THE CORRELATION BETWEEN RATE OF OXIDATION AND POTENTIAL IN IRON SYSTEMS

BY L. MICHAELIS AND C. V. SMYTHE*

(From the Laboratories of The Rockefeller Institute for Medical Research, New York)

(Received for publication, August 21, 1931)

It is well known that in general the rate of a chemical reaction has no relation to the free energy change accompanying this re- action. The change of free energy depends only on the state of t.he reactants before and after the reaction concerned. The rate, however, depends on the intermediate states also and can be varied by cata1yst.s. If, however, we take any given reaction and break it up into sufficiently elementary processes we must find one process for which there are no intermediate steps and for this process the rate might be in some way or other intimately related to the free energy change. As far as biological oxidations are concerned we know that in many cases one intermediate step seems to be the formation of an iron, or at least some heavy metal, complex. This is followed by a change of valence on the part of the iron. This change of valence is a very elementary process and it is this process with which the present paper is concerned. We shall endeavor to show that the rate at which this process occurs in a few simple systems runs parallel with the normal potential of the system in which the process occurs.

During the course of previous work (1, 2) it became apparent that the order of the normal potentials of various reversibly oxi- dizable and reducible iron complex systems closely paralleled the order of the rates at which the ferro members of the systems were oxidized by molecular oxygen. By the normal potential of a system we mean the potential of a solution containing equal parts of the ferro and ferri forms of the complex. This normal potential de- pends on pH so we wish to restrict our comparison between the

* National Research Council Fellow in Biochemistry. 329

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

330 Iron Systems

potential and the rate of autoxidation to iron systems at constant pH. It is necessary, of course, that the pH chosen be compatible with the existence of the complex.

It is not asserted that in general potential and rate run parallel. Our assertion is that if we examine the normal potentials of a series of iron complex compounds at any constant pH, the order of the potentials is the same as the order of the rates of autoxidat,ion of the ferro form of these complexes. The more negative the po- tential, the more rapid is the oxidation by molecular oxygen.

In a previous paper (1) it was shown that the rate of oxidation of ferrous compounds by molecular oxygen was greatly dependent upon the electrical condition of the ferrous iron. Ferrous ion was oxidized much more slowly than the electrically neutral ferrous iron atom contained in various compounds. In a subsequent paper (2), the normal potentials of various iron complex systems were investigated. Before demonstrating the validity of the above rule for these data we shall consider certain iron complexes not investigated in the earlier papers, for these cases will be part,icu- larly demonstrative.

First we may consider the case of some cyanide complexes of iron. As previously pointed out (l), although in ferrocyanide the iron atom is tightly held in an unionized condition it is still difficult for it to be oxidized. This was considered to be due to the fact that the iron atom is surrounded by four negative charges and in order for an electron to escape from the kernel of the iron it must move out against this negative atmosphere. This is a different consideration than the charged condition of the iron atom itself and one cannot,, offhand, predict how important it will be in comparison with positive charges on the iron. However, the only thing that need concern us here is whether or not the aut- oxidizability and the normal potential run parallel. We have seen that ferrocyanide is not readily autoxidizable; then it should be a comparatively poor reducing agent and its normal potential should be relatively positive. As Fig. 1 shows, the potential is quite positive. As the normal potential of a system containing such polyvalent ions is very sensitive to changes in the ionic strength we have to make the comparison of various compounds of this kind at a constant ionic strength. We shall compare them under the conditions established by dissolving a very small amount

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

L. Michaelis and C. V. Smyt,he 331

of the complex in a 1.0 M KC1 solution. Under these conditions the potential was found to be +0.480 volt when expressed against the normal hydrogen electrode.

