3366
N. KELSOKINGAND M. E. WINFIELD
Vol. 83
TABLE I11 THROUGH ELECTRON TRANSFER IN THE SYSTEM [Co(PDTA)]--[Co(PDTA)]' RESOLUTION T = 100'; pH 2.0 am
% Resolution" k ( M - 1 hr. - 9 6 Calcd. Racemic, 0.02 M D-1,0.04 h!f ..... -0.202 84 1.9 Racemic, 0.02 M D-1, 0.02 h f -0.138 - .138 75 2.2 D-1, 0.02 M Racemic, 0.02 M .I33 .138 75 2.2 D-1, 0.02 hf Racemic, 0.03 M - .119 .116 70 2.1 0 Per cent. resolution is the ratio of the calculated concentration of the more abundant isomer in the equilibrium mixture to the concentration of the initial racemate. Each value of k is the average of more than two experiments. ([Co(PDTA)I-), a
([Co(PDTA)I)-, b
Expt.
-
-
*
tribution might vary anywhere from about 51% (~-L)-49% ( ~ 4to) something like 99+% ( ~ 4 ) 1% (L-l), since the kinetic equations will be obeyed so long as the rotation is instantly proportional to the concentration of the cobalt(I1) complex. From the nature of the steric interaction of the methyl group with the remainder of the ligand, an increase in the size of the metal ion should increase this repulsion, for this would fold back the methylene hydrogen atoms into the methyl group. One therefore expects the methyl group to be a t least as selective for ions bigger than cobalt(111) as it is for that ion. Since the results given here prove the cobalt(II1) complex to exist a t equilibrium as 99+% D-I (or L-d) isomer and since cobalt(I1) is larger than cobalt(III), it is only reasonable to conclude that the cobalt(I1) complex is essentially all (99+%) of the single preferred configuration. From the foregoing, a novel method of resolution may be suggested for certain very special systems. I n systems retaining configuration, these configurations (d and 2) are distributed statistically between the exchanging species as a result of electron transfer (just as an isotopic tracer is).
I n consequence, an excess of an optically active ion will confer a high degree of resolution on the second ion (originally racemic) a t equilibrium. This has been demonstrated (Table 111) by following such solutions to their equilibrium rotations, which may be calculated from equation 8 where a and b are as previously defined and [a111 and [a1111are the rotations of one molar solutions of the divalent and trivalent complexes. In all cases the experimental and predicted values for the equilibrium rotation agree closely. Other novel resolution methods based on equilibria have been demonstrated among c ~ m p l e x e s . ~ ~ ~ ~ ~ Acknowledgments.-The support of the National Science Foundation is gratefully acknowledged. The authors also wish to express gratitude to their colleagues for helpful discussions and council. Thanks are especially due Drs. W. B. Hadley, J. G. Calvert and W. J. Stratton for their interest in this work. (12) F. P. Dwyer, M. P. O'Dwyer and E. C. Gyarfas, Nature, 167, 036 (1951); F. P. Dwyer and E. C. Gyarfas, ibid., 168, 29 (1951). (13) D. H.Busch, J. A m . Chcm. Sot., 77, 2717 (1965).
[CONTRIBUTION FROM THE DIVISIONOF PHYSICAL CHEMISTRY OF THE CHEMICAL RESEARCH LABORATORIES, AUSTRALIAN COMMONWEALTH SCIENTIFIC AND INDUSTRIAL RESEARCH ORGANIZATION, MELBOURNE, AUSTRALIA ]
The Liquid Phase Hydrogenation of Cyanocobaltate(11) BY N. KELSOKINGAND M. E. WINFIELD RECEIVED DECEMBER 27, 1960 Solutions containing [COII(CN)~]Jreact reversibly with molecular hydrogen to give a colorless hydride ion of probable structure [CO~I~(CN)E.H]~-, which has considerable stability in alkaline water or methanol, or in the solid state. The same product results from reduction of [CoII(CN)s]J- electrolytically or by means of NaBH4. It is formed also in the first step in the aging of [CoII(CN)s]s-, which proves to be disproportionation to hydride and [ C O ~ ~ ~ ( C N ) ~ rather O H ] ~than a simple dimerization as formerly believed, Since cations accelerate both disproportionation and H2 uptake, the influence of added salt on the latter reaction is not readily analyzed but is probably a combination of cation promotion and a general salt-effect diminished by loss of reactant by disproportionation. Cs+ provides a clear example of excellent promotion of gas uptake vitiated by excessive stimulation of disproportionation. K + is the best compromise promoter in the series of alkali metal ions. Use of methanol in place of water as solvent increases the rate of H2 uptake 4-fold (taking into account a 10-fold increase in gas solubility), Descriptions are given of the sequence of reactions which follow mixing of Coclz and KCN solutions and of the spectra of essentially pure solutions of [COII(CN)~!- and [CoIII(CS).,H]'-. The hydride has a single absorption band in the visible and ultraviolet regions, situated a t 300 mp. The HZ evolution and exchange reactions, which have been observed in mixtures of CoClt and KCN by previous investigators, can b e expl'rined in terms of [CoIII(CN)sH]*- as the source of the liberated gas.
