JOHN
[CONTRIBUTION FROM
THE
F. ENDICOTT A N D HENRYTAUBE
Vol. 86
DEPARTMEKT O F CHEMISTRY, S T A N F O R D UNIVERSITY, S T A N F O R D , CALIFORKIA]
Kinetics of Some Outer-Sphere Electron-Transfer Reactions BY JOHN F. ENDICOTT' A N D HENRYTACBE RECEIVED DECEMBER 3. 1963 Kinetic studies have been made of reactions of Ru(NHs)s*- with complex ions of the type (XH:l)5Co(III)L, for a variety of ligands L These reactions are necessarily of the outer-sphere type, and the relative specific rates of reaction are comparable to the relative rates of reduction of the same Co( 111) complexes by other agents which require the same type activated complex. h few limited comparisons between some of the results of this study and the Marcus theory of electron-transfer reactions have been possible.
Introduction
an electron pair to constitute "1as a bridging group) prior to electron transfer does not occur. The rates measured for the various reactions mentioned are considered in the light of Marcus' theory4 which suggests a means of correlating rates for reactions of the outer-sphere type.
Investigations of the rates of outer-sphere electrontransfer reactions in recent years have demonstrated a remarkable consistency in the relative rates of reaction of a number of reducing agents with a series of substituted cobalt(II1) Marcus has developed a treatment applicable to outer-sphere reExperimental actions which has shown promise of being a t least semiquantitative.4 - 8 General Procedures and Treatment of Data.-The reductants C r ( d i p ~ ) ~ ? +is important among the reducing used in this study were air sensitive, and all solutions were deaerated with nitrogen which had been scrubbed by passing it agents for which kinetic data have been obtained and through a series of solutions containing Cr*+(aq). All reactions which are presumed to react via an outer-sphere actiwere run in nitrogen atmospheres. vated complex. However, there is cause for serious In most of these studies, the reductant was added as the last doubt that it does, in fact, react by this type of mechreagent by nieans of a syringe t o the deoxygenated solution of oxidant, neutral salt, and acid. In some of the early studies, anism. The complex is very substitution-labile, so the reductant was added from a buret to a glass mixing device the possibility exists t h a t even when the activated attached t o a spectrophotometer cell. This mixing device was complex contains three dipyridyls for each C r 2 + a so designed t h a t oxidant and reductant could be kept separate coordination position is exposed by opening one of the until mixing was achieved by tilting. In some experiments the chelate rings. By contrast, the reducing reagent reductant was generated in one compartment of the mixing device, then mixed with oxidant. Solutions were thermostated in RU("~)~%+, which is featured in the work reported a water bath a t 25" before mixing. Kinetic determinations here, is found to undergo substitution very slowly were made by observing the change of optical density a t a given compared to the rate a t which i t can be oxidized; and wave length as a function of time. In some studies the rate of the product of one electron oxidation is R U ( ? V T H ~ ) ~ ~ +change . of optical density a t two or inore wave lengths was followed during a single run or in successive runs. In a few inThe reagent Ru(%H3)6'+ thus seems to be of the stances we have been able to follow both the disappearance of genuine "outer-sphere" type, and since i t is quite R U ( R - H ~ )a~t ~275 + tnp ( E -TOO)11 and the decay of the cobalt(II1) reactive to cobalt(II1) complexes, we have considered species. Reasonable agreement between determinations of i t worthwhile to study the kinetics of a number of such specific rates a t different wave lengths was always observed. reactions. The results provide a firm experimental In determining the rates of reduction of nu( ? X s ) 6 3 f the change in optical density a t 2 i 5 nip (emex for I M-1
CONSISTESCV OF MARCUS THEORY set.-'
i 3 X lo*" 2 x 102"
3.3 1
x
x
10-3"
sec
kll,
-1
1 x 10-2~ 5 2 X 10-5' 1 x 10-2b 38
Kl*d
jjz
9 x 107 3 . 2 x lo1@ 1 3 x 106' 1 . 4 x 10-2'
0 2 0.11
0 53 0 9
kg2
(calcd ) , .M-' sec
-!
