HERBERT H. UHLIGAND GLENNE. WOODSIDE production might lead to the blocking of destru'ctive reactions through the device of maintaining other reactions a t constant electrochemical affinities. It seems likely that new light could thus be thrown on the problem of corrosion inhibition. For example, the electrochemical mechanism which we have proposed with Pourbaix14 for the inhibition (14) M. Pourbaix and P. Van Rysselberghe, Corrosion, 6, 313 (1950).
VOl. 57
of the corrosion of iron by nitrites, chromates or even oxygen when present in sufficient amount lends itself to interesting Further study on the basis of irreversible thermodynamics. Let us note in this connection that, through the approximate procedure of linearization of polarization curves the practical range of usefulness of the above considerahons can be considerably enlarged. We ho e to return to this and other topics of theoretical and appied electrochemical thermodynamics in later communications.
ANODIC POLARIZATION OF PASSIVE AND NON-PASSIVE CHROMIUM-IRON ALLOYS BY HERBERT H. UHLIG AND GLENNE. WOODSIDE Corrosion Laboratory, Department of Metallurgy, Massachusetts Institute of Technology, Cambridge, Massachusetts Received November 86, 1868
Anodic polarization data for 0-16,770 Cr-Fe alloys in 3y0NalSOl are reported. Critical current densities for passivity accompanied by sudden reduction of anodic dissolution and a pronounced noble potential range from 150 ma./cm.* for iron to 0.05 ma./cm.Z for 11.6% Cr. At 12% Cr and above, a critical current no longer exists; instead any small polarizing current produces a marked shift of potential in a noble direction. Small corrosion currents, therefore, can equipotentialiae the surface and stifle the corrosion process in accord with the electrochemical mechanism of corrosion. Since 12% Cr and above mark the stable passive compositions (stainless steels) in solutions like NazSOd, i t appears that the property of socalled passive alloys to equipotentialiae themselves is an important characteristic of passivit and perhaps an essential condition to their remarkable resistance to corrosion. Coulometric measurements show that atout 0.014 coulomb/cm.z or less is concerned with the passivation process for the 9.2-16.770 Cr alloys and probably for other compositions as well. This corresponds a t most to a few monolayers of subatance on the passive surface. Thick oxide films, therefore, cannot possibly be the primary cause of passivity.
When iron is polarized as, anode a t moderate current densities in dilute sodium sulfate or sulfuric acid, the major reaction is Fe -+ Fe++ 2e. As the current is increased, a critical value is reached at which the potential suddenly changes to a value more noble by about 2 v., and the anodic reaction changes to
+
20H-
-+
+ HzO + 2e
1/z02
accompanied by the reaction Fe -+ Fe+++
'
+ 3e
Iron is said to become passive, both because it assumes a more noble potential and because oxygen evolving a t its surface resembles the behavior of a noble metal electrode. Also, the dissolution rate is appreciably less than that of an active iron anode. Soon after the polarizing current is stopped, the electrode again assumes the active state. If the current is gradually decreased, the active state is achieved a t a current density somewhat lower than is necessary to achieve the passive state when applying progressively higher currents. Conditions that apply for achieving anodic passivity of iron-chromium alloys substituted for iron have not yet been studied. Information of this kind should be of value to a better understanding of passivity, particularly since passivity is established in a more stable form when the chromium content reaches 12% or more, This composition establishes the basis for the modern stainless steels.
Experimental Arrangement Polarization data for the Cr-Fe alloys were obtained using the cell shown in Fig. 1. This was placed in an air thermostat maintained a t 25 f lo. Two circular platinum cathodes 3.1 cm. in diameter are located 1 cm. from either side of the alloy anode. The latter measures approximately 1 cm. square by 2 mm. thick. The electrolyte IS 3% NazSOd.
