tions, being always situated a t the same wave length. The spectra of the cupric salts in concentratcd halide solutions are far more complicatcd. In Figure 2 are given two broad absorption Rpectra in potassium and sodium chloride solutions. They penk a t 258 and 255 mp, respectively. [It appears that the spectrum shown in Figure 11 of (6) and indicated to be that of Cu(1) in 6 M hydrochloric acid, is really that of a Cu(I1) solution.] The Amsx of CuBr2 in KBr solutions was found by Fromherz et al. (3) to be at 293 mp, and in a later publication (9)a t 281 mp. The latter also observed one or two additional absorption bands in the longer wave length range, in highly concentrated lithium bromide solutions, and noted that the cupric salt decomposes to a large extent (about 19%) into the cuprous complex and bromine. These observations offer an explanation for the inconsistency in the stated peak positions, as well as for the exceptional broadness of the absorption spectra. It is suggested by the present authors that the broad bands obtained with the cupric halides result from the overlap of three distinct spectra-namely (with chloride solutions), the cupric complex at h,.”250 mp; the cuprous complex at A b . = 272 m l ; and the long wave length tail of chlorine, peaking in KCl solutions at 234 mp (corresponding to the
Vrband in crystalline KC1) (curve 111, Figure 2). The equilibrium: C U + ~(complex)+ Cu+ (complex) leans to the right in the alkali halide solutions studied according to the order: HCl < NaCl < KC1 < Kbr. Further proof for the above proposition has been obtained by trying to shift the equilibrium one way or the other, as follows : When chlorine gas is bubbled through halide solutions of Cu(II), the copper band is entirely hidden under the strong absorbance of the chlorine (curve IV, Figure 2). The excess of chlorine may be removed by careful boiling, whereupon the copper band, now peaking a t shorter wave lengths, reappears. The spectrum thus obtained (curve V) peaks a t 251 mp, i t is appreciably narrower than that of the untreated Cu(I1) solutions, and, in common with curve 111,does not show the rising tail of the second absorption band of Cu(1) at shorter wave lengths. (Solutions containing excess chlorine are sensitive to ultraviolet light, and their absorbance diminishes during a measurement.) When nitrogen gas is passed through a Cu(I1) solution, the absorption peak, as well as the minimum aksoktion at shorter wave lengths, are gradually shifted to longer wave lengths, clearly, owing to the equilibrium shift toward
the increased formation of Cu(1) (curve
VI). Last, when chlorine was bubbled through a Cu(1) solution and then boiled (to remove excess), absorption spectra identical in form and peak position to the Cu(I1) solutions were obtained. This is shown by dots close to curve I. Figure 2 also shows the minimum absorbance a t 220 mp of Cu(1I) solutions in comparison to a minimum appearing a t about 245 mp with Cu(1) solutions. LITERATURE CITED
(1) De Sesa. M. A.. Rogers. L. B.. Anal.
.-,(B) 3 , l (1929).
.
.
’
(4) Kanzelmeyer J., Freund, H., ANAL. CHEW 25, 18& (1953).
(6) Merritt, C., Jr:, Hershenson, H. M., Rogera, L. B., Zbzd., 25,572 (1953). (6) Yamamoto, Y., Bull. Inst. Chem. Research, Kyoto Unav. 36, No. 6, 139 (1958).
ABRARAM GLASNER PINCHAS AVINUR Department of Inorganic and Anal tical Chemistry The Sebrew Univeralty Jerusalem, Israel RECEIVEDfor review March 27, 1961. Acce ted A ril 11, 1961. Taken from the $h.D. tfesis to be submitted b P. Avinur to the Senate of The Hetrew University.