3
30 6 1.3 X 2 6 X
a This study. * K. 1.. Krishnamurty and X . C. Wahl, J . A m . Chem. Sor., 80, 5921 (1958) .I.Anderson and N.A. Bonner, I'bzd.,76, Standard oxidation potentials from ref. 13 and as reported in this study. e .I.Zwickel and H . Taube, J . d n z . Chem 3826 (1954). Soc., 83, 793 (1961). Bjerrumh has determined the C O ( ~ " ~ ) ~ * + - C ~ ( " potential ~ ) ~ ~ + a t 30' in 1 N NHINOBas -0.055 v. arid in 2 N S H 4 N 0 3as -0.035 v . The value from ref. 13 is closer to what we would expect a t infinite dilution. Since this calculation is so sensitive to the values of E@and since the other values of Eo employed here are referred to infinite dilution, we feel -0.1 v . is a better estimate. If we had chosen -0.15 v., kna would have been 2.5 X and 3 X A - 1 sec. - 1 in the respective calculations. 0 k2? calculated in reaction 1 above. J Bjerrum. "Metal Ammine Formation in Aqueous Solutions. Theory of Reversible Step Reactions," 1'. Haase and Son,Copenhagen, 1941
govern the rates are still too poorly understood to justify confidence on this point. The difference in the pattern of reactivity for Cr2+(aq)compared to the other reducing agents may not be a matter of difference in mechanism per se, but the difference both in mechanism and in rates may reflect ways that the electronic structure of the different metal ions can affect their chemical behavior. Particularly for reactions of the inner-sphere type we may expect relative rates to reflect the individualities of the reaction partners. An interesting point which first was reported by VlcekJ is that the reductions of the diacetatotetraammine-cobalt( 111) complexes a t the dropping mercury electrode involve IH+], and, in fact, in the acid range covered in Vlcek's experiments, two hydrogen ions are called for in making up the activated complex for the reduction of cobalt(I11). The observations for ?h(xH3)62' and the dropping mercury electrode do differ for the reduction of the diacetato complexes; the former reagent exhibits a two-term rate law, but no term second order in [ H + ]appears. The reduction by Cr'+(aq) has a rate law similar to that exhibited by R U ( N & ) ~ ~ although +, the reaction with Cr'j+(aq) is of the inner-sphere type. The Results in Relation to Marcus' TheoryMarcus4has proposed a correlation of rates of oxidationreduction reactions proceeding by "outer-sphere" mechanisms with the rates of electron exchange for the component couples and the driving force for the reaction. Ti7hen the work terms are small4 as is likely in the case for reactions between ammine and aquo complexes, the correlation takes the form
where k l z is the observed specific rate of the oxidationreduction reaction, k , , and k2' are the isotopic exchange specific rates for the component couples, K12 is the equilibrium constant for the reaction, and
(2 is the collision frequency for the uncharged species which we, as Marcus did, will take to be 10" -1R-l sec. - I . )
Unfortunately, despite the large number of measurements which have been made, many of the parameters needed for a searching test of Marcus' theory a remissing. However, some tests of consistency can be applied. These consist of calculating some quantity such as a self-exchange specific rate from independent sets of data to learn whether reasonably constant (or, if only limits can be set, consistent) values are obtained. T h e notation to be used is this: subscript "1" will refer to the reducing agent which reacts and subscript "2" to the oxidizing agent. If kz2 in eq. 1 is regarded as the unknown, it will be noticed that the unknown appears also in the factor f 1 2 , fi2, however, is not very sensitive to the value of k Z 2 thus , the value of kz2 can be extracted from a set of known values of klz, K12,and k , by a process of iteration. Two such sets of calculations are reported in Table IV. For the two reductions of R u ( N H ~ ) ~ the ~ + ,calculated values of kz2 are consistent and are in reasonable agreement with the observation that the rate of the reaction R u ( N H ~ ) ~ C f ~R~U+ ( N H ~ ) ~is*4+ X lo2 -Ifp1 set.