the adsorbed iron. The chronopotentiograms obey the potential-time relation to be expected for adsorbed reactants. Adsorption isotherms for reactants adsorbed on platinum electrodes can b e determined with the aid of chronopotentiometry. The isotherms allow an estimate to be made of the effect of adsorption of reactants on the chronopotentiometric constant.
A
OF ~ N ~ T A N C E have S been reported by Lorenz and coworkers in which a substance reacting a t an electrode is adsorbed substantially (IW14). It has been suggested that such reactant adsorption may play an important role in the kinetics of electrode reactions (4, 6-9, 15, IO), so that the nature and properties of adsorbed reactants are of general interest. Previous studies of adsorbed reactants have usually been indirect because of the necessity of separating effects due to the adsorbed species from the (generally predominating) effects due to the unadsorbed reactant in the bulk of the solution (9, 12-14). Recently the ovidation and reduction of reactants adsorbed on platinum electrodes were studied directly by chronopotentiometry (5’). This report presents the results of further evperiments with Fe(I1) and Fe(II1) adsorbed on platinum dectrodeq. SUMBER
1498
ANALYTICAL CHEMISTRY
EXPERIMENTAL
Apparatus. T h e chronopotentiometric circuitry followed standard practice (IO). A Moseley autograph recorder (Model 3 3 ) was used to record the ehronopotentiograms. The potential difference between the working and reference electrodes was supplied to the input of a high-impedance follower amplifier having a gain of unity The amplifier was constructed from plug-in analog computer amplifiers following a circuit of DeFord ( 5 ) . The output of the amplifier was supplied to the input of the recorder. The working electrode was a piece of 0.030inch platinum wire sealed in soft glass. The area of the electrode was 0.15 sq. em. ‘The auxiliary electrode was a 100-sq em. cylindrical platinum gauze electrode The working electrode was positioned in the center of the cylindrical auxiliary electrode to ensure a uniform current density. Electrode geometries that provided less uniform current densities resulted in chronopotentiograms with distorted shapes. Potentials were measured with respect to a calomel electrode saturated with sodium chloride, but are reported us. the usual potassium chloride-saturated calomel electrode. It is necessary to pretreat electrodes prepared from commercially available high-purity platinum wire in order to obtain significant adsorption. The procedure consists of very light platinizing of the electrode by making i t the cathode in a 0.2M chloroplatinate solution and passing 0.1 to 1 coulomb per sq. em. of electricity through the electrode. The rewlting electrode surface should not be visibly darkened by the deposited platinum The potential of a properly prepared electrode can be forced from f0.6 to +0.1 volt us. S.C.E. in deaerated 1F sulfuric acid in less than 0.5 second with a constant current of 100 ,ua. per sq. cm. In
cases where the electrode response is not this sharp because of excessive platinization, the response can be improved by partially dissolving the platinized platinum in aqua regia. Procedure. T h e procedure for adsorption experiments consisted of immersing t h e electrode in solutions of Fe(I1) and/or Fe(II1) for measured time intervals, removing the electrode and rapidly washing i t thoroughly with distilled water, placing i t in an iron-free solution of deaerated perchloric or sulfuric acid, and recording chronopotentiograms corresponding to the adsorbed Fe(I1) and Fe(II1). Reagents. Solutions of Fe(I1) and Fe(II1) perchlorate were prepared and standardized as described previously ( I ) . Solutions were prepared from triply distilled water and measurements were performed in oxygen-free solutions. The temperature of all solutions remained within 2’ of 25’ C. RESULTS AND DISCUSSION
A platinum electrode that is exposed to solutions of Fe(I1) and Fe(II1) adsorbs both oxidation stafes of iron on its surface. The electrode may he removed from the iron solutions :ind washed free of unadsorbed iron without removing the adsorbed iron. The washed electrode may then he placed in an iron-free perchloric or sulfuric acid solution and chronopotentiograms corresponding to the absorbed iron obtained. That the ehronopotentiograins actually correspond to adsorbed iron is readily demonstrated. For one thing the transition times arc independent of whether or not the solution is stirred during the recording of chronopotentiograms. I n addition, if the direction of the chronopotentiometric current is reversed a t the t,ransi-
04-
-2
03-
’. $ 0
021
(0
,
w
010-
02-01
-
TIME
.