It is possible to substitute one of these cyanide groups of ferro- cyanide for any one of several other groups and Davidson (3) has measured t)he potentials of some of the derivatives. We would expect from our considerations that the substitution of an NOz- group for a CN- group would not change the pot!ential very mark- edly and to the extent t,hat it did change it we would expect the value to become more positive. The value reported by Davidson,

12

I -0.6 -0.4 -0.2 0 so.2 to.4 to.6 to.8 t1.0 t.

volt5

FIG. 1. The normal potentials of various iron systems

0.516 volt (3), is in fact somewhat higher. We would also expect that the replacement of a CN- group by an NH3 group would appreciably lower the potential, for this reduces the negative charge of the molecule by 1 unit. The value found by Davidson, when measured in an excess of ammonia, was 0.374 volt (3). Likewise we would predict that if a CN- group is replaced by Hz0 t,he resulting potential will be considerably lower and the complex in its ferro form will be more inclined to autoxidation. The latter is true. The potential, however, according to Davidson is not more negative than that of ferrocyanide, but even more positive.

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

332 Iron Systems

He reports the potential of this aquo compound to be 0.491 volt (3). This value is higher t,han that of the ferrocyanide. Of the various compounds to be considered in this paper this penta- cyanoaquoferroate is the only one that is out of its predicted place. We may then justifiably harbor a suspicion concerning the accuracy of this value. This suspicion is further increased by the fact that Davidson reports difficulty in measuring this potential. He attributes this difficulty to the formation of an intermediate compound containing 3 molecules of the oxidized and 2 of the reduced form.

On the basis of the above suspicion we have repeated the po- tential measurements on this compound and for the reasons fully set. forth in the experimental part we believe that Davidson’s value is in error. As pointed out later, the assignment of an Eo value to such an irregular system involves some assumptions. If these assumptions be granted then our Eo for the system is 0.418 volt at pH 6.6. This value puts this system in its expected place in our arrangement (Fig. 1). Even if this value is not correct our results clearly show that Davidson’s value is much too high and that this system is, at least, not a contradiction of our rule. With- out, insisting too much on the fact that our corrected potential value shifts this complex to the expected place in the order of aut,oxidizability, one can say that this complex is not a simple complex with only one iron nucleus and therefore not comparable with the others and not suitable for the application of our rule.

A corollary of the theory proposed is that all complex ferro cations, in which the iron has retained its positive charge, should be difficult t,o oxidize, and that a mjxt,ure of these with the cor- responding ferri complexes should have a very positive potential. One beatitiful example demonstrates the validity of this state- ment. The ferro complex of CX, or’-dipyridyl, investigated by Blau (4), when written in modern fashion, according to Werner’s scheme, has the constitution shown in Formula I.

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1,. Michaelis and C. V. Smyt,he 333

Formula I

This complex is ent,irely stable toward oxygen and so difficultly oxidizable that not even bromine, but only chlorine or perman- ganate, can oxidize it to the ferric state. Exposure to the air even in alkaline solution for any length of time does not oxidize this red ferro complex into the blue ferric.

We attempted to measure the potential of the dipyridyl com- plex by titrabing ferrous dipyridyl with an aqueous solution of chlorine. The potential range was very posit’ive indeed, being much more positive than 1.0 volt, even much more positive than that of the system ferric chloride + ferrous chloride, but no steady potential could be obtained. These unsteady potentials may be due in part to the fact that all noble metal electrodes no longer work as indifferent reversible electrodes when the potential range approaches that of the oxygen electrode and in part to the fact that t’he ferri complex is not a very stable compound but liable to intramolecular rearrangement in such a way that the organic part of the complex is oxidized and the iron reduced to the ferro state, as has been observed by Baudisch (5). Another contri- buting cause may be that the chlorine used as oxidant will in part directly attack the organic part of the molecule in addition to oxidizing the iron.