-', 2o since rates of reactions of ruthenium(II1) are not very sensitive to the ligands. For the reductions of C O ( N H ~ ) it ~ ~is+ necessary to compare the calculated values of kz2 with the upper limit of 8 X Jf-' sec. -I a t 65' which Stranks?' has set for the rate of electron exchange between C O ( N & ) ~ ~and + C O ( N & ) ~ ~ +At . 25', the upper limit can reasonably be set as 3.3 X 10-l2 Jf-l set.-'. The value of k22 estimated from the V2+(aq) reduction of Co(XH3)e3+ is much greater than the upper limit set by experiment; however, this may only reflect on the inaccuracy of the values of the oxidation potentials a t infinite dilution (see footnote f , Table I). The value for kz2 calculated from the Ru(NH3)6?+reduction of C O ( " ~ ) ~ ~ + ,however, is inconsistent both with the upper limit set by experiment and with the value of kz2 calculated from the V*+(aq)-Co("3) 6 3 + reaction. This discrepancy still exists even if k l l for the R U ( N H ~ ) ~ ~ + - R U ( ~ H ~ ) ~ ~ + isotopic exchange is taken to be l o 3 greater than the calculated value used in the determination of k22 from reaction 4 (Table IV). The AFO term arising in Marcus' theory is the standard free energy of the elementary electron-transfer step4 and the standard oxidation potentials employed (20) J F. Endicott and H T a u b e . J , A m , Chem. S O L .84, , 4981 (1982). ( 2 1 ) D . K . S t r a n k s , Discussions F a r a d a y Soc., 29, 7 3 (1960). N . S. Biradar and D. R . S t r a n k s , T u a n s . F a r a d a y S o r . , 58, 2421 (1963).
LITHIUMIN TRIETHYLSILANE-ETHYLAMINE REACTION
May 5 , 1964
in the above calculations may reflect the differences of the spin states of cobalt(I1) and cobalt(II1) as well as the free energy of the elementary electron-transfer step.?’ However,. no such variation of oxidation potential seems sufficient to account for the discrepancy observed between values of kz2 calculated for the V 2 + and the Ru(”3)6’+ reactions. Marcus’ theory is not invalidated by failing to account for the rates of reaction in a few particular cases. It does offer promise’4,25of correlating rates of electron transfer in a large number of “normal” cases, suggesting t h a t some special effects are operating in the reactions of C0(”3)6~+ with reducing agents; one will need to go beyond the theory t o discover what these special effects are. I n the reaction between Co(”3)e3+ and Ru(NH3),j2+ the over-all free energy of activation should reflect some compromise between the standard oxidation potential ( 2 2 ) This point has been discussed with regard t o t h e standard oxidation potential of Co(NHz)aOH22+-Co(NH3)sOH9 +.*a ( 2 3 ) A Haim and H . T a u b e , J . A m . Chrm. Soc.. 86, 499 (1963). (24) G. Dulz and K.Sutin, I n o r g . Chem., 1 , 917 (19F3) ( 2 5 ) R. J Campion, N . Purdie, and PI’. Sutin, ibid., in press.
[COSTRIBUTIOSFROM
THE
1691
of C O ( N H ~ ) ~ * + - C O ( N Hand ~ ) ~the ~ +standard oxidation potential (- 1.8 v.I3) of C O ( H Z ~ ) ~ ~ + - C O since (H~~)~~+ the ultimate cobalt(I1) product is Co’+(aq). Evidently the immediate cobalt(I1) species, once formed, decays rapidly into the equilibrium cobalt(I1) species. The other ruthenium(I1)-cobalt(II1) reactions reported in this paper seem to be more favorable thermodynamically. 26 Acknowledgments.-We wish to express our appreciation to Messrs. L. Bennett, R. Butler, D. Huchital, C. Andrade, R . Jordan, R. Linck, and J . Stritar, and Dr. E. S. Gould, who supplied a number of the complex compounds used in this study. The research described was supported by the National Science Foundation under Grant 6-20954. Funds for the purchase of the Cary Model 14 spectrophotometer were supplied by Grant G-22611 from the National Science Foundation. (26) For example, t h e standard oxidation potential of
+
C O ( N H ~ ) ~ O= H~ C *O~( S H ? ’ ~ ~ O He~- ~ + has been estimated t o be -0.33 v.27 (27) R. G . Yalman, Znorg. Chem., 1, 16 (1962).