Figure 1 Alternate cathodic and anodic chronopotentiograms for adsorbed Fe(ll1) and Fe(ll) Electrode was exposed to a solution 0.2M in Fe(lll) and Fe(ll) for 50 sec. washed, and ploced in 1 F H&O, supporting electrolyte Current density, 65 pa. cm. 2
tion times, a series of cathodic and anodic chronopotentiograms results in which the transition time following each current reversal is approximately equal to the transition time preceding the reversal. Figure 1 shows a set of chronopotentiograms obtained in this way. The equality of successive anodic and cathodic transition times proves that the reactants undergoing the oxidation and reduction remain on the electrode. The magnitude of each successive pair of anodic and cathodic chronopotentiograms gradually decreases as the adsorbed reactants slowly desorb and diffuse into the solution. The time required for complete desorption of the adsorbed iron in I F perchloric or sulfuric acid solution decreases with the amount of iron adsorbed; but, with adsorbing solution concentrations of 0.05P or greater, only a negligible fraction of the iron is desorbed in the minute or two required to carry out a chronopotentiometric experiment. Even when the electrode is forced to potentials where extensive oxygen or hydrogen evolution occurs, the iron remains adsorbed on the electrode. COMPARISON OF DIFFUSION- AND ADSORPTION-CONTROLLED CHRONOPOTENTIOGRAMS
There are important morphological differences bet-iveen chronopotentiograms corresponding to adsorbed and diffusing reactants. These differences are demonstrated in Figure 2, which contains eiamples of both kinds of chronopotentiograms recorded with the same current density. Curve 1 resulted from the reduction of adsorbed Fe(II1); curve 2 from the reduction of Fe(II1) in a solution 0.39 m M in Fe(I1) and Fe(1II) in 1F HzSO4. The most striking difference in the tn-o chronopotentiograms is the shnrper potential inflection in curve 1. This results because once the adsorbed Fc(II1) is reduced to Fe(I1)
TIME
Figure 2.
Chronopotentiograms for reduction of Fe(l1l)
1 . Adsorbed Fe(lll) from a 0.2M solution of Fe(I1) and Fe(lll) 2. Diffusing Fe(1ll) in a 0.35mM solution of Fe(ll) and Fe(llll Supporting electrolyte, 1 F H ~ S O ~ Current density, 150 pa. cm.-l
ail of the current is available for the next electrode reaction, the evolution of hydrogen. The potential inflection is less sharp in curve 2 because a major fraction of the current continues to be conmmed in the reduction of Fe(III), even after the transition time, because more Fe(lI1) continuously diffuses to the electrode. Chronopotentiometric Wave Equations. Adsorption- and diffusion-controlled chronopotentiogranis should not obey the same potential-time equation. The equation for a reversible cathodic chronopotentiometric wave in which semi-infinite linear diffusion controls the supply of reducible species to the electrode is (10):
where Eli4is the potential of the electrode when t = 7/4, T is the transition time, and the other symbols have their customary significance. Equation 1is written for the case where no reduced species is initially present in the solution. The corresponding wave equation for a reversible reduction in which the reactant and product are adsorbed on the electrode is:
where El,2 is the potential of the electrode when t = 7/2. ( I t may be noted in passing that whereas El, depends on the square root of the ratio of the diffusion coefficients for the oxidized and reduced species, E112 is independent of diffusion coefficients.) A number of diffusion and adsorption chronopotentiograms for the reduction of Fe(II1) were analyzed according to Equations 1 and 2 by making plots ) In t / of E us. In t 1 / * / ( ~ 1 / 2 - t 1 / 2and ( ~ - t ) , respectively. Figure 3 shows