Introduction Cobaltous chloride added to excess aqueous KCN yields a greenish solution containing the pentacyanocobaltate(I1) which can be re(1) M. A. Descamps, Com$l. rend., 67, 330 (1868). (2) D. N. Hume and I. M. Kolthoff, 1. Am. Chrm. Soc., 11, 867 (1949). (3) A. W.Adamson, ibid., 1 3 , 6710 (1951).
duced by various means to an ion whose color and absorption spectrunl have not previously been determined correctly because of masking by other compounds present. Treadwell and H ~ b e r a, s~ well as later investigators, have regarded the ion as a complex of Co', the expected product of a 1electron reduction of [COII(CN)~]~-. (4) W. D. Treadwell and D. Huber, Hclo. Chim. A d o , 26, 10 (1943).
Aug. 20, 1961
LTQUID PHASE HYDROGENATION OF CYANOCOBALTATE(II)
3367
TABLE I Complex
VISIBLE AND ULTRAVIOLET SPECTRA OF CYANO-COMPLEXESOF CO" AND CO'" AT 25' Wave length (A mp) and molar extinction coe5cient (e) Solvent
7
280 5300 279.6 4800 296 >2600
316 428 618 967 Water X 231 [CO"( CN)s]'530 49 7 298 (263) E 6800 -1000 317 430" 620" 905" Dry methanol X 232 (263) [Co"(CN)r]Se 5800 -1000 580 49 N7 290 Dimethyl sulfoxide X 356 396 1110 [Co"(CN)s]'E >330 >300 >200 [CoIII(CN)a]*Water X 305 e 610 [COI~~(CN)SOH]*-Water X 235 380 e -7000 360 The solubility in methanol is insufficientt o permit accurate determination of X and e in the visible region of the spectrum.
T h a t molecular hydrogen could be used for the reduction appears to have been first noticed by I g ~ c h i . The ~ reaction has since been studied by us6-@and by several other groups10-12 with the object of determining the nature of the active molecular species and how it combines with HP. The rapidity of gas uptake, even a t Oo, suggested that the structure was uncommonly well adapted t o the "activation" of Hz; knowledge of the molecular structure of reactant and product might therefore lead to a better understanding of hydrogenation ~ a t a l y s i s . ~ ~ ~ ~ ' ~ J ~ ~ ~ ~ I n the present paper we wish to describe experimental work mentioned in previous communicationssJV8 and whose publication has been delayed by difficulties of interpretation. Recently we have succeeded in obtaining a clean spectrum of the hydrogenated product, which permits a re-assessment of our earlier results and those of Mills, Weller and Wheeler." It provides also the key to understanding the exchange reaction described by Mills, et al., and the hydrogen evolution reaction discussed by King and Winfie1d.O Meanwhile the demonstration by nuclear magnetic resonance measurementsl2 that the reduced complex does indeed contain a hydrogen atom has resolved the major doubt regarding its structure.