A tubulus, sealed a t each end with asbestos fibers and filled with the same solution is mounted on one side of the anode close to the metal surface. The opposite end is immersed in a 0.1 N KC1 solution into which an Ag-AgCl electrode also in 0.1 N KCI is immersed. Potentials were measured using a vacuum tube potentiometer. No correction was made for liquid junction potentials, since these were assumed to be approximately constant throughout the measurements and, in any case, were small. Current was supplied usually by a high-vqltage B battery through a high resistance, thereby maintaining constant current, actual values of which were read using a precision (fl/,%) milliammeter and also, for very small currents, by the potential drop across a precision 100-ohm resistance. For many measurements, the sodium sulfate solution was first freed of chlorides by adding silver sulfate and removing excess silver by electrolysis for several hours between platinum electrodes. Contaminating heavy metals were removed simultaneously. The pH was then adjusted to 7 by adding sodium hydroxide. This urification procedure did not appear to be critical, since vaLes of potentials were the same within the experimental error whether or not preelectrolysis was carried out. Electrode materials consisted of electrolytic iron, and of laboratory melts of chromium-iron alloys. Alloys collected from three sources gave consistent results indicating that small differences in impurities are not important. Thermal history, however, was felt to be a significant factor, and, hence, all the alloys were heated in. helium at 1000° and water quenched. Analyses are gwen In Table I. Before measurements were made, the solutions were deaerated about one hour by bubbling nitrogen through them. Nitrogen was previously purified by passing it over copper turnings maintained at 400'. The gas entered the cell through a sintered glass disc providing a fine dispersion of gas bubbles, but during measurements the gas was bypassed through a glass tube at the side of the cell, so that bubbles did not impinge on the anode. This was necessary because stirring had some effect on the critical current for passivity, greater current being necessary the higher the stirring rate. Preparation of Electrodes and Procedure.-The anode was attached to a nickel wire by spot-welding and then sealed into a glass tube using de Khotinsky Cement to cover the exposed wire and the spot-welded area. The electrode
Mar,, 1953
ANODICPOLARIZATION OF PASSIVE AND NON-PASSIVE CHROMIUM-IRON ALLOYS 281
TABLE I ANALYSES OF ALLOYS Alloys A, D, E , J were supplied by courtesy of the Electro Metallurgical Corporation. Alloys B, F, G, H, I were prepared in this Laboratory from ferrochrome and low metalloid iron with additions of manganese and silicon. Alloy C was supplied by courtesy of Armco Steel Corporation. All alloys were forged, followed by rolling or swaging, then water quenched from 1000". A B C D E F
G H I J
Cr, %
C,
2.84 3.54 5.43 6.93 8.32 9.22 11.59 12.22 13.84 16.70
0.007 .03 ,099 .016 .018 .09 .07 .07 .06
%
Mn,
Si,
0.22
0.003
%
%
VARIABLE
0.34 .34 .30 .34
.18 .17 .19 .25
45V
PT. C A T H O D E S PROBE
WATER SEAL
ELECTRODE
.007
was next pickled for a minute or less in hot 15 vol. % "0s5% vol. HF, washed successively in three vessels of nitrogensaturated water and quickly placed in the cell. The nitrogen was bubbled through the NazSOa solution for an additional 30 minutes in order to supplement the deaeration procedure carried out previously. Within this time, the opencircuit potential of the anode achieved a reasonably steady value. A small current was then impressed below the critical value for passivity until a constant measured potential was achieved, the corresponding potential was recorded and the current stopped. After the original open-circuit potential was regained, the measurement was repeated using a higher current. Eventually a current was reached at which the potential became considerably more noble and remained noble. This was designated as the critical minimum current for passivity. A preconditioning treatment of the anode using a sma!l current below the critical appeared necessary for reproducibility of measurements, perhaps in order to clean off the metal surface of adsorbed gas, ions, oxide, etc.; otherwise observed critical currents were erratically lower. The preconditioning usually gave an open-circuit potential slightly more active than before, confirming that the anode was cleaned by this treatment. Cathodic treatment of the electrode was less satisfactory, probably because of hydrogen entering the metal lattice where it could later escape and affect anodic polarization. Reproducibility was no better than the scatter obtained in usual measurements of passivity. The critical current was established to f 30/, for iron, but could be determined with progressively less precision as chromium content of the alloy increased. One factor entering the larger percentage scatter for higher chromium alloys was the very small current necessary for passivity. Electrodes polarized at currents below the critical quickly achieved the original open circuit value when the polarizing current was cut off, or reached a few hundredths volt more active, as mentioned previously. When the electrodes were polarized above the critical value for passivity, particularly for higher chromium-iron alloys, the original potentials were not regained, but instead some value more noble than the original. The critical currents on immediate repetition of the polarizing experiments were then always less. It was necessary to re-pickle the alloys in order to reproduce the original potentials and currents, especially for alloys of greater than 5% chromium.