a pair of cathodic chronopotentiograms and the corresponding logarithmic plots on a n expanded potential scale. Curve 1 in Figure 3 is an adsorption chronopotentiogram that was obtained by immersing the electrode for 100 seconds in a solution 0 . M in both Fe(I1) and Fe(II1) in I F HBOd, washing the electrode, placing it in a deaerated 1F Ha04 solution, oxidizing all the Fe(I1) adsorbed on the electrode chronopotentiometrically, reversing the direction of the current at the anodic transition time, and recording curve 1. Curve 2 was obtained with the same electrode (free of any previously adsorbed iron) a t the same current density (42 pa. in a 0.2mM solution of Fe(II1) in I F H&304. Lines 3 and 4 in Figure 3 are plots of E us. log t/(.-t) and E vs. log t1’2/(~112- P2),respectively. The good linearity of line 3 and the fact that its slope is 0.061 volt (0.059 volt being the theoretical value) demonstrates that curve 1 obeys Equation 2 The poorer linearity and greater slope (0.074 volt) of line 4 indicate that the reduction of unadsorbed Fe(II1) proceeds less reversibly than that of aclsorbed Fe(II1). Analysis of chronopotentiometric waves of adsorbed substances by means of plots of E us. log t / ( ~ - t ) offers the possibility of establishing the nature of the adsorbed reactants in cases where they can exist in more than one state. For example, chronopotentiograms for the reversible reduction of iodine adsorbed on platinum electrodes will obey different equations depending upon whether the adsorbed iodine exists as molecules or as dissociated atoms. If the adsorbed iodine exists as iodine molecules the adsorption chronopotentiogram would obey the following equution :
VOL. 33,
NO. 1 1 ,
OCTOBER 1961
1499
TIME
0.E
I
I
I
I
I
A
-
I
~lOsec---l A
0.5
L
0
=-. 4
c! cn > YI
w 0.4 I R O N CONCENTRATION, M O L A R
Figure 4. Isotherm for adsorption of Fe(ll) and Fe(lll) on a platinum electrode at 25' C.
0.2 1.5
1.0
0.5 0 -0.5 LOGARITHMIC TERM
-1.0
Figure 3. Two cathodic chronopotentiograms and their corresponding logarithmic plots Chronopotentiograms for reduction of: 1. Adsorbed Fe(lll) 2. Diffusing Fe(ll1) t) for curve 1 3. Plot of E VI. log t / ( T f l / * J for curve 2 4. Plot of E. VI. log t1/*/(+2 Curve 4 was shifted b y 0.1 volt on potential scale for clarity
-
where E' is independent of t but depends on the current density. If the adsorbed iodine exists as iodine atonis, the adsorption chronopotentiogram would obey Equation 2. Studies of the iodine-iodide couple are in progress. Concentration Dependence of Adsorption. Measurement of transition times for the reduction and oxidation of adsorbed substances is a convenient means for determining the amount of adsorbed material. A series of measurements of transition times following exposure of the electrode to solutions of varying concentrations should enable one to construct ihe corresponding adsorption isotherm. Attempts t o carry out such a series of measurements, however, met with a number of difficulties. First, adsorption equilibrium is not rapidly established a t the electrode surface. Experiments showed that 15 to 20 minutes are required for the amount of iron adsorbed to attain its equilibrium value. In addition, Fe(I1) adsorbed on a platinum electrode surface is extremely susceptible to oxidation by atmospheric oxygen so that to establish the relative amounts of Fe(I1) and Fe(II1) adsorbed it would be necessary to carry out the transfer of the electrode from the iron solution to the iron-free solution in an oxygen-free atmosphere. The adsorption isotherms in Figure 4 were obtained with an electrode that was freshly oxidized and reduced, immersed in a series of Fe(1I)-Fe(I1I) solutions for 100 seconds each, and then
1500
ANALYTICAL CHEMISTRY
-