Results Aging in vacuo.-By working in dilute aqueous solution, taking considerable care to exclude reactive gases, i t is possible to mix CoClZand KCN in a Thunberg tube to obtain a solution whose absorption spectrum is essentially that of the pentacyanocobaltate(I1) ion (Table I and Fig. 1). There is negligible contribution from ions other than [co11(cN)613-and CN- as judged by the constancy (5) M . Iguchi, J . Chcm. Soc. (Japan), 6S, 534 (1942). (6) N. K . King and M E. Winfield, Biochcm. Biophys. Acta, 18, 431 (1955). (7) M. E. Winfield, Reus. Pure A p P f . Chcm. (Auslraiia), 6, 217 (1955). ( 8 ) J. Bayston, N. IC. King and &I. E. Winfield. "Advances in Catalysis," Vol. IX, Academic Press, Inc., New York, N. Y.,1957, p. 312. (9) N. K. King and M. E. Winfield, J . A m . Chcm. SOC.,80, 2080 (1958). (10) S. W. Wcller and G. A. Mills, "Advances in Catalysis," Vol. VIII. Academic Press, Inc., New York, N. Y.,1956, p 183. (11) G. A. Mills, S. Weller and A. Wheeler, J . Phys. Chcm., 63,403 (lg59). (12) W. P.Gri5th and G. Wilkinson, J . Chcm. Soc., 2767 (1959). (13) J. Halpern, Quart. Reus. Chcm. Soc , 10, 483 (1956). (14) J. Halpern, "Advancer in Catalysis," Vol XI, Academic Press, I n c . , New York, N. Y.,1959,p. 301.
of extinction coefficients when the cobalt concentration is varied through the range 0.0003 to 0.0072 M. 200
250
333
500
I
I
I
40000
30000
20000
IO00
V (cm-I),
Fig. 1.-Absorption spectra of [Co1I(CN)~Ia-and oi [Co11r(CN)6H]*- in water a t 25'. The hydride solutio*i was obtained by reducing [ C O I ' ( C N ) ~ ] with ~ - NaBHd utitil negligible amount of the CoI1 complex remained.
Under the conditions mentioned above and at 25' the [COII(CN)6]3- ion is remarkably stable. The aging process" is so slow that a t a cobalt concentration of 0.0024 M and with a cyanide to cobalt ratio ( K ) of 10, the complex declines in concentration only 1% in two days. We may therefore assume that all of the change that occurs in a few hours when Hz is admitted is due to reaction with HP, with a negligible loss of [ C O ~ I ( C N ) ~by ] ~ -aging. We have already mentioned0 that the 967 nip band can be used as an accurate measure of the [ChII(CN)~]s-concentration. It lies remote from the bands of other complexes likely to be present. Of the remaining [CoII(CN)~JBbands only the one a t 280 mp is useful quantitatively, and then only
N. KELSOKINGAND M. E. ~VINFIELD
3368
50
Vol. 83
-
-1
40 -
R = 40 -'
-!
%
0
CONTROL
w 30-
A
PLUS
u
;r
0
3
"
Ll CI
NaCI
1-
20 -
A
PLUS LI Cl
/ 10 -
I
0
5 TIME
10 ELAPSED
15
5 TIME
20
(MIU)
I
I
10 ELAPSED
5
I
20
(M!lu)
Fig. 3.-Heterogeneous reaction with HZ showing the effect of alkali metal ions. R = 4.0; other conditions as in Fig. 2 .
Fig. Z.--Ilomogeneous HZuptake by mixtures of aqueous CoC12 and KCK in the presence of alkali metal ions. X (mole ratio CN/Co) = 8.0; 20 micromoles Co in 2.8 ml. total vol. ; 1 atm. Hz; 1'; 90 micromoles of added salt.
the equation
in exceptional circumstances The 316 mp peak is particularly susceptible to the presence of impurities, a trace of 0 2 for example being sufficient to obliterate the peak leaving an indeterminate shoulder. Hydrogenation with H2.--When [COII(CN)~] 3is prepared in the presence of 1 atm. of Hz, and in the above-mentioned concentration range 0.0003 to 0.0072 M , the 967 mp peak falls rapidly a t first, then gradually for several days, the change being accompanied by formation of a colorless complex which absorbs a t 305 mp (Table I and Fig. I). There is no other product detectable spectrophotometrically. Since rise of the 305 mp peak is accompanied by decline of the nearby 316 mp band of [COI~(CN)E,]~and since traces of O2 can result in the appearance of two complexes which absorb a t 310 and 311 mp, some care is required in carrying out the experinlent. Analysis of the spectra obtained after hydrogenation appears to have reached equilibrium indicates ~ - finally that almost all of the [ C O I I ( C X ) ~ ] is converted to the 305 mp complex, which we shall L e . , a cobaltic complex write as [COIII(CN)~H]~-, in which one of the ligands is the hydride ion H-. The structure is discussed later. Manometric Determination of Hz Uptake.Iguchi6 obtained an uptake of 0.8 hydrogen atom per atom of cobalt, and Mills, Weller and \{-heeler" have found values near 0.95 under favorable circumstances. Thus the experimentally determined stoichiometry is close to that expected from