Adsorption as a Source of Inconstancy of the Chronopotentiometric Constant for Short Transition Times SIR: Recently, Bard (4) presented experimental data demonstrating that the chronopotentiometric constant, ir1I2/C, increases markedly at short transition times for a number of reactions at platinum electrodes. To account for this increase in the values of i9l2/C, Bard suggested that the short (10-l to lo-* second) transition t i e s were lengthened relative to longer (1 to 10 seconds) transition times as a result of oxidation of the electrode, or the time necessary to charge the double layer, or because the effective area of the electrode increased at short transition times because of the surface roughness of the electrode. This communication offers evidence that a major catme of the increases in if1‘2/Cobserved by Bard was adsorption of the species undergoing the electrode reaction on the surface of the electrode. Experimental values of i~l‘z/C for the oxidation of iodide ion in 1F Ha04 are presented by Bard for 10, 6, 1, and O.lmM solutions of iodide ion. I n the
extends into the solution from the case of iodide ion, oxidation of the surface of a chronopotentiometric workelectrode can be ruled out aa a conis approximately equal to ~ electrode / tributing factor to the increases in i ~ ~ / ing (Dt)1/2,where D is the diffusion coC because the transition times were efficient of the species that is reacting a t measured at a potential (0.7 volt us. the electrode and t is the time since the S.C.E.) where no appreciable oxidation electrolysis was started. The roughness of the platinum electrode could have of the electrode surface will begin to occurred. Qualitatively, t.he deviations from cause increases in the effective area of constancy in the chronopotentiometric the electrode when distance (Dt)’/* becomes commensurate with the average constant resulting from double layer distances from “hilltops” to “valleys” charging are very similar to deviations on the electrode surface. The possicaused by reactant adsorption (vide bility of contributions from roughness infra) and it is not always n simple causing increases in W/Z/C is more matter to distinguish between the two likely as T decreases; however, for a effects. However, Bard calculated that under the conditions he employed the constant value of T, i ~ 1 ” / C should be observed increases in i ~ l / ~ /were C much independent of concentration because (Dt)l I 2 is independent of concentration. too large to be even approximately Thus, chronopotentiograms obtaiied accounted for by contributions to T from the charging of the double layer. for solutions of varying concentration Electrode roughness can also be ruled with the current density adjusted in out as a major source of the increases in each case to keep T constant should give ~ T ~ / ~by/ Cmeans of the following values of i~l’q/C that are independent argument, which is due to Lorenz (6): of the concentration even for rough The distance that the diffusion layer electrodes. Figure 3 in Bard’s paper VOL. 33, NO. 8, JULY 1961 e
1123
( 4 ) shows clearly that the values of ~ T ’ / ~ for / C iodide oxidation evaluated by maintaining 7 constant still increase markedly as the concentration of iodide decreases. This increase cannot be attributed to roughness of the electrode surface. Lorenz and Mtihlbarg (6) have demonstrated that iodine, triiodide ion, and iodide ion (as well as silver ion and ferric ion) are adsorbed on platinized platinum electrodes. In fact, the exprrimental technique employed by these authors depends upon the fact that in the presence of adsorption the values of ir112/C increase as 7 decreases until, for sufficiently short transition times, the product i7, corresponding to the amount of material adsorbed, becomes constant. The common practice of evolving oxygen and then hydrogen on a platinum electrode prior to its use in an experiment results in the surface of the electrode becoming slightly platinized (1). I t is to be expected that many substances will be adsorbed on platinum electrodes treated in this way. Recent experiments in these laboratories have confirmed the presence of such adsorption (2) and found it to be especially pronounced in the case of iodide ion. A platinum electrode that had been oxidized and then reduced was immersed in solutions of iodide ion in 1F H a 0 4 for periods of 20 to 200 seconds. The electrode was removed from the solution, thoroughly washed, wiped dry with an adsorbent tissue, and placed in an air-free 1F HzSO~ solution that was being rapidly stirred, and an anodic chronopotentiogram was recorded with a cathode ray oscilloscope. The resulting chronopotentiogram clearly showed the presence of iodide ion on the surface of the electrode. The transition time was independent of whether or not the solution was stirred, but it decreased in