used to record first a cathodic [to convert all adsorbed iron to Fe(II)], and then an anodic chronopotentiogram. The anodic transition times were used to calculate the total quantity of adsorbed iron. Experiments showed that after immersion in the iron solutions for 100 seconds the amount of iron adsorbed on the electrode reached about 60 to 70% of the value that would have prevailed a t equilibrium. The absolute values obtained for total adsorbed iron in Figure 4 are thus lower than the equilibrium values. However, the general shape of the adsorption isotherm should correspond to that of the equilibrium isotherm because the rate of approach to adsorption equilibrium appeared to be approximately independent of the concentration of iron in the adsorbing solutions. Figure 4 shows that the adsorption isotherms for both perchloric and sulfuric acid solutions of Fe(I1) and Fe(II1) are approximately linear up to a concentration of about 0.3F. Also the amount of iron adsorbed is approximately the same from perchloric and sulfuric acid solutions. Mechanism of Electron Transfer at Electrode. A recent report (2) showed t h a t in IF perchloric acid solutions containing equal concentrations of Fe(I1) and Fe(III), the exchange-current density a t a platinum electrode deviates from the expected linear dependence on the concentration. Norms1 behavior was observed in a
0 1 F HClOa solution A 1F H2SOa solutions
1F sulfuric. acid solution of Fe(I1) and Fe(II1) ( 2 ) . The character of the deviation in perchloric acid suggested that specific adsorption of Fe(1I) and/or Fe(II1) on the platinum electrode could have been responsible for the anomaly. The present evidence demonstrating that Fe(I1) and Fe(II1) are adsorbed on platinum electrodes raises the question of the mechanism of electron transfer a t platinum electrodes in solutions of Fe(1I) and Fe(II1). The exchange current could flow entirely by means of the adsorbed Fe(I1) and Fe(II1):
+ e-
-
Fe(III),d. Fe(II),d.
Fe(II).d, Fe(III),d, f e +
(4)
(5)
In this case increases in the concentration of Fe(I1) and Fe(II1) in the body of the solution would increase the exchange current only because of the corresponding increase in the concentration of adsorbed Fe(I1) and Fe(II1). I n the previous investigation of the concentration dependence of the exchange current in Fe(I1) and Fe(II1) solutions in 1F HzS04 and 1F HCIOa (2), when the platinum electrode was rapidly transferred from a solution with an equilibrium exchange current density of 30 ma. to a more dilute solution with an equilibrium exchange current density of 5 ma. cm.-2, the lower exchange current density reached a steady equilibrium value within a minute or t-ivo. The present study has shown that the Fe(I1) and Fe(II1) adsorbed on the platinum electrode require a much longer time to be desorbed. Since the exchange current responds to changes in the concentration in the bulk of the solution much faster than does the conccntration of the adsorbed Fe(I1) and Fe(II1) Reactions 4 and 5 cannot constitute the major mechanism of electron transfer. The rapid influence on the exchange
current of changes in the bulk solution concentration could reshlt from the folloning pair of reactions occurring seriatim:
+
Fe(III)dd, e-%
Fe(II),d,
(6)
followed by Fe(II),d,
+ Fe(II1) 5 Fe(III),d, + WII) (71
with the opposite pair of reactions fctr the anodic exchange current. However, the mechanism according to Reactions 6 and 7 appears to be ruled out by the following argument: This mechanism would lead to a secondorder dependence of the exchange current on the bulk concentrations of Fe(1I) and Fe(II1) in the concentration range where the adsorption isotherm is linear because doubling the concentration of iron in the solution would also double the concentration of adsorbed iron, thus quadrupling the rate of Reaction 7. Yet a linear concentrationexchange current dependence is observed in sulfuric acid solutions of Fe(I1) and Fe(II1) throughout the concentration range where the adsorption isotherm appears to be linear ( 2 ) And in perchloric acid the deviation from linearity is away from, not in the direction of, a second-order dependence ( 2 ) . The mechanism of Reactions 6 and 7 is thus not consistent m-ith the ehTerimental results. The total exchange current could also consist of two components flowing in parallel at the electrode according to the following mechanism: Fe(II1) Fe(III),d,
+ e - -,Fe(I1)
(8)
+ e-
(9)
-
Fe(II),d,