2[Co'I(CN)sI3- f Hz F! ~[CO"'(CN)E.H]~if the reaction were to go to completion. IYith dilute solutions in the presence of 1 atm. of Hz a ~ ] ~ - a t equilibrium. few % of [ C O ~ I ( C N ) remain Since the rate of aging, and its extent, are temperature dependent, the experimentally observed activation energy and frequency factor for Ht uptake are of uncertain significance. Values obtained a t R = 4 over the range 1-10' are 6 kcal./ mole, and 5 >( l o 3 l./hr.//mole of cobalt, respectively. The order of effectiveness of alkali-metal cations in accelerating Hz uptake is Cs+ > K + > Na+ > L i t as shown in Fig. 2. After a brief period the rate in the presence of Csf declines due to considerable loss of cobaltocyanide ion by disproportionation (discussed later). A sustained high rate of reaction with Hz can be achieved by using KCl as promoter. It was demonstrated later that R b + lies between Cs+ and K + in promoting HZuptake and also in producing the absorption peak a t 380390 mp which in our previous paper9 we have related to the H2evolution reaction. Qualitatively the salt effect is the same in the presence of precipitated cobaltous dicyanide, a t am R value of 4 (Fig. 3). Here however the apparent rate in the presence of Csf tapers off sharply due to Hzevolution, which after 10 min. exceeds H? uptake. Reversibility of Hydrogenation.-If after reaction with HPthe residual gas is pumped off, the 967 mp hand can be observed to increase (after allowing
Aug. 20, 1961
LIQUID PHASE HYDROGENATION OF CYANOCOBALTATE(II)
3369
for loss of water during pumping). Complete re- phosphate or Nujol mulls. In the ultraviolet region the absorption of KeCo2(CN)lois not clearly covery ._- of the original cobaltous complex is not distinguishable from that of the products of dispossible. A solution of [CoII(CN)61a- allowed to stand for proportionation and oxidation a t the surface of the crystal. several days in Hz a t 0" will, when warmed to 25', Methanol as Solvent.-On mixing CoCla and evolve €32 with simultaneous increase in absorption KCN solutions in methanol which contains a little a t 967 mp. An alternative demonstration can be made by water (A.R. methanol without further drying), the adding ethanol. l2 The experiment is convincing absorption spectrum resembles that obtained in if the cobaltous complex is treated with NaBH4 water, with the following exceptions: rather than Hz, thus achieving virtually complete (i) The 230, 316 and 967 mp bands are disconversion to [COIII(CN)~H]~--. On addition of placed 2 mp to higher wave lengths. (ii) There is appreciable disproportionation, ethanol there is a copious evolution of H2 accompanied by precipitation of considerable K ~ C O Z -as in water containing CsC1. (iii) The reduced product of the disproportiona(CN)IO. Reversal can also be observed on turning off the tion is less soluble than the 380 mp complex and current during electrolytic reduction of [CoII- tends to precipitate out. I n the absence of oxygen i t can be separated and dissolved in water with no (CN)6I3-. change in wave length of the absorption band (305 Reduction by Sodium Borohydride.-When NaBH4 is used in place of Hz the reaction is more w). rapid and complete, again yielding the 305 mp (iv) A weak band a t 510-530 mp is readily complex (Fig. 1) with no other products readily detectable for several minutes after mixing, a t 2". At room temperature i t is far less apparent. The detectable in the spectrophotometer. Electrolytic Reduction.