magnitude hhc longer the electrode n a s allowed to stand in the sulfuric acid solution beforo T was measured. I t is unlikely that the amount of adsorbed iodide determined in this way corresponds rxactly to the amount that is adsorbed when the electrode is a t equilibrium in an iodide solution. However, this measurement does determine a lower limit for the amount of adsorption. For example, immersion of the electrode in a 5mM solution of iodide in 1F lI2SOd for 100 seconds (no diffcrence was noted for immersion tinics between 50 and 200 seconds) resulted in a subsequent anodic transition time of 0.02 second a t 550 pa. per sq. cm. with the electrod? in air-free 1F H&04. Therefore, a t least 11 pcoulonibs per sq. cm. of iodide are adsorbed on the electrode in a 5mM iodide solution in 1F HZSOd. In chronopotentiometric experiments with a 5mM solution of iodide ion in 1F HzSOd, Bard observed that ir1l2/C increased from about 400 pa, set."* mM-l a t 7 = 1.0 second to about 500 pa, cm.+ sec.’lzmM-’at 7 = 10+ second. If, as a first approximation, it is assumed that the adsorbed iodide is only oxidized near the end of the transition time for the unatlsorbed iodide that diffuses to the electrode, an equation derived by Anson and Lingane (9) can be used to calculate the amount by which the transition time is increased due to the adsorbed iodide. If & is the amount of adsorbed iodide (11 microcoulombs cm.-2), i is the current density (25,000 pa. om.+ a t 7 = lo-* sec.), 7 is the transition time (-lo+ second), and T is the amount by which 7 is increased because of the adsorbed iodide,
Substitution of the known values of Q, i, and T in Equation 1 leads to a value of 2.3 X lo-* second for T. Subtracting this value of T from T and recalculating iT1l2/C gives 440, which is to be compared with 500, the value obtained with no correction for adsorption, and 400, the constant value obtained at longer transition times. The same calculation shows that for transition times longer than 1 second the contribution from adsorbed iodide is less than 5 X IO-s second, so that the effect of adsorption on T is negligible. Since the value used to evaluate the amount of adsorbed iodide gave only a lower limit for &, it is not to be expected that this correction procedure would completely eliminate the dependence of is112/C on 7 . Nevertheless this calculation shows that adsorption of the species fhat is reacting a t the electrode can be largely responsible for observed increases in the values of ~ T ~ /as~ 7/ decreases. C Unless data indicating the absence of any significant adsorption are available, this source of the lack of constancy frequently observed for values of i ~ ~ l cannot ~ / C be ignored. LITERATURE CITED (1) Anson, F. C., ANAL. CHEM.33, 934 ’( 1961). (2) Anson, F. C., J . Am. C h a . Soc., in
press. (3) Anson, F. C., Lingane, J. J., Ibid.,
79, 1015 (1957). ( 4 ) Bard, A . J., ANAL. CHEM. 33, 11 (1961). (5) Lorenz, W., 2. Eleklrochem. 59, 730 (1955). (6) Lorenz, W., Mulhberg, H., Ibid., 59 736 (1955); 2. physik. Chem. ( N . F.I 17, 129 (1958).
FRED C. ANSON Gates and Crellin Laboratories of Chemistry California Institute of Technology Pasadena, Calif. RECEIVED for review February 13, 1961. Accepted April 24, 1961. Contribution 2682.
Thermogravimetric Measurements SIR: In a recent paper, Newkirk (3) discussed an impressive number of limitations and sources of error that may be found in thermogravimetry. I should like to point out two other sources of error arising from a wrong use of thermogravimetry. A superficial interpretation of thermograms has led many authors to speak of a so-called “temperature of onset of decomposition.” The theoretical basis 1124
ANALYTICAL CHEMISTRY
behind these statements is insufficient. From the Arrhenius law: k P Ae-E/RT it may be deduced that the rate of any reaction, with a few exceptions, decreases with the temperature. But there is absolutely no reason why this rate suddenly becomes zero, although i t may be that no apparatus can detect it below a certain value of the tempera-
ture, characteristic of the apparatus, not of the reaction. Only a melting point may constitute such a temperature of onset of decomposition, because the kinetic mechanism which takes place in the liquid may be stopped in the solid phase. Of course equilibrium temperature under a given pressure is quite a different phenomenon. However, many authors have made careful determinations of the decom-,