with the opposite pair of reactions for the anodic exchange current. The mechanism embodied in Reactions 8 and 9 is analogous to the mechanism proposed by Laitinen and Randles (9) to account for a n anomalous exchange current behavior they observed in solutions of trisiethylenediamine)cobalt(II~ and -cobalt(III) at a mercury electrode. These authors showed that the expression for the eschange current in this case includes a n additional term involving the concentrations of the adsorbed reactants. For the case of solutions containing equal bulk concentrations C of Fe(I1) and Fe(III), the esrhange current density is given by = nF[k,C
+
kads
TZa(l
- z)fl-a)] (10)
where io is the exchange current density, k. and kada are the rate constants for Reactions 8 and 9, respectively, r is the surface concentration of iron (in both osidation states), z is the fraction of adsorbed iron in the Fe(II1) state, and a is the transfer coefficient for
el c l,ron exchange between adsorbed F;,II) and Fe(II1). Equation 10 leads to a linear de?Fndence of the exchange current on h e bulk concentration of Fe(I1) and Fe(II1) when the second term on the right is negligible compared with the first, or when the product r ( 2 ) . ( 1 - ~ ) ( 1 - ~is) a linear function of concentration. I n the concentration range where the adsorption isotherm is linear, r is a linear function of concentration and 5 and or may be relatively independent of concentration Thus the linear dependence of the eschange current on concentration observed with Fe(I1)Fe(II1) solutions in sulfuric acid is consistent with the parallel mechanism (Reactions 8 and 9) regardless of the relative magnitude of the two terms in Equation 10. The data in Figure 3 indicated that adsorbed iron is reduced more reversibly than unadsorbed iron. This means that k a d s in Equation 10 may be considerably larger than k.. However, the evidence cited above on the rate of response of the exchange current to changes in the bulk concentrations of Fe(I1) and Fe(II1) indicates that the contribution to the total exchange current from the adsorbed reactants is small compared to the contribution from the unadsorbed reactants. It appears that even though kadsis larger than k,, r is small enough compared to C so that the product k,c is larger than kad. r. Effect of Adsorption on the Chronopotentiometric Constant. The chronopotentiometric constant ~ T “ ~ /isC independent of i, T , or C so long as semiinfinite linear diffusion is the only means by which a reacting substance is supplied to the electrode, and the current results exclusively from faradaic processes occurring a t the electrode (10). If the substance that undergoes the electrode reaction is adsorbed on the electrode, the value of i ~ ~ ’ ~ increases ,lC as i increases or C decreases. Lorenz (11) and Reinmuth ( 1 7 ) have derived equations designed to allow the calculation of the effect of adsorption on the observed values of ~ T ” ~ / ’ C . illathematical intransigencies caused both authors to limit their derivations to special cases. These cases include: reaction of all adsorbed species before the species in the bulk of the solution; reaction of all of the diffusing species a t the surface of the electrode before the adsorbed species; a linear adsorption isotherm for which adsorption equilibrium is maintained a t the electrode surface during the chronopotentiogram. None of these cases corresponds t o the behavior of the Fe(I1)-Fe(II1) couple. Figure 2 shows that the reduction of adsorbed Fe(II1) takes place throughout the same potential range where Fe(II1) in the bulk of the solution would be reduced so that the first
and second cases do not apply. Fifteen to 20 minutes is necessary to attain adsorption equilibrium a t the electrode so that the third case is also inapplicable. Lorenz (11) considered the special case in which both adsorbed and diffusing substances react a t the electrode concomitantly. He derived a n equation giving the dependence of the chronopotentiometric constant on the transition time with the assumption that the total constant current is composed of two constant components corresponding to the reaction of the diffusing and adsorbed substances, respectively. This assumption will not correspond exactly to what really goes on a t the electrode with the Fe(I1)Fe(II1) couple, but i t is certainly more nearly correct than the other cases cited above. Lorenz’s equation can be written in the following form:
where (i~1/2/C),~, is the observed value of the chronopotentiometric constant, ( i T l i Z / C ) d l f f is the true chronopotentiometric constant that would be observed in the absence of adsorption, and r is the surface concentration of the adsorbed reactant. According to Equation 11 the true chronopotentiometric constant can be obtained from the intercept of the linear plot of ( i ~ 1 / 2 / C ) , ~us., ~ - 1 ’ 2 , while the surface concentration of the adsorbed substance determines the slope of the line. I n the case of a linear adsorption isotherm, plots of ir112/C V S . should be identical for all concentrations. The second term in Equation 11 becomes significant compared to the first term a t sufficiently short transition times, but the charging of the double layer a t the electrode often limits the experimentally realizable transition times to values too large for the contributions from adsorption to be observed. This is the case with the Fe(I1)Fe(II1) system. I n 1F HC104 the chronopotentiometric constant for the reduction of Fe(II1) is approximately 2.2 x 10-1 amp. cm.-2 sec.1/2 molar-’. The linear adsorption isotherm in Figure 2 has a slope of 3 X loF3 coulomb cm.-2 molar-‘ = nFr/C. Thus, one calculates that the second term in Equation 11 will be small compared t o the first term for transition times longer than about 10 ms. (milliseconds). Experimentally it is observed that the chronopotentiometric constant is independent of T for transition times of 5 ms. or greater with solutions from 0.004 to 0 . 4 N in Fe(II1) in 1F HC1O4. The accuracy of the measurement of transition times shorter than about 5 ms. is poor because of conVOL 33, NO. 11, OCTOBER 1961
* 1501
tributions from the charging of the double layer. Thus it is not possible to check Equation 11 with the Fe(I1)-Fe(lI1) couple because the adsorption is not sufficiently pronounced. Nevertheless, it is to be expected that for cases where adsorption is more extensive and both adsorbed and diffusing forms of the reactants are simultaneously reduced (or oxidized), Equation 11 will be the best approximation for calculating true chronopotentiometric constants. LITERATURE CITED
(1) Anson, Fred C., ANAL. CHEM. 33, 934 (1961). (2) Anson, Fred C., Zbid., 33, 939 (1961).
(3) Anson, Fred C., J . -4111. C‘hem. SOC. 83, 2387 (1961). (4) Brdicka, R., 2. Elektrochem 48, 248 (1942); Collectzon Czechoslw. Chem. Communs. 12, 522 (1947). (5) Deford, D. D., Div. of Analytical Chemistry, 133rd Meeting, ACS. San Francisco, 1958. (6) Delahay, P., Trachtenberg, I., J. Am. Chem. SOC.80, 2094 (1958). (7) Frumkin, A., Abstr. No. 172, Enlarged Abstracts of Papers of the Theoretical Section, Electrochemical Society Meeting, Philadel~hia.1959. (8) Eaitinen, H. A:, Mosier, B., J . A m . Chem. SOC.80, 2363 (1958). (9) . . Laitinen, H. A,, Randles. J. E. B..’ Trans. Faraday Soc. 51, 54 (1955). (10) Lingane, J. J., “Electroanalytical Chemistry,” Chap. XXII, Interscience, New York, 1958. (11) Lorenz, W., 2. Elektrochem. 59, 730 (1955).
(12) Lorenz, \Y., XIuhlbcre. El.. l b z d . . 59, 736 (195.5). (13) Lorenz, W., Xu1hlberg, H , Z.p h y s i k Chem. X.F. 17, 12:1 (1958) (14) Lorenz, W.,Schmalz, b. 0 , 2 . Elektrochem. 62, 301 (1958). (15) Matsuda, H., Delahay, P., Colleck~on Czechoslov C h e w Communp 12. 2977
(1960).
RECEIVEDfor review April 14, 1961 Accepted July 3, 1961. Division of Analytical Chemistry, 140th Meeting. ACS, Chicago, Ill., September 1961 Contribution KO.2665 from the Gates and Crellin Laboratories of Chemistry, California Institute of Technology, Pasadena, Calif. Work supported by the U. S. Army Research Office under Grant NO. DA-ORD-3 1-124-61-G91.