-The result is similar to band is also seen when pure &Coz(CN)lo is disthat obtained with Ha, although it is much more solved in methanol a t low temperatures. It is a difficult to achieve a clean spectrum. Thus the reasonable assumption that the ion responsible CoI complex claimed by Treadwell and Huber4 to for the absorption is [Co2(CN)lo16-. be the reduction product is our 305 nib complex, Although [ C O ~ I ( C N ) ~is] ~much less soluble in which we believe to be [ C O I I ~ ( C N ) ~ H ] ~ - - . A.R. methanol than i t is in water, the solubility is "Self-Reduction."-Formation of the 305 m,u adequate for determination of absorption spectra complex in solutions containing [COII(CN)~]~-,and rates of Hz uptake and evolution. When the without the addition of reducing agent, may be reagents and solvent are dried further, the soluobserved in certain circumstances, for example bility becomes quite small (about 0.00025 mole/l.). when CsCl is added. It is then accompanied by I t is of interest that in the driest solutions which we a rise a t 375-390 mp, shown later to be due to a are able to achieve there is negligible shift in wave complex whose absorption peak is a t 380 mp and length of the absorption bands compared with those whose structure is [ C O ~ I I ( C N ) ~ O Hor ] ~ - [CoIII- seen in pure water (Table I). There is little dispro(CN)50HzI2-. portionation (about 6% per day in 0.00024 M It can be demonstrated spectrophotometrically solution a t room temperature), and its slow progthat two moles of the 967 mp complex yield one of ress seems to be accompanied by decreasing soluthe 305 and one of the 380 mp compound, assuming bility of [COII(CN)~]~-, suggesting that the disall three to be monomeric. Thus the simple dis- proportionation is removing the last traces of water proportionation mechanism for the loss of para- from the solvent. Since no new absorption bands magnetismll on aging proves to be correct. The are seen, i t is unlikely that both of the ions [CoIIview previously held by hlills, Weller and (CN)6I3- and [CoII(CN)60H2l3- can exist in soluWheeler" and by us that dimerization was responsi- tion. ble was based on inability to detect the products of Hydrogenation in Methanol.-Uptake of Hz in disproportionation We were unable to distinguish A.R. methanol is accompanied by increased absorp] ~ - the 316 tion a t 305 mp, as in water, so that the product is the 305 mp band of [ C O ~ I I ( C N ) ~ Hfrom mp band of [ C O I I ( C N ) ~ ]and ~ - the 310 nip band of again assumed to be [COIII(CN)~H]~-. failed to ob[COIII(CN)~]~-, while Mills, et d., Manometric measurements show that the rate of serve the band near 380 mp. The latter is clearly reaction is much greater than in water: for example, visible if the region is scanned continuously during the experiment. Mills, P t al., may have used more 46 times greater a t 21" in 1 atm. of H?, with R = 16 dilute solutions for their spectrophotometry than and a cobalt concentration of 0.005 M . Allowing for their paramagnetic studies, or the 380 mp com- for the 10-fold greater solubility of HOin methanol, plex may have already been converted to [CoIII- we may conclude that the concentration of the reactive cobalt species is 4 times higher in methanol (CN)sl3- before the spectrum was examined or that the active species is 4 times more reactive Absorption Spectrum of K&oz(CN)lo.-The purple crystalline diamagnetic3 form of potassium due to an enhanced salt effect. It is clear from the pentacyanocobaltate(I1) has a visible absorption remarks in the previous section that the proportion band a t 532 m p when a crystal is examined in air.16 of cobalt present as [ C O I I ( C N ) ~ ]is~ -less in methWith the wave length shifted a little toward the anol. The initial rates of gas uptake a t a given value red the peak is also readily detected in tricresylR are proportional to approximately the second of (15) Measurement carried out by Dr J. Ferguson, of the Division of power of the total cobalt concentration, as in Chemical Physics, Australian Commonwealth Scientific and Industrial water.8 Research Organization, using a microspectrophotometer.