Now we may draw a graph (Fig. 1) comprising all pot,entials of the various systems, including those just discussed. We do not have measurements for all the systems at any one pH, but we can compare the range of the potentials, say at pH 5.0. The order of the potentials is: dipyridyl, chloride or sulfate, (CN)6N02,

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Iron Systems

(CN)e, (CN),HzO, (CN),NH, salicylate, malonate, oxalate, pyro- phosphate. This is precisely the order of autoxidizability. The ferro dipyridyl complex is not autoxidizable at all, the free ferrous ions of an acidulated solution of ferrous sulfate, as well as the ferro- cyanide complex, are very stable, but the next, pentacyanoaquofer- roate and pentacysnoaminoferroate, are definitely autoxidizable, bhough at a relatively slow rate. Ferrous salicylate and malonate are readily aut’oxidizable. A solution of ferrous salt dissolved in an excess of salicylate at pH 5.0 and exposed to air gradually develops the violet color of the ferri complex. The rate of this oxidation is of a medium, measurable order of magnitude. Ferrous oxalate and pyrophosphate absorb oxygen so rapidly that the rate can scarcely be measured in a micro respiration apparatus. Thus, although we cannot offer precise quantitative data for the different rates of autoxidation, the order of magnitude differs so largely that the validit#y of our st,atement is obvious. At any other pH some of the compounds can be compared with each other in a similar way and with the same result. As we compare various iron complexes at any other constant pH, the number of such complexes compatible with this pH may be smaller but those available for a comparison will always be in hhe correct order.

The physiological significance of the rule stated above may be expressed as follows. The various porphyrin iron complexes fulfil at least two different functions. Some of them serve as oxygen carriers, others as catalysts for oxidat,ion of other substances. The best known oxygen carrier, hemoglobin, is a ferro compound. To fulfil its function as oxygen carrier, the iron of hemoglobin must not be oxidizable by molecular oxygen. The oxygen compound of its ferro state must be stable in this ferro state. On the other hand the catalytically acting heme pigments can fulfil their func- tion only if their iron nucleus is autoxidizable in a reversible way. We must then distinguish autoxidizable and non-autoxidizable heme pigments. Thus hemoglobin is not autoxidizable, but de- natured hemoglobin is. Such considerations suggest the ques- tion, upon what, measurable properties does the autoxidizability of an iron complex depend? It seemed desirable to investigate t,his problem for simpler iron complexes with the hope that what- ever results were obtained could be applied to the heme pigments later. Potentiometric measurements on the heme pigments have

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

L. Michaelis and C. V. Smythe 335

been performed by Conant and coworkers (6) but the data pre- sented by these authors do not seem to be sufficiently definite to justify the application of the principle stated above, so we must restrict ourselves to merely mentioning the physiological point of view which underlies the rule stated in this paper.

EXPERIMENTAL

The sodium pentacyanoaquoferroate [Fe(CN)sHzO]Nas was prepared by treating sodium nitroprusside with hydroxylaminc

1 2 h 4 5 6 7 8 Hours

FIG. 2. Oxygen consumption of sodium pentacyanoaquoferroate solu- tion. The experiments were carried out in a Warburg respiration apparatus at 22”. Each flask contained 1.6 cc. of 0.02 M solution. If the solution is made more acid than about pH 4.0 the rate of oxidation again decreases. This decrease of autoxidizability is accompanied by a color change of this ferro solution from the usual yellow color to a blue-green color, similar to the color of the ferri compound.

hydrochloride according to the method of Hofmann (7). The corresponding ferri compound was prepared by oxidizing the ferro compound with bromine water at low temperature, again in the manner described by Hofmann (7). These procedures led to a yellow powder for the ferro preparation and a very dark blue

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

336 Iron Systems

powder for the ferri compound. The aqueous solutions have about the same colors as the powders.

In agreement with Davidson (3) it was found that when a solution of the ferri compound was mixed with an excess of the ferro compound the color of the ferri solution immediately faded out. This is not due to an optical mixing of the colors, as can

560

FIG. 3. The effect of pH on the potential of the system sodium penta- cyanoaquoferroate-sodium pentacyanoaquoferriate, established by mixing the two compounds. 0 pH 4.6 (0.2 M acetate buffer), X pH 6.0 (0.2 M

acetate buffer), A pH 8.0 (M/15 phosphate buffer).

easily be determined, but to some intermediate compound for- mation.