Potentiometric Titrations of Certain lnorga nic Substances by Titanium(ll1) Chloride and Chromium(I1) ChIo ride in N,N-Dimet hy Ifo rma mide JAMES F. HINTON and HAZEL M. TOMLINSON The Chemisfry Deparfmenf, Temple University, Philadelphia, Pa.
b Nonaqueous solvents are widely used in acid-base titrimetry and in titrations involving precipitations and complex ion formation, but few redox titrations involving inorganic substances in such solvents have been reported. This paper describes the procedures and results using Tic13 and CrClz in the titration of iodine, copper(ll), iron(lll), antimony(V), and bromine in N,Ndirnethylformarnide (DMF); CrClz also reduced titanium(1V) and iodine monochloride quantitatively. Other investigations in this field are currently in progress.
o
LITERATURE is available concerned with redox titrations in nonaqueous solutions except the studies in glacial acetic acid made by Stone (2, 3) and Tomicek (5-7) and their coworkers, and in liquid ammonia by Watt et al. (8-12). I n this laboratory, investigations not yet completed have included titrations of CuClz. 2H20, Iz, Br2, FeC13.6Hz0, and SbC& by TiCI, and CrCl? in DMF; in addition, IC1 and TiCL have been titrated with CrC12. Using Ti(II1) as titrant, potentiometric and visual titrations have a limit of error less than 10 parts per thousand. The limits of error in potentiometric titrations with CrClz are similar. Titanium chloride has been titrated by SbC15 in DRIF to a limit of error less
402
ANALYTICAL CHEMISTRY
than 1%. In each case in TiCl3 titrations, potentiometric and visual end points coincide and titrations may be made with nearly the same precision visually as potentiometrically. TITANIUM CHLORIDE AS REDUCTANT
Experimental. REAGENTS.Unless freshly opened, Fisher’s DMF (certified reagent grade, containing less than 100 p.p.m. of water) was stored for a t least 3 days ovei baked Linde Molecular Sieve 5A before use. All substances used for preparation of solutions were of Baker & hdamson reagent grade except the TiCla, with a typical analysis of 97% active reagent, which was kindly furnished by Union Carbide Co. Iodine and CuClz 2Hz0 were used as primary standards. APPARATUS.I n the titration setup, nitrogen of water-pump grade was passed successively through Burrell Oxsorbent, calcium chloride, magnesium perchlorate, past a lubricating oil manometer which served also as safety valve, into a closed glass cylinder encasing the upper parts of two burets with Teflon cocks, and hence through an exit needle between the burets. This needle was sufficiently flexible to effect stirring and providc a n atmosphere of nitrogen when placed in a 1-em. Beckman cell; this cell could be readily placed under either buret when a purely visual or spectrophotometric determination was made. Potentials were observed with either a Beckman Model G or a Leeds & LTorthrup Model 7664 pH meter.
A 20-ml. bulb and electrodes of 28gage B & 8 platinum wire were used. The buret-electrode ( I S ) assumes a definite potential in a given concentration of any one salt, but this value is unpredictable. A three-way stopcock permitted a continuous atmosphere of nitrogen whether or not the solution WE stirred by this gas. CHARACTERISTICS OF TITANIUM CHLORIDE SOLUTIONS.Restandardizations of solutions of TiCla in DhSF were required each day of use. Very di1ut.e (0.0005F) solutions, prepared by adding TiCL by buret to DhIF in a I-em. Beckman cell, showed a marked decrease in absorbance over a 5-minute period; the extent of change >%-as not affected by the quantity of nitrogen used for stirring. When Tic12 solutions were titrated by I2 solutions, considerably more Tic13 was consumed for a given amount of iodine solution than. in the reverse titration. This was to be expected since transfer of TiC1, solution from one environment to another resulted in decrease in titer, and the reaction with iodine mas very slow, permitting a long exposure to the effect of the new environment. When Tic13 solutions were titrated rapidly by CuC12,2Hz0 solutions, the amount of TiC13 required above that required for the reverse titration was negligible. This was expected since the reaction of Cu(IT) and Ti(II1) is very rapid. Solutions of Tic13 mere undfected by standing in a buret for more than 4 hours in close proximity to 1 2 solution in the other hrirct. TVatcr in the iodine
