Atmospheric Corrosion

A model of electrolytic corrosion is de- veloped. It is shown that electrically conducting channels threading through the oxide layer and connecting anodic and cathodic areas, obey the equation for a reactant being catalyzed by its product. The resulting autocatalytic equation is compared with available experimental data and found to be widely applicable and capable of unifying many experimental observations.

In an earlier paper (1), the theory that atmospheric corrosion obeys the equation for an autocatalytic reaction was briefly sketched. Here we develop the theory in more detail and examine its applicability in specific cases. A characteristic feature of atmospheric corrosion is the occurrence of alternating wet and dry periods. Thus, periods when an exposed metal surface is covered by a thin conducting film are followed by periods when the surface is dry. Unless surfaces are completely uniform there will be areas which preferentially catalyze the reactions which extract electrons from the metal and are, therefore, cathodic and other anodic areas where positive metal ions pass into solution donating their electrons to the metal. These electrons then stream toward the cathodic areas with the circuit being closed by ionic conduction through the moist film. Influences which interrupt this circuit stop electrolytic corrosion.
Typically, hydrogen ions are discharged on to the metal surface at the cathodic areas. What happens to the resultant hydrogen atoms then depends upon conditions. If no oxidizing agent such as oxygen is present, the hydrogen atoms may pass into the metal causing embrittlement or they may be discharged as hydrogen molecules. A marked slowing of the evolution of hydrogen molecules with a corresponding slowing of corrosion can be brought about by the addition of compounds containing a group V element such as As203. The arsenic poisons the catalytic sites which otherwise catalyze the reaction 2H-H2 (gas). Oxygen, by combining with hydrogen atoms to make OHor H202, provides another means of escape of the hydrogen atoms. This outlet may be eliminated by excluding 02 or by reacting it with reducing agent such as hydrazine or sulfite ion and so removing it. Interrupting the corrosion circuit in the moist conducting film by excluding electrolyte, or by painting the metal surface with pitch or some adherent paint, or by accumulation of an adherent nonconducting precipitate, also slows corrosion.
The potential, Av, driving the current through the liquid film is the thermodynamic potential between the anodic and cathodic areas after one has subtracted out the potential drop across the two Helmholtz double layers. For a sufficiently poorly conducting film the potential drop across the Helmholtz double layers can be neglected. In any case we have for the current, I, per unit area between anodic and cathodic 245 areas I = Avota. [1] Here a is the conductance per square centimeter between anodic and cathodic areas. Consider for example an iron plate covered by a weight of zinc, W, in grams per square centimeter and let N, and N2 be the anodic and cathodic fractions of the surface. Let us suppose that (a) the surface of the metal is covered by a moist membranous film having n channels per square centimeter connecting different areas on the metal surface and that the channels have a mean conductance KX where K is the specific conductance in the channel and X is the crosssectional area of the channel divided by its length. We suppose further (b) the chance that a channel terminating on any area of the metal is proportional to that area. Then the average conductance per square centimeter, a, between anodic and cathodic areas is a = nKXNlN2f,. Here, fi is the fraction of the time the oxide film is wet enough to be conducting. Eq. 1 then takes the form d~(Niwz96,5OO) = AvnKXNlN2f,. [2] Here z is the valence of zinc; 96,500 is the Faraday in coulombs; and M is the molecular weight of zinc. Here r is the half-life, i.e., the value of the time when N, = '/2 Rearranging Eq. 5, we get N, (1 + exp( k (Tt))) and N =( expQ-( -) [3] [4] [5] [6] [7] Eq 6 is also the equation for autocatalysis and for survival of a homogeneous human population (1). In the electrolytic rusting process, anodic area (reactant) is changing over into cathodic area (product). For conducting channels to implement this electrolytic corrosion, a channel must connect the two types of areas so the rate of reaction is proportional to the reactant, N1, in this reaction, times the product, N2, which is the condition for autocatalysis.
It is to be expected that the half life, r, should obey the equation Here, rn is the half life for 1 g/cm2 of protecting metal and w is the actual weight of this protective metal coating. If the plate starts out corroded or scratched to an extent that corresponds to an age t = ro, then this is taken care of in Eq. 8 by writing r = riw -ro. On the other hand, if the surface is completely homogeneous there can be no electrochemical corrosion and the appropriate value for To might be very large and positive. For the usual case which we will encounter.  (Fig. 1) was used to determine k and r1. The dotted curves represent the average values of k and iT from the rectified raw data (Fig. 1). 0, raw data from Brunot Island; , N2 = [1 + exp(k(rit/W))] -1; ..., N2 = [1 + exp(k( -t/W))] -1.
O= 0 or very nearly so. Introducing T = TrW into Eq. 7 gives N2 = (1 + exp(k (wri -t))) = (1 + exp(k Ti_ - [9] and Eq. 5 takes the form For a fixed value of N2 we can write ln( N)= C = k--) orw= t(<k) [11] In Fig. 1, Eq. 10 is applied to data for the fractional extent of rusting, N2, of galvanized sheets as a function of the time, t, in years, in the industrial atmosphere of Brunot Island near Pittsburgh, Pa., as reported by Reinhard (2) and by Burns and Bradley (3). The excellent straight lines obtained provide impressive support of our theory of atmospheric corrosion. Fig. 2 is the result of 11 similar plots for the "progressive development of rust at Brunot Island, on 4 X 6-inch steel specimens plates with cadmium and zinc and on corrugated galvanized sheets" (4). Plots of the duration of protection (in years) of zinc-coated steel at Sheffield, England against ounces of protective covering per square foot yields a straight line as Eq. 11 predicts (4). The weight of zinc coating in ounces per square foot plotted against years of exposure to produce initial rust gives straight lines with characteristic slopes for eight different regions as Eq. 11 predicts (4). If in Eq. 9, we insert the constants k = 2.72, T = 3.54, and w = 2.50 which correspond to the values found from the line farthest to the right in Fig. 1, and we do the same for the other curves, we find the calculated and experimental curves are practically indistinguishable (Fig. 3)  tells us about the nature of the channels in the oxidized layers through which the corroding currents flow. The average value of k for galvanized sheets is 3.56 (expressed in ounces per square foot per year). This value must be multiplied by 0.983 X 10to convert it to g cm-2 sec'. Thus, we get 3.48 X 10-9 = 0.3 (fif2)/l (1.9 X 10-6)/(2 X 96,50Q)65.37. Here, we have taken the effective voltage as 0.3 which is approximately the reversible potential; 1.9 X 10-6 is the specific conductance of water; 65.37 is the atomic weight of zinc; 96,500 is the Faraday. We have redefined nX -_ f2/21 where f2 is the fraction of the surface covered by conducting channels and 1 is the average length of the channels. The factor 1/2 comes in because both ends of a channel must lie on the metal surface. If we estimate that a channel is in a conducting state fi = 1/10 of the time on Brunot Island, we get 1 = f2/283 and if we suppose further that f2 = 1/100, i.e., one hundredth part of the surface is conducting, we get 1 = 3500 A. Without more definite information to pin things down, one can only say that such a length of channel would not be surprising. Our model of conducting channels, which thread through the oxide coating and connect, catalytically, anodic and cathodic areas, is consistent with the practice of letting galvanized sheet weather a while before blocking the conducting channels with a suitable additive and with pitting and preferential etching at dislocations and grain boundaries.