GEORGEHERBERT JOKES LABORATORY, THEUSIVERSITY OF CHICAGO, CHICAGO, ILLINOIS]
The Role of Dissolved Lithium in the Reaction between Triethylsilane and Ethylamine BY ARLENVISTE AND HENRYTAUBE RECEIVED DECEMBER 18, 1963 The reaction of triethylsilane with ethylamine is not significantly catalyzed by dissolved lithium itself. Instead the reaction appears t o be catalyzed by small amounts of a strong base such as lithium ethylamide, presumably present as an impurity in the lithium-ethylamine solution.
Introduction In 1934 Kraus and Nelson’ reported that triethylsilane reacts with ethylamine in the following way. ( CaHj)&H
Our main interest was in the possibility of genuine catalysis by lithium; if this were to take place, the reaction would offer a convenient means of studying the kinetics of a reaction involving a metal in solution.
Li + C 2 H b S H 2+ ( C z H s ) a S i h ” C ~ H+ s HZ
Experimental The stoichiometry with respect to Et3SiH and Ht Reagents.-Anhydrous ethylamine (Eastman No. 506) was was established, and the product EtaSiNHEt was isodried with metallic lithium and then distilled on the vacuum line lated and identified. By adding the lithium to a soluinto a storage vessel where it was kept over lithium wire until use. After each use, the ethylamine was distilled back t o this tion of Et&H in E t N H 2 , they found t h a t the reaction storage vessel. For most experiments, triethylsilane (Peninsular begins as soon as the blue color due to lithium develops, ChemResearch N o . 436) was sealed off in fragile Pyrex ampoules and (qualitatively) the rate of reaction increases as the under vacuum. Triethylfluorosilane (Peninsular ChemResearch Li concentration is increased. It was shown t h a t an S o . 2747) was redistilled. When used, it was outgassed on the initial mole ratio of Li to Et3SiH of either 1 or 0.1 vacuum line and condensed into the reaction flask. Small pieces of lithium metal (Lithium Corporation of America) were sealed results in catalysis. Accordingly, they concluded off in fragile Pyrex ampoules under vacuum. that “the reaction is a homogeneous one and the Apparatus and Procedure.-About 95% of the 420-ml. volume lithium evidently serves merely as a catalyst.” available t o the reaction system consisted of a glass vessel which Later, in a paper devoted to preparative work could be submerged in a thermostated bath in a dewar. T h e utilizing this and similar resctions, Dolgov, et d.,* reaction vessel was connected t o a manometer for pressure measurements. The solution and the gas above it were stirred with a offered the view that the reaction is instead basemultipole magnetic stirrer enclosed in an evacuated glass jacket catalyzed, the base being the alkali amide formed by and driven by a variable speed motor. Two evacuated glass the reaction of the alkali metal with the solvent. tubes served as ampoule breakers. Each was sealed to a reagent ampoule (Li or Et3SiH)and contained a small bar magnet allowPresumably the possibility of base catalysis was suging it t o be manipulated by a horseshoe magnet outside the reacgested in part by kinetic studies of the reaction between tion vessel. The reaction vessel was thermostated in a stripa trialkylsilane and alcohol or water, beginning with silvered dewar. The bath liquid (usually methanol) was periodithat of Price. cally circulated through a cooling coil in a Dry Iceemethanol (1) C . A. K r a u s and W . K. S e l s o n , J . A m . Chem. Soc., 66, 195 (1934). (2) B. N Dolgov, N . P. Kharitonov, and M . G. Voronkov, Z h . Obshch. K h i m . , a i , 678 (1954). (3) F. P. Price, J . A m . Chem. Soc., 69, 2600 (1847).
slush in a separate dewar by means of a self-priming pump, which was controlled by a thermistor temperature controller (Yellow Springs Instrument Co., Model 63RB). For most of the experiments, temperature stability was about 1 0 . 1 ” . Total drift in