N. KELSOKINGAND M. E. WINFIELD
3370
Although the results in methanol are suggestive that [COP(CN)~O]~is the ion which reacts with HP, experiments in which the height of the 530 mp band was determined as a function of time showed no correlation between [Coz(CN)1ol6concentration and rate of HS uptake. No ions which might conceivably react with Hzwere consistently found in greater concentration in methanol than in water. Dimethyl Sulfoxide as Solvent.-The pentacyanocobaltate(I1) ion has a distinctly different absorption spectrum in dimethyl sulfoxide, indicating reaction with the solvent. At first the principal product is a 1110 mp complex which has an intense band a t 296 mp (Table I). It ages rapidly, reacts weakly with Hz and strongly with OP. Aging Reactions in Water in the Absence of Reactive Gas.-At high cobalt concentrations (>0.1 M ) the aging is rapid16 and results in loss of most of the [ C O I I ( C N ) ~ ]as ~ -indicated by decline of the 967 mp band. Visual and spectrophotometric observations indicate that the sequence of events initiated when strong solutions of CoCI, and KCN are mixed proceeds as
+ 2CN-
[Co"(OH&,]'+
--f
+
CO(CN)~ 6Ht0
(1)
At the surface of the precipitated dicyanidel? the Co(CN)P appears to be converted for a brief period to K&!O~II(CN)~O,thus coloring the particles redg 2Co(CN)2
+ GKCN +K~CO~"(CN)IO (2)
The K&!oz"(CN)lo then dissolves in the surrounding water to give the dissociated dimeric ion, which rapidly monomerizes &Coi"(CN)io
a [Co2"(CN)iola-
[Cor''(CN)io]'-
+ 6K'
?f 2[Co"(CN)5]'-
(3) (4)
Particularly in strong solutions it is likely that disproportionation occurs partly by the direct route via the dimer
+
[COP(CN)IO]'- Hz0 --f [CO"'( CN)sH] 8 -
+ [ CO'"( CN)rOH]'- (5)
and the remainder vk the monomer
+
+
[COI'(CX)~]'- H-OH [CO"(CN)~]*- ---t [CO'~'(CN)sH] 1- f [CO"'( CN)sOH] a-
+ CN-
---f
[Co1I'(CN)~]'-
+ OH-
catalyst ~ [ C O ~ ~ I ( C N ) ~ H ] ~ - ~ [ C O * ~ ( C X ) ~ H] ~z - (8)
7
(7)
It is unexpectedly slow, suggesting that appreciable activation is required for addition of the sixth cyanide group. A tendency for the reduced product of disproportionation to generate an equilibrium concentration of HS becomes evident if a catalyst is present or the solution heated@ (16) This had been shown earlier by Mills. Weller and Wheeler*' by following the decline in paramagnetism. (17) In the early literature i t is regarded as a hydrated salt which m a y exist in either of the two forma Co(CN)v2H10 or Co(CN)a*3H,O. Since it is undoubtedly a complex and contains Co", it is either CoII[CoII(CN),].nH*Oor a complex in which the ratio of cyanide to cobalt 1s not exactly 2. A new analstir Is desirable.
+
Since the liberated CoII complex can recycle v i a reactions 6-8, the net result is the one which has been so often observed when solutions of [CoII(CN)5I3- have been heated 2H20
+ 2[Co"(CN)s]'- + 2CN- +
~ [ C O " ' ( C N ) ~ ] ' -4- H2
+ 20H-
When the solutions are cool and free from precipitate, little or no HZ is generated and the final stage of aging yields polymers which may take the form of threads or globules of colorless jelly. Since the products have negligible absorption in the visible or near ultraviolet regions, the reaction is difficult to follow spectrophotometrically. Poor mixing of solutions, causing a local deficiency of cyanide, can result in precipitation of the green complex regarded by Descamps' as KZCo[Co(CN)e] and by Griffith and Wilkinson12 as C O ~ [ C O ~ ( C N )Since ~ ~ ] . as we have already reportedg the amount of green complex formed is proportional to the amount of alkali added but not to the concentration of KC1, neither formula is convincing. It is possible that the green compound analyzed by the above authors was the one which forms on aging. At temperatures below 3" we have observed a black precipitate, cobalt metal or cobalt oxide, which disappears when the temperature is raised a few degrees. If aging in dilute solution is found to exceed a few yo loss of {COII(CN)~]~per day, i t can be assumed that added salt or impurities is responsible. Oxygen is the most common cause. The absorption spectra of the products of oxygenation and subsequent oxidation will be discussed in a later paper. Significance of the Salt Effect.-Mills, Weller and Wheeler" have pointed out the strong influence of added salt on the observable kinetics of homogeneous Hz uptake, so strong that we may write approximately I
(6)
a reaction for wh ch close approach of the two negatively charged ions without formation of the Co-Co bond is probably adequate. The next phase of the aging process is the irreversible formation of the very stable hexacyanocobaltate(II1) ion [Co'I'(CN)sOH]'-
i'ol. s 3
= k[K+][Co]
(9)
up to values of [K+]/[Co] in the neighborhood of 25 (Fig. 4, ref. 8). Here Y is the reaction rate, [K+] the total potassium ion concentration and [Co] the concentration of [CoII(CN)513- (equal a t the beginning of the experiment to the total cobalt Concentration). When the only source of K + is KCN, R has the value [K+J/[Co] and expression 9 becomes I =