Results A plot of potential versus logarithm of polarizing current for iron and all the alloys is given in Fig. 2. Open-circuit potentials of the alloys are placed for convenience on the ordinate corresponding to 0.0001 ma./om.*. These potentials are slightly more noble as the chromium content increases. No particular pattern seems to be followed by polarizing potentials below the critical current density, but above the critical value, data fall on a typical Tafel overvolt,age plot corresponding to oxygen evolution. The
SINTERED DISC
G L A S S c.
Fig. l.-Sketcxof
polarization cell and auxiliary equipment.
equation for excess potential over the reversible oxygen potential, or in other words the Tafel equation for oxygen overvoltage 1) is given by q = 0.83 0.10 log i where i is in milliamperes/cm.2. I t is notable that a critical current density is no longer detected for alloys above 11.6% chromium. With the 12.201, chromium alloy or above, any applied current produces a potential more noble than before and as the current is increased the potentials become progressively more noble. At no time is there a sudden break in potential as is observed for the 11.6% or lower chromium alloys. This difference of behavior was carefully checked by repeated measurements on alloys containing 9 40 14% chromium. These results imply that any small polarizing current will sensitively shift potentials of alloys containing more than 12y0 chromium in a noble direction, but for alloys of less than 12% chromium, appreciably more noble potentials are achieved only at current densities equal to or greater than the critical. However, critical current densities, when they exist, correspond to a substantial corrosion rate, so that such currents are not reached naturally in NazSOd solution. At 0.1 ma./cm.z, for example, which is the critical current density for 9.2% Cr alloys, the corresponding corrosion rate is 250 m.d.d. (milligrams/sq. decimeter/day) or 0.045 i.p.y. (inches penetration/ycar), a value which exceeds the normal corrosion rate of iron (approximately 25-50 m.d.d.). The relation of critical current density for passivity to chromium content is given in Fig. 3. Obviously, as the chromium content approaches 12%, the current necessary for passivity falls to very small values. Any corrosion process, therefore, accompanied by electric current in the neighborhood of 0.05 ma./cm.z is expected to passivate the 1J.6% chromium-iron alloy, and, hence, e uivalent corrosion rates (in Na2S04) are never greater 8 a n this value. But alloys above 12% chromium become progressively more noble for much smaller currents. It is obvious, therefore, that any corrosion process will quickly tend to equipotentialize the 12% Cr alloy surface, a tendency which in itself stifles the corrosion process in accord with the well known electrochemical theory of corrosion. Moreover, these alloys also tend to assume a noble potential in part because of the equi otentialieation process. Tge property of passive alloys to equipotentialize themselves through anodic polarization at low current densities appears to be an'important characteristic of passivity and
+
HERBERT H. UHLIG AND GLENN E. WOODSIDE
282 1.9
I
I l l ]
I
l
l
l
l
I l l 1 1
1
1
1
1
,
1
VOl. 57 I'
I l l
I
I
'1
PASS1 V E
1.3
1.1
-
-
0.9-
0.7 -
.
6 .9*/e
-
Cr
/
2. 8 % Cr J
0.5-
-
v)
.5
0.3-
8.3% Cr /
W
.OOOl
Fig. 2.-Anodic
,001
01
0.I 1.0 IO CURRENT D E N S I T Y ( M I L L I A M P / S O . CM).
IO0
polarization characteristics of Cr-Fe alloys in 3% Na2SOa: potentials us. Ag-AgC1 in 0.1 N KCl, 25"
perhaps an essential condition to their remarkable resistance to corrosion. The relation of passivity to polarization has been discussed previously by Mears and Brown.l*Z
the time was extrapolated to this value of potential. The uncertainty of extrapolation was apparently less than the uncertainty of true surface area of the electrode, which may well have varied by a factor of two from a surface pickled once to another pickled several times. In addition, the values of coulombs/cm.Z are high to the extent that the anode reactions continue throughout the time
Amount of Substance on the Passive Alloy Surface.-Some measure of the number of equivalents concerned with change from active to passive state was obtained by coulometry. Preliminary data were obtained by recording the time, using a stopwatch, for the potential to progress from opencircuit values to an arbitrary value in the passive TABLE I1 range. This could not be accomplished so simply MEASUREOF EQUIVALENTS NECESSARYTO PASSIVATIC for iron or the low chromium alloys because critical CHROMIUM-IRON ALLOYS current densities for passivity are high and the corTime t o reach 1.2 V. US. Agresponding times for passivation very brief. HowAgCl in 0 . 1 N KC1 Coulombs/ Cr, ever, for alloys of greater than 9% chromium, the cm.2 % Ma./cm.z (sea) currents are small and, hence, several minutes are 70 0.013 0.2 0.18 required to achieve passivity. From time-current .010 57 .18 measurements, the number of coulombs accom125 ,023 11.6 .18 panying the passivation process can be calculated. 100 .018 12.2 .18 Information of this kind is assembled in Table 11. .012 1350 ,016 In cases where the potentials did not reach 1.2 v., .Oll 60 13.8 .18 which was taken as the arbitrary passive potential, (1) R. B. Mears, Trans. Electrochem. Soc., 96, 1 (1949). (2) R. B. Mears and R. H. Brown, J . Eleetyochem. Soc., 97, 75 (1960).
