Mechanism of Corrosion of Iron in Sodium Chloride Solution

Mechanism of Corrosion of Iron in Sodium Chloride Solution. E. G. R. Ardagh, R. M. B. Roome, and H. W. Owens. Ind. Eng. Chem. , 1933, 25 (10), pp 1116...
0 downloads 0 Views 800KB Size
Mechanism of Corrosion of Iron in Sodium Chloride Solution Effect upon the Potential of the Iron of Adding Organic Substances to the Brine E. G. R. ARDAGH,R. M. B. ROOME,AND H. W. OWENS School of Engineering Research, University of Toronto, Toronto, Canada It is probable that the apparent HE study of the characA dependable procedure has been developed by which reproducible measurements of the overvoltage is due partly to this ter, causes, and prevenresistance and partly to the adOf t h e r u s t i n g Of potential of a n iron electrode in a brine can be ditional work required to disiron and steel, though of comparatively recent date, has alObtained* This Procedure has been in charge hydrogen ions against an ready inspired a vast literature. making a general survey of the inhibiting effect adsorbed layer of inhibitor as on the corrosion of iron in brine of different against pure iron (83). Evans (6),F r i e n d (9), Pollitt A measure of the effective(IC), and S p e l l e r (18) are the classes of organic compounds. Certain classes ness for both the above types a u t h o r s Of handbooks On this of organic compounds, either singly or in cornof protective film should be obsubject. inCertain metals-for example, bination, ha.ce been shown to be tained from the rate a t which hibitors. The mechanism of this inhibiting iron atoms pass into s o l u t i o n aluminum, copper, and leadas ions. One would expect this are protected from more than a action has been partially elucidated. rate to be proportional to the very superficial corrosion in air or s o l u t i o n pressure of the metal in water by the development of a tightly adhering, impervious film of oxide, basic sulfate, or basic with respect to the surrounding solution (16). If, then, an carbonate (6). A similar protective film of oxide is formed on organic inhibitor lowers the potential of an iron electrode in a iron that has been rendered passive by immersion in chromate brine solution, we may reasonably expect the rate of corrosolution or in concentrated nitric acid (82). Evans has also sion to be lowered also. A potential-time curve should give shown that even on exposure to air a film of iron oxide forms valuable information with respect to the effectiveness of the on iron which gives temporary protection against rusting (4,inhibitor. Furthermore, such a curve may also be so inThe destruction of iron and steel by rust is usually more terpreted as to tell a t least a little of the story concerning rapid in salt solutions (refrigerator brines, sea water, etc.) the mechanism of the reaction which results in the breakthan in ordinary fresh water. The usual inhibitor employed down of the film and the resumption of the normal rate of in refrigerator brines is sodium or potassium chromate (go), corrosion. the protective action of which depends upon the production EXPERIMEKTAL PROCEDURE of a high hydrogen “overpotential” coupled with the formaThe determination of the potential of iron against brine is tion of a continuous film of oxide of iron. Certain organic compounds have been shown to be effective admittedly not an easy task. It was only after a long series as inhibitors in acid pickling baths (3, 16, 1 9 ) ; hence we may of failures that reasonably reproducible figures were obtained. find among the vast array of organic compounds either indi- (Shortly after commencing the work, the authors discovered that a beginning had already viduals or -types w h i c h will been made in attacking this prove effective i n h i b i t o r s very problem, 1 ) . against rusting in salt soluPREPARATION OF SITROtions. A systematic search GEN FREEFROM O X Y G E N . for such compounds is not, The first r e q u i s i t e was a however, as easy a task as supply of nitrogen freed from the search f o r i n h i b i t o r s every trace of oxygen. This against acid attack where the oxygen-free nitrogen served r a t e of hydrogen evolution a triple purpose: (1) as a serves as a simple yardstick. means of transferring water R h o d e s and K u h n (16) and solutions from one vessel have s h o w n that c e r t a i n to another, (2) as a stirring heterocyclic compounds conand mixing agent, and (3) as taining a nitrogen atom in the a means by which water and ring are effective inhibitors in solutions absolutely free from acid pickling. Apparently a oxygen could be prepared. layer of large organic moleTo remove the oxygen (0.6 cules is deposited upon or adto 0.7 per cent), the nitrogen sorbed by the iron surface from the cylinder supply was and forms a blanketing layer passed through copper gauze which raises the e l e c t r i c a l resistance a t the i n t e r f a c e FIGURE1. APPARATUS FOR PnODUCING OXYGEV-FREE h e a t e d to redness in a silica tube one inch (2.5 cm.) NITROGEN AND WATER of m e t a l a n d s o l u t i o n . 1116