In working with systems such as this there are several factors that must be closely controlled. The first, which has already been discussed, is change in ionic strength. Anot,her danger arises from the fact t’hat the ferro compound is oxidized by the oxygen

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

L. Michaelis and C. V. Smythe 337

of the air. This autoxidizability is shown in Fig. 2. Such autoxi- dizability itself may be taken as an indication of the fact that the potential of the system is more negative than that of the ferro- cyanide system. If this autoxidizability is not guarded against, its effect on the titration curve obtained by mixing the two forms is to flatten it at both ends and to place the entire curve in a too

340, I I I I 1

300 f&--~ ‘!ct+ti

Per cent reduction FIG. 4. The titration of sodium pentacyanoaquoferroate with reduced

rosinduline. 1 0.16 mM in volume of 25.0 cc., O- 0.08 mM in volume of 25.0 CC., -0 0.008 mM in volume of 25.0 cc. The curves drawn are those calcu- lated for two steps of oxidation, each involving 1 electron. To refer these potentials to the normal Ha electrode add 242 millivolts.

positive range, since the ratio of ferri to ferro is always greater than that calculated. The third factor that must be cont,rolled is the pH. This was not done in Davidson’s work. In t.he case of ferrocyanide it is not important to work at constant pH, for the potential of this system, at pH more than 5.0, is independent of pH. However, this is not true for the aquo compound under consideration as Fig. 3 shows. The solutions of the ferro and ferri

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

338 Iron Systems

compounds have quite different acidities. Measurements with a glass electrode showed that a 0.01 M solution of the ferro compound had a pH of 8.5. A similar solution of the ferri compound had a pH of 6.0. If one wishes to maintain a constant pH during the mixing of such solutions the necessity for buffering is evident. A small part of the difference shown by the curves in Fig. 3 may be due to changes in ionic strength, but since all solutions were 1.0 M

with potassium chloride this change would be small. In order to meet the above requirements we chose the following

conditions. The ferri compound to be used was weighed out and dissolved in 1.0 cc. of water. To this were added 3.0 cc. of a ~/15 phosphate mixture and 22.0 cc. of 1.0 M KCI. The pH of such solutions was 6.6. This was titrated wit’h a solution of reduced rosinduline of appropriat,e concentration. The results obtained are shown in Fig. 4.

The most striking t.hing about the curves is that they are much too steep as compared with those of other complex iron systems. The change from the ferri to the ferro form should be a 1 electron change and this should correspond to a potential change of approxi- mately 30 millivolts for a S-fold change in the rat.io of the ferri to the ferro form. The actual potential change is about twice this. The only apparent explanation for this is that both the ferri and the ferro compounds exist in solution in a polymerized bimolecular form and that an intermediate compound containing both ferri and ferro molecules also exists. From the fact that the curves are symmetrical about the mid-point we may assume that the intermediate compound present at this point contains the ferri and ferro forms in equal proportions. If this be true then the potential at this point is the Eo of an imaginary system con- t,aining only the ferri and the ferro forms of the complex. As may be seen, this is 0.418 volt at pH 6.6. If we consider t,he E. potential for the first step consisting of the oxidation of the bi- molecular ferro compound to the mixed ferro-ferri compound, this is even much more negative.

The three experiments in Fig. 4 represent three different concen- trat’ions. Apparently, from the mid-point onward the three curves are accurately the same, but from t,he beginning to the mid-point there is considerable deviation. This finds a ready explanation if we assume that the polymerized ferri form is in equilibrium with a

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

L. Michaelis and C. V. Smythe

monomolecular ferri form. Any change in concentration or other conditions would affect this equilibrium and hence the potential.