,006
16.7 Extrapolated.
.007
1700" 1920"
.010 .013
L
Mar., 1953
ANODICPOLARIZATION OF PASSIVE AND NON-PASSIVE CHROMIUM-IRON ALLOYS
of measurement ; hence, some current goes .toward reactions other than those necessary to passivity. The data probably deserve repetition, therefore, a t such time as the true current density can be determined parallel with true surface area measurements for each electrode and with corrections for extraneous reactions. The average value of coulombs/cm.2 required for passivation of alloys listed i n Table I1 is 0.014. This value can be compared with 0.08 coulomb/cm.2 necessary to passivate iron in 2 N NaOH at a current density of 0.01 ma./cm.2, as reported by Kabanov, Burstein and Frumkin.3 The passivation process in NaOH, of course, need not be the same. Within the present experimental error of such determinations, there is no difference in the equivalents of substance necessary to passivate alloys below as compared with alloys above 12% chromium. The break in the potential versus current density curve for low Cr alloys and its absence for greater than 12y0 Cr alloys, therefore, is not related to any large difference in the equivalents of anodic reaction accompanying passivation. The value 0.014 coulomb/cm.2 corresponds to 1.2 x gram of oxygen/cm.2 of apparent alloy surface. A monolayer of close packed oxygen atoms corresponds to 0.15 X 10-6 g./cm.2 true surface, so that a roughness factor of 8 would be required if one assumes an adsorbed monolayer of oxygen atoms as primary cause of passivity. In this connection, it was recently determined that the amount of oxygen adsorbed on passive 18-8 stainless steel previously pickled corresponds to about 1.8 atomic oxygen layers (or 1 atom plus 1 molecule layer equivalent in either instance to 0.27 X gram oxygen/cm.2 true surface),4so that the roughness factor assuming this situation need be only 4.5. It can be concluded, therefore, that the amounts of oxygen required to anodically passivate chromium-iron alloys in sodium sulfate are not large, and, (3) B. Kabanov, R. Burstein a n d A. Frumkin, Discussions of the Faraday Soc., 259 (1947). (4) S. S. Lord, Jr., and H. H. Uhlig, submitted t o J . Electrochem. sac.
PER CENT,
283
CHROMIUM.
Fig, 3.-Critical current densities for pa!sivity Cr-Fe alloys in 3% NazSOa,25
.
of the
hence, thick oxide films cannot possibly be the primary cause of passivit,y. This is true of alloys just below as wellasabove 12% Cr. Theviewpreviously expressed at various times6that passivityis primarily accompanied by chemisorbed films in the order of monolayers of atoms or molecules is consistent with the presently reported facts. Acknowledgment.-This research was supported by the Shell Fellowship Committee of the Shell Companies in the U. S. to whom the authors express their appreciation. (5) I. Langmuir, Trans. ElectrocAem. Sac., 29, 260 (1916); J. Chem. Soc., (London) 518 (1940); H. ’H. Uhlig, Trans. A.I.M.E., 175, 710 (1948); Chem. Enu. News, 24, 3164 (1946); H. H. Uhlig, editor, “Corrosion Handbook,” John Wiley and Sons, Inc., New York, N. Y., 1948, pp. 24-33; H. H. Uhlig, J. Electrochem. Soc., 97, 215C (1950); leta tal Interfaces,” Am. Sac. Metals, 312 (1952).