October, 1933

INDUSTRIAL

A S D

ENGINEERING

CHEMISTRY

1117

(Figure 2). removed from the electrode chamber through The apparatus was then thoroughly flushed out n-ith oxygen-free nichrome wire. After leaving this small furnace, t h e nitrogen nitrogen. One hundred cubic centimeters of the 0.1 N sodium was passed through a glass tube containing alkaline (potas- chloride were forced by nitrogen pressure into the mixing vessel sium hydroxide) pyrogallol, prepared as recommended by and thence into the elect'rode chamber. The Armco wire electrode was washed with alcohol followed Hempel to avoid liberation Of carbon monoxide. This tube by water. I t was then laid on a piece of cardboard and polished was given a genble upward inclination, and to its upper end by rubbing vigorously in one direction with alundum cement was attached a n inverted gas-washing bott,le t o t r a p any picked up on a piece of cotton batting moistened with distilled water. The same p o l i s h i n g spray that might possibly be agent w a s u s e d throughout, formed. This part of t h e since iron polished with alunapparatus assembly is shown dum is cathodic towards iron polished with emery and both in Figure 1. are anodic towards unpolished iron (26). After polishing, the APPAR.4TUS FOR DETERMISe l e c t r o d e was washed with ISG THE POTENTIAL OF IROK d i s t i l l e d Lyater, placed in a AGAISST BRINE. The electrode clamp, and exposed to the air vessel and calomel electrode for 5 minutes ( + 5 seconds). were placed in a constant-temI t was then introduced into the perature bath in Tvhich the temelectrode chamber (care being p e r a t u r e was maintained a t taken to keep the e l e c t r o d e 20" =t0.1" C. The detnils of from touching the glass during this a s s e m b l y are s h o m in insertion) and kept immersed Figure 2. in the 0.1 AT sodium chloride A Leeds & Northrup type K for one hour, the solution being potentiometer and box galvaagitated continuously with ninometer F 4469 were employed trogen during the entire period. to measure the potential difThis t r e a t m e n t brought the ference between the iron elecelectrode to a standard condi. potassium trode and the 0.1 V tion, easily reproduced. chloride-calomel cell. At the end of the hour, the The m e r c u r y used in the e 1e c t r o d e was again exposed calomel cell was purified careto the air for 5 minutes and fully. First, air w a s b l o w n then, w i t h o u t any f u r t h e r through the mercury for some time. I t was then allowed to polishing, reimmersed in the FIGURE 2. ELECTRODE \-ESSEL A S D C.4LO>IEL salt solution. The potentiomef a l l in a s t , r e a m of m i n u t e ELECTRODE ASSEMBLY t e r i T a s read every minute droplets through a 5-foot (1.5for 5 minutes and then every meter) column of 1:1nitric acid, and, finally, it was distilled under diminished pressure three times. 5 minutes for 50 minutes, the salt solution being agitated conThe calomel and potassium chloride were Kahlbaum's analytical tinuously, as before, by a stream of nitrogen bubbles. reagents. The electrode was the usual type, platinum connected This set of readings constituted the reference run, and the to a copper lead by means of mercury. potential-time curve plotted from these readings constituted the The connections throughout were glass seals where possible; reference curve (Figure 3). elsewhere glass was held to glass by pure gum-ruhber tubing Akt the end of the reference run the electrode was washed coated on the outside with collodion. The rubber stoppers in nith tyater and again exposed to the air for 5 minutes. During the three vessels shown in Figure 2 were rendered gas-tight by this interval the electrode chamber was emptied through S sealing with