If we employ the same conditions we can duplicate the curves shown, either by reducing the ferri compound with reduced indigo tetrasulfonate instead of rosinduline or by mixing the ferri and ferro compounds. It is more difficult to obtain as good results by oxi- dizing the ferro compound for at the temperature employed (30”) no good end-point is obtained, probably due to irreversible oxi- dation of the cyanide by the strong oxidant which it is necessary to employ (Cl2 or BrJ. In Fig. 4 the resulm for this complex are plotted for three experiments with varied initial concentration of the ferri compound which is titrated with reduced rosinduline as a reductant. The drawn out curves are calculated on the assump- tion that there are two steps of reduction each involving the addi- tion of 1 electron without any change in the molecular size. The agreement of the calculated and the observed values is cer- tainly not as good as usual in stable reversible systems. The slight though obvious dependence of the upper end of the curve on the initial concentration was consistent in repeated experi- ments. On considering the instability of this complex one might be inclined to attribute these deviations to impurities due to secondary decomposition products. If this be true the only possible interpretation is that this complex is bimolecular and exists in the ferro-ferro, ferro-ferri, and the ferri-ferri states. 1 molecule of Hz0 may be considered to function as a bridge holding together two Fe(CN)6 radicals.

It is scarcely thinkable that the slight deviations from the theoretical curve might require a complete abrogation of this interpretation. It cannot be decided however whether the de- viat.ions are due to the mentioned impurit’ies or to the fact t,hat the association of each 2 molecules to a double molecule is not entirely complete but in equilibrium with the monomolecular state. This would involve a dependence of the potential on the absolute concentration of the substance, as has been pointed out in pre- vious papers.

SUMMARY

It is shown that for a series of iron compounds the autoxidiza- hility of the ferro compound at a given pH closely parallels the

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

340 Iron Systems

normal oxidation reduction potential of the system ferro com- pound-ferri compound, at the same pH. The more negative the potential, the greater is the autoxidizability.

It is shown that the normal potentials of various iron systems reported in the literature are, with one exception, in good agree- ment with the above statement and with the potent,ials expected from a knowledge of the structure of the compounds.

The potential of the system sodium pentacyanoaquoferroate- sodium pentacyanoaquoferriate, the excepbion noted above, has been reinvestigated. We are unable to confirm Davidson’s value of 0.491 volt, expressed against the hydrogen electrode. Our data indicate that this is a very irregular system and the inter- pretation of the results involves some assumptions. These as- sumptions being granted, the exceptional position of this complex vanishes.

BIBLIOGRAPHY

1. Smythe, C. V., J. Biol. Chem., SO, 251 (1931). 2. Michaelis, L., and Friedheim, E., J. Biol. Chem., 91, 343 (1931). 3. Davidson, D., J. Am. Chem. Sot., 50, 2622 (1928). 4. Blau, F., Monatsh. Chem., 19, 647 (1898). 5. Baudisch, O., Biochem. Z., 92, 189 (1918). 6. Conant, J. B., J. Biol. Chem., 67,401 (1923). Conant, J. B., Alles, G. A.,

and Tongberg, C. O., J. Biol. Chem., 79, 89 (1928). Conant, J. B., and Tongberg, C. O., J. Biol. Chem., 86, 733 (1930).

7. Hofmann, K. A., Ann. Chem., 312, 1 (1900).

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from

L. Michaelis and C. V. SmytheIRON SYSTEMS

OF OXIDATION AND POTENTIAL IN THE CORRELATION BETWEEN RATE

1931, 94:329-340.J. Biol. Chem. 

  http://www.jbc.org/content/94/2/329.citation

Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

alerts to choose from all of JBC's e-mailClick here

  ml#ref-list-1

http://www.jbc.org/content/94/2/329.citation.full.htaccessed free atThis article cites 0 references, 0 of which can be

by guest on August 24, 2020

http://ww

w.jbc.org/

Dow

nloaded from