kR[Co]*
Thus our earlier findingla that r is proportional to [CoI2when [Co] is varied while R is held constant, should be interpreted to mean that the rate is first order in [COII(CN)~]~-, a t least in the dilute solutions which we employed. I n the same papers we interpreted the observed kinetics to support [COII(CN)~]~-, present in very small amounts in equilibrium with [COII(CN)~]~-, as the ion which reacts with Hz. The absorption band thought due to the hexacyanide ion has since proved to be that of a product of disproportionation. A comprehensive search has shown that if [CoII(CN)6l4- can exist in solution i t is a t concen-
Aug. 20, 1961
LIQUID PHASE HYDROGENATION O F CYANOCOBALTATE(I1)
337 1
trations below the limit of detection by spectro- mal ligand position. The absorption spectrum of photometric means, even in the presence of meth- the reduced cobalt complex (Fig. l), with its single anol or CsC1. I n addition Mills, Weller and band, situated in the near ultraviolet region, is Wheeler" have shown that added salt in the form better explained in terms of an octahedral than a of KC1 is about as effective as excess KCN in ac- 5- or 4-coordinate structure. That i t contains not more than 5 cyanides per celerating H, uptake, thus confirming the irrelecobalt can be shown by the lack of precipitation vance of [COII(CN)~]~to the reaction mechanism. The above statement regarding the equivalence when the hydride is formed in high yield in an R = of KC1 and KCN as promoters requires amplifica- 5 solution. Proof that there are no less than five tion. By taking appropriate rates from the curves cyanide ligands is still lacking. As evidence given in Fig. 4 of Bayston, King and Winfields against a lesser number may be cited the very and in Fig. 2 of the present paper, i t can be esti- small activation energy of H2 uptake. If the reacmated that KCN is about 15y0more effective than tion required dissociation of a cyanide ligand, an KC1. The discrepancy is accounted for as follows: activation energy of 20-50 kcal./mole of [CoII(i) KCN increases the @H of solutions of [CoII- (CN)6I3- would be expected. Structure of the Oxidized Complex.-Spectro(CN)6I3-, and KOH is known5 to be 10-20% more effective than KCl in accelerating Hz uptake; (ii) photometrically and chemically the more oxidized spectrophotometric measurements (unpublished) of the two disproportionation products appears to show that KCN induces less disproportionation be identical with the 380 mp complex which has been prepared by several investigators20v21by difthan KC1. It will be noticed in Figs. 2 and 3 that the varia- ferent methods and regarded as [ C O ~ ~ ' ( C N ) ~ O H ] ~ - . tions in promoter effectiveness through the scale It can be formed also by irradiation22of pure of cations from Li+ to Cs+ do not match with aqueous K3Co(CN)6 and shown to contain five those which have been observed in Hzevolution cyanide groups per cobalt atom, with OH or H20 and in the disproportionation reaction by which it as the only possible ligands for the sixth position, is p r e ~ e d e d . ~In particular the anomalously large providing that the complex is monomeric. We effect of Cs+ on disproportionation does not find a have chosen to write it as the hydroxo-complex parallel in Hz uptake, suggesting that a different since the pH in most of our experiments is 10 or property of the cation is utilized (perhaps hydration more. The absorption spectrum given in Table I is number in disproportionation and ionic radius in that of the purified complex prepared either by H P uptake). Structure of the Reduced Complex.-The prod- irradiation of [COIII(CN)O]~-or oxidation of uct of electrolytic reduction4 of [ C O I I ( C N ) ~ ~ ~[Co1I(CN)6l3-, followed by column chromatoghas usually been considered to be a CoI complex, raphy.22 I n the presence of 20% ethanol the 380 probably [CoI(CN)5I4--, since five cyanide ligands mp peak is shifted to 382 m~ and in absolute alcohol would confer on the cobalt the electronic structure t o 388 m,u. When the complex is formed in dilute of the inert gases. We have regarded it instead solution by disproportionation of [ C O I I ( C N ) ~ ] ~ as a hydride of CoIII since i t can be formed by induced by added CsCl, the band is a t 382-396 reaction of [ C O I I ( C N ) ~ with ] ~ - H2without apparent mp. The displacement from 380 mp may be exliberation of protons.8 It can be shown that the plained by overlapping of the 428 mp band-of residcomplex persists in the presence of considerable ual [Co"(CN)513-. KOH and therefore does not behave as an acid. I t should be noted that the p H behavior of mixGriffith and Wilkinson12 have recently shown by tures of CoClz and KCNg will depend largely on nuclear magnetic resonance measurements that the proton affinity of [Cor11(CN)601