soft de Khotinsky cement and by covering the latter (Figure 2) by ivater suction and thoroughly flushed out with with beeswax. The stirrer was run by a water motor. TKO nitrogen. thermometers were kept in the bath. The agar-agar jelly in the A charge consisting of a kno1Yl-n weight of inhibitor dissolved bridge was made by dissolving 3 grams of agar in 100 CC. of hot in alcohol, mixed Tvith sufficient 0.1 N sodium chloride to make distilled water and then stirring in 40 grams of potassium chlo- 100 cc, of the mixture, having been previously prepared and ride. The mixture, poured into the bridge while hot, set to a stored in the mixing vessel, was run into the electrode stiff jelly. The ends of the agar-potassium chloride bridge were chamber. The electrode was reimmersed in this solution and drawn down somewhat to prevent shifting of the jelly by gas connected t o the potentiometer, and readings were taken as Pressure. The electrode chamber and mixing vessel had a cabefore for one hour, the solution being continuously stirred by pacity of 250 cc. each. nitrogen. This constituted the run with inhibitor present The iron electrodes used were made from 6-inch (15.2-cm.) (pigure 3). lengths of Armco wire 0.009 inch (0.229 mm.) in diameter. After this run, the mixing vessel and electrode chamber were The wire was sealed into soft' glass tubing, 4 inches (10.2 cm.) nashed out iyith alcohol, follolved by distilled I,.ater to of the wire being allowed to protrude into the electrode vessel. any inhibitor remaining before another run , ~ made. ~ ~ s The electrical connection was made in the usual way with mercury into which dipped the copper lead from the potentiometer. EXPERIMENTAL DIFFICULTIES.At the beginning of the P R E P A R A T I O S O F OXYGEK-FREE W A T E R . The apparatus investigation great difficulty \vas experienced in obtaining is shown in Figure 1. T h e 12-liter, short ring-neck flask was results Tvhich checked iyith each other Tyithin the reasonable half filled with distilled mater, a handful of glass beads added, limits of experimental error. In every case the trouble was and t h e air flushed out with nitrogen from both the 12- and traced t o the presence of oxygen, When even a trace of 5-liter flasks. The water was heated, suction applied t o t h e oxygen was present,, the potential-time CurTTeS were irregular receiver by the usual x a t e r pump, and distillation begun. in shape and gave too high a value for the limiting or equilibT h e first liter of condensate was rejected. When about 5 rium potential. It R-as found that, in order to obtain relit'ers had been collected, nitrogen was run in until a Positive producible results, the oxygen content of the nitrogen had pressure was indicated, and t h e taps were then closed. t o be kept below t h a t determinable as copper oxide, upon PREPARATIOK OF STANDARD 0.1 A' SODIUMCHLORIDE. passing the gas over heated copper gauze. If the electrode Kahlbaum's C . P. sodium chloride (5.846 grams) was added was kept immersed t h a n 1.25 hours during the pret o the liter flask, through which a current of nitrogen was liminary treatment, a slightly higher result was obtained. floTving (Figure 2), then some OxYgen-free water, and, after During these experiments great care was taken to keep the solution, oxygen-free water t o t h e mark. T h e solution was p~ of all the solutions at 7 ( * ()I), the adjustment being thoroughly mixed b y a current of nitrogen. made with 0.01 N sodium hydroxide or hydrogen chloride, STASDARD PROCEDURE. The mixing vessel and electrode and t h e pH checked against standard color tubes or, in some chamber were rinsed out with oxygen-free R-ater which was cases, even more precisely by the quinhydrone electrode.

in outside diameter

18 inches (45.7 cm.) long wound lvith

INDUSTRIAL AND ENGINEERING CHEMISTRY

lll&

Arfew experiments were made with solutions which were not neutral to show the effect of the p H of the solution upon the electrode potential of iron. Figure 4 shows that a low p H results in a high curve-that is, with only a small decrease of potential from that of the reference curve. On the other,hand, a high p H results in a low curve. This is in agreement with the known fact that the rate of corrosion of iron is less in alkaline solutions.

75 0 720

710 700 6 90 $680 s 670 660

.f

2 dsa 640

? 630

26201

o

I

5

I

io

I

15

I

2a

I

2s

I

I

I

20 35 40

I

4s

I

I

Vol. 25, N o . 10

alcohol (freed from aldehyde by silver nitrate and redistilled) to sufficient salt solution to produce one liter of 0.1 N sodium chloride solution. THEORETICAL DISCUSSION.It has generally been assumed that the potential shown by a bimetallic specimen against a calomel cell is that of the most reactive metal present. This is true only if, in the short-circuited cell produced, the cathodic greatly exceeds the anodic polarization, and the cell resistance is high. Evans (4) has shown that with coppercoated iron a compromise potential is obtained which depends upon the relative areas of the two metals. Applying this to the case of an electrode surface of iron and iron oxide, he concludes that the potential of a metal, covered with an oxide film, will in general lie intermediate between the values given by uncoated iron and solid oxide. The more porous the oxide film, the nearer does the potential approach that of iron. He says, furthermore, ‘‘a final more positive potential indicates immunity from rusting; a middle value, slight rusting; a more negative value, a profuse rusting.”

5a 55

7-iPIZ.rnnutrs

FIGURE3

I n order to obtain results which may be taken as comparable, it is essential that the solutions to which the results refer must have their hydrogen-ion concentrations controlled within very narrow limits (7,10,I S , 1 7 , 2 4 ) . By thus controlling all the above factors, reference curves which checked within 5 millivolts could be obtained a t any time. The concentration of the sodium chloride solution, of course, had a marked effect on the potentials obtained, a more concentrated solution giving a larger potential-that is, a potential more negative than the calomel cell and more negative than the usual reference potential. SERIESI, EXPERIMENTS WITH ACRIDINEAS INHIBITOR I n developing the apparatus and standard procedure, acridine was used as inhibitor. This heterocyclic compound is known to be an effective inhibitor in acid pickling solutions (16).

A series of runs was made with the nominal aim of finding, if possible, the optimum concentration for this particular inhibitor. The maximum concentration possible t o employ was 7 millimoles per liter. 1

1

0 5

1

1

1

1

1

1

1

1

1

l

IO

/5 20 25 30 35 40 45 50 55 TIMf-minutes

FIGURE 4

The acridine was introduced as an alcoholic solution, and in the runs with the higher concentrations it was feared that the introduction of such relatively large quantities of alcohol to the electrode chamber solution would have an effect which would, of course, be absent from the reference runs. I n order to obviate such errors, in runs 46 to 52, inclusive, the salt solution was made up by adding 150 cc. of 95 per cent

FIGURE 5

The potential of iron against brine is more negative than that of the film of iron oxide against brine. If we assume from the work of Evans that the potential of the iron electrode a t any time is the resultant of the individual potentials of iron and iron oxide as determined by their relative areas, then it follows that a potential-time curve will indicate how the areas of iron and iron oxide are changing relative t o one another and hence will give some indication of the liability of the metal to corrode (not the rate of corrosion). Thus we shall secure information with respect to the protective effect of the skin under known conditions. The potential-time curves obtained in Series I are of an exponential form, and empirical equations may be built up and evaluated for any particular run as shown in Figure 5. Let us assume two cells acting in parallel and opposed to each other to represent an iron electrode coated with oxide immersed in sodium chloride solution and connected to a calomel half-cell, through the electrolyte internally, and through a potentiometer externally Call the e. m. f. of the cell (oxideisodium chloride solutionlpotassium chloride solution) calomel electrode) El and its resistance R1. Call the e. m. f. of the cell (ironlsodium chloride solutionlpotassium chloride solution~calomelelectrode) ES and its resistance RQ. Then if i is current in outer circuit: ROi

+ Rii - Ei + Ez

=

The potential difference between Q and W

0

\

October, 1933

:.

INDUSTRIAL AND ENGINEERING CHEMISTRY

In(L - E')

P (as shown on potentiometer, Figure 5 )

1119

= a -

Kit

Upon plotting log ( L - P ) against time, straight lines are obtained (Figure 6). Sow, as the relative areas of oxide and iron change with It was observed] however, that the log (L - P ) vs. time time, the resistances R1and R2 change. The resistance, RI, graph sometimes consists of two straight lines, which preof the oxide increases as the oxide disappears] while that of sumably result from two reactions, one of which largely the iron decreases as more iron is uncovered. At the start determines the slope of the first straight line; the other, the R? > R,, and at this stage R1 may be taken as negligible. slope of the second. Therefore P = El (approximately). At the finish Rl > R2, Of the two reactions one possibly arises from the passing into solution of metallic iron beneath the oxide skin as a result of chlorine ions passing through the interstices in the iron oxide film, the other possibly from the removal of the oxide film itself. One of these reactions, probably the solution of the iron, was almost negligible and in most of the runs with acridine could be regarded as zero. I n the particular cases in which this reaction did not approach zero, an equation of the form L - P = a be-KJt r e - K zt describes the curves very well. It would facilitate further investigation t o develop a constant which would be a measure of the inhibiting powers of a compound. Such a constant was evaluated as follows: Values of log ( L - P ) were plotted against time for the reference run. The slope of this straight line was measured and denoted by K . This value is a measure of the rate a t which the protective film is destroyed-that is, the rate a t which the iron loses its passivity. For the run with inhibitor O J 2 3 4 5 6 7 8 9 / 0 minutes present, the slope of the log ( L - P ) vs. time straight line was denoted by K1. The difference between K and K1FIGURE6 that is, the difference in the rates a t which the protective and a t this stage Rzmay be taken as negligible. Therefore effect of the two films changed-is obviously a constant and is a convenient measure of the value of the compound as an P = E? (approximately). inhibitor. This constant is denoted by N (Figure 7). l(specific resistance) d(thickne-) Reiistnnce = A (area)

+

+

Let us assume ZI approximates Z2, and dl approximates d2. This assumption may appear t o have little justification, but Evans' results ( 7 ) for copper-coated iron indicate that, for a case in which the cell resistance is high, this assumption is justified in an endeavor to account for the results obtained.

26 78

o

L 0G.C ONCENTRATION

a2 0.4 a6

aa

IU

0.

0. 0.

Then

0.4 h

2

0

s

20 0.3

0

R2 -=.-=

R1 Then

=

+ R2

Ai ,41 4-A:

(area af oxide) total area

~ o x l d e+

02

totaiarea

CONCLN TRATlONrnriYmols/iitrrc

FIGURE 7

-) E,,,,

hasuming that the decomposition of the oxide is a reaction of the first order, it is proportional to the percentage of total area that is remaining as oxide. Then = Be-K1' A (total area) :. B = A .: A1 = .4ebKit A2 = A ( l - e-K1t) Similorly :. P = ( e - K 1 t ) ~ l (1 - e-K1t)E2 = E, -- e - X 1 t (E2 - E l ) :. EB- P = e-K1t (Ez- El) L - p = ae-Klt 0 1' where L = limiting potential in the negative direction-that most negative potential of the reference run a = E? - El, a constant

when

2

area of oxide total area area of iron exposed

(

?

The inhibiting effect is expressed by the equation L - P = e f N t and multiplies the rate of oxide destruction, L - P - e-Kt Thus the potential-time curve with inhibitor present may be represented by L - P = e!N-K)t. This equation was found to fit the results obtained very well. The results of Series I are shown in Table I. I

t = 0, A , =

TABLEI. DATAON EXPERIMENTS WITH ACRIDINE RUN 43,44 32,33 122.123

+

is, the 4

CONCN.

MilZimoles/liter 0.5 1.0

3128 117,115 4O,4la 29,30,42 47,50,51.52 48,49 Saturation point without alcohol

1.5

2.0 2.5 3.0 4.0 5.0 6.0 7.0

N (Av.) 0.0804 0.1000 0.1190 0.1215 0.1270 0.1300 0,1400 0.1480 0.1566 0.1634

IKDUSTRIAL AND ENGINEERING CHEMISTRY

1120

Vol. 25, No. 10

TABLE11. CL.4SsIFICATION OF COMPOUNDS EXCELLENT Acridine iodoethane Acridine iodomethane Acridine iodothane: Sodium salicylate Naphthoquinone sodium salicylate Naphthoqulnone Acridine iodoethane: Sodium cinnamate Naphthoquinone Sodium salt of eosin

+ + +

+

+

FAIR

GoOD Acridine Quinoline iodoethane Tannin Hydrorinnamic acid Sodium salicylate Sodium cinpamate Naphthoquinone Oxime of Michler’s ketone Sodium salt of fluorescein Nicotine sulfate Alizarin Sky Blue

INERT

POOR

8-Naphthoquinoline Benzoquinone Bcet aldehyde Benzaldehyde Quinaldine 8-H ydroxyquinoline salioylic ,acid 8-Hydroxyqujnoljne 8-Hydroxyquinoline gallic acid citric acid 8-Hvdroxyquinoline cinnamic acid 8-Hidroxyquinoline Quinine hydrochloride Cinchonine hydrochloride Nicotine

Benzophenone o-Toluidine Aniline Benzidine @-Naphthylamine Phenylhydrazine Ethylene glycol Glycine Resorcinol Pyrogallol Pyridine Quinoline Anthracene Brown Brilliant PurDurine Caffeine

+ + ++

Acetone Acetophenone Allyl alcohol Glycerol Anisole Anethole Potassium x a n t tlate Grape sugar Soluble starch Phenol a-Aminophenol E t h y l alcohol Hydroquinone Phenol~hthalein

R

11, EXPERIMENTS WITH ORG.4NIC C O M P O U S D s OTHER THAN ACRIDINE The success with acridine encouraged a systematic search for more efficient inhibitors. In the course of this search many compounds were tried; some were allied to, and others were very different from acridine in their composition and structure. I n Table I1 the compounds investigated are classed according to the effect of small additions upon the potential of the iron electrode against sodium chloride solution. These classes correspond to the following average differences between the observed and reference potentials a t the end of one hour:

The potential-time curves for the sodium salts of eosin and fluorescein also showed reversed initial slopes, but the potential finally rose again (Figures 4 and 8). This effect was probably due, not to the fast rate of adsorption, but to the fact that the sodium salts of eosin and fluorescein were being adsorbed by the iron oxide instead of by the iron surface (12). Freundlich calls this “retrogressive adsorption” ( 8 ) . The similarity of the fluorescein and eosin curves no doubt arises from the resemblance in composition and constitution of the compounds. The greater effectiveness of eosin is doubtless entirely due to its higher molar weight. The combined use of three of the most effective inhibitors so far found-vis., acridine iodoethane, naphthoquinone, and sodium cinnamate (as shown in Figure 8)-gave a lowering of potential of 299 millivolts, which, even a t the end of Mzllzvolls Millisolts .~ Excellent 200-370 Poor 10-50 335 minutes, showed no signs of decreasing. A mixture of Good 100-200 Inert 0 naphthoquinone and sodium salicylate, each present in the Fair 50-100 concentration found to give approximately a maximum PROPERTIES F.4VORABLE TO INHIRITION O F RUSTING. effect, gave an even greater lowering of potential (370 milli1. In the case of nonelectrolytes,both high molecular weight volts). In both the above examples it is possible that proand sparing solubility in water are essential characteridics of portions other than those selected will be found still more good inhibitors. 2. Certain configurations and groupings are particularly effective. These researches are being continued in an endeavor to effective Drovided the compound falls under 1. Among these are heterocyclic compounds-withnitrogen in the ring as already discover even more effective combinations of inhibitors. observed by Speller and Chappell (21) and by Rhodes and Kuhn (16) in the acid pickling of iron and steel. Such compounds are especially effective when certain side chains are directly attached SERIES 111. h h C H A X I S M OF ACTIONO F ISHIBITORS t o the nitrogen in the ring. Quinone structure is of less importance than heterocyclic structure, but the combination of In this series of experiments a quantitative study of the the two in the same compound results in effective inhibition. relations between the inhibiting effects of certain similar 3. In the case of electrolytes, the order of effectiveness is practically that of the lyotropic series for the anions; e. g., compounds-namely, acridine, quinoline iodoethane, quinaldine, and P-naphthoquinoline-was made. The results sodium chromate is more effective than sodium salicylate. 4. Combinations of inhibitors do not exert addition effects. of this series are given in Tables I, 111, and IV, and Figure 7 . SERIES

T 4BLE 111. INHIBITING EFFECTSOF CERTAIN COMPOUNDS CONCN. N (.4v.)

RCX

RUN

Mzllzmoles/liler QUINOLINE IODOETAANE

128 125 135-136 126

0.5 1.0 2.0 5.0

153 155 149-151

0.50 0.75

0,065

160-164 156-157

0.50 1.00

0.090 0,109

1.00

CONCN. N (Av.) Millimoles/liler

138-139 141 131-132

0.129 B-NAPHTHOQCIXOLINE 0.0755 152 146 0.0880 0.0940



10.0

7.0

0.135 0,139 0.150

1.50 2.00

0.1050 0.1100

2.00 2.50

0.076

8.0

QGINALDINE

TI ME-minutes

FIGURE8 POTENTIAL-TIMECURVES OBTAINED FOR THE h f o s T EFFECTIVE INHIBITORS. The initial slope of the potentialtime curve for naphthoquinone was of opposite sign to the corresponding slope of the reference curve (Figure 8). This initial slope may be explained by supposing the naphthoquinone to be adsorbed on the iron surface more rapidly than the initially present iron oxide film is destroyed.

TABLEI\‘.

0.033 0.053

158-159 162-163

0.081

COMPARISON OF EFFECTS OF VARIOUSCOMPOUNDS

C oM P ou N D Acridine Quinoline iodoethane

5 -3.37 -3.12

K 34.2 35.0

a COMPOUND 8-naphthoquinoline -3.20 -1.85 Quinaldlne

K 34.2 34.3

Upon plotting the values of N against the concentrations of the inhibitors for these runs, the curves are found to be all of the same general shape. The general relation between the concentration and N is C = aeKN,where C = concentration, a and K are constants, and N has been defined earlier.

I N D U STR IA L A N D E N G IN E E R IN G C H E M I ST R Y

October, 1933

Table IV shows that, for the four compounds given, K is the same. Constant N may belinked up with the properties of the compound to which it refers as follows: Since c

=

,,eKN

.:

1nC

=

a

+KN

Butler (Z), using the results obtained by Gouy (11) in his experiments on electrocapillarity, calculated the following expression for the change of work of adsorption a t a metalliquid interface due to change in the potential across the interface: W = 1/z(A - B)EZdVB ( P A - P B ) E ~ L ’ B where W = change of work of adsorption Subscript A refers t o metal salt Subscript B refers t o adsorbing molecule E = field strength = constant X voltage A , B = moments of polarization due to field E due = volume of molecule B P A , P B= extra moments of polarization due to a polar grouping in molecule B

+

This change of work, W , may be considered as causing a dilution of the solution with respect to molecule B.

... co= c e ( M V Zf R

R

where C = final concentration CO = initial concentration V = potential across interface M = -~ ( A - B)E’~VB; = ( P A - P E ) E ~ B kt

c + M V ~+ RV

.:

In c0 = But In CO = a

.:

N =

kt

+ + MV2 + R V )

KM (C - a

K

We see from this that N depends upon (1) the molecular volume of the adsorbing molecule (dug), (2) polar or nonpolar characteristics of the adsorbing molecule, (3) initial potential across the metal-liquid interface (I.), and (4) a, an experimental constant, which will probably equal C. Since K was approximately the same for the four compounds studied, it apparently depends upon some common property of the four compounds, whereas a depends upon some property which is specific for each compound. It seems reasonable to assume that any group of compounds, the members of which are of the same type, would give a series of parallel exponential curves. Throughout any one such series K would be the same. SERIES

Iv, MECHANIShl OF ORGANIC COhIPOUh-DS DIFFERS

THATOF CHROMATES It has been shown by Evans (7) and others that the inhibiting effect of potassium chromate on the corrosion of iron in brines is due to the formation of an oxide film on the iron. The inhibiting effect of the organic inhibitors discussed in this article has been found to be due to adsorption of the inhibitor on the electrode surface. In order to obtain additional evidence in support of the opinion that the mechanism of inhibition in the case of organic inhibitors differs from that of alkali chromates, the variation of inhibiting effect was investigated-that is, of N with varying concentrations of potassium chromate. As will be seen from Figure 7 , the relation between N and C is a straight line for potassium chromate and consists of two straight lines intersecting a t a concentration of 2.0 millimoles per liter. On the other hand, the relation between N and C for typical organic inhibitors has already been shown in this article to be exponential. FROM

SERIESV, EFFECTOF STIRRING When an organic inhibitor is present, the iron becomes coated with the organic compound. The extent to which this will happen is, as noted in Series 111,dependent upon the

1121

concentration of the compound in the solution and more particularly in the solution immediately surrounding the electrode. Inhibition, therefore, should be dependent upon the rate a t which the dissolved compound diffuses towards the vicinity of the electrode. A number of tests were run, in some of which the solutions were stirred by nitrogen as in the standard procedure outlined previously. In the remainder the solutions were not stirred. The results of these tests confirmed the assumption made regarding the effect of stirring; that is, the tests which were not stirred gave lower values for N for approximately the first 5 minutes, but beyond 5 minutes the value of 11‘ for the stirred and unstirred tests was the same.

SUMMARY OF RESULTS AND CONCLUSIONS A dependable procedure has been developed by which (1)

reproducible measurements of the potential of an iron electrode in a brine can be obtained. (2) This procedure has been employed in making a general survey of the inhibiting tendency on the corrosion of iron in brine of different classes of organic compounds. (3) On this basis certain classes of organic compounds, either singly or in combination, have been shown to have a strong tendency to inhibit the corrosion of iron in brine. (4) An attempt has been made to elucidate the mechanism of the inhibiting action of these organic inhibitors and to use this information in discovering more effective inhibitors. ( 5 ) The difference between the mechanism of inhibition as exercised by these organic inhibitors and that of alkali chromates has been clearly brought out. (6) Further work upon the application of organic inhibitors to engineering and industrial problems is now being carried on by the authors; this work embodies the principles here set forth. ACRNOWLEDGhIENT

A spool of Armco wire used in the experimental work was kindly furnished by the manufacturers.

LITERATLTRE CITED (1) Bastow and McAulay, J.C u m . SOC.,1929, 85. (2) Butler, Proc. R o y . Soc (London), A112, 129 (1926); 113, 594 (1927); 122, 399 (1929). (3) Chappell, Roetheli, and McCarthy, IKD.ENQ. CHEM., 20, 582 (1928). (4) Evans, U. R., C a n . Chem. Met., 16, 296 (1931). ( 5 ) Evans, U. R., “Corrosion of Metals,” Arnold, 1924. (6) Evans, U. R., J . Chem. SOC.,1927, 102C-40. (7) Ibid., 1930, 1361-74. (8) Freundlich, “New Conceptions in Colloidal Chemistry,” p. 118, Methuen, London, 1926. (9) Friend, J. N., “Corrosion of Iron and Steel,” Longmans, 1911. (10) Fujihari, T r a n s . Am. Electrochem. SOC.,49, 327 (1926). (11) Gouy, Ann. chim. phys., [7] 29, 146 (1903); [SI 8, 291 (1906); [8] 9, 75 (1906). (12) Haber, “Beitrage z. Kenntnisse d. Flockung u. Adsorption Hydrophobic Sols,” 1925. (13) Kohman and Sanborn, IKD. EN. CHEM.,20, 1373 (1928). (14) Pollitt, A. A., “Causes and Prevention of Corrosion,” Benn, 1923. (15) Rawdon and Tucker, Bur. Standards J . Research, 3, 375 (1929). (16) Rhodes and Kuhn, 1x0. EKQ.CHEM.,21, 1066 (1929). (17) Shipley and McHaffie, Can. Chem. Met., 8, 121 (1924). (18) Speller, F. N., “Corrosion, Causes and Prevention,” McGrawHill, 1926. (19) Ibid., p. 279. (20) Ibid., p. 360. (21) Speller and Chappell, Chem. &. M e t . Eng., 34, 421 (1927). (22) Tammann and Sotter, 2. anorg. allgem. Chem., 127,257 (1923). (23) Warner, J. C., T r a n s . Am. Electrochem. Soc., 55, 287 (1929). (24) Wilson, J. A , , IKD. ENU.CHEM.,15, 127 (1923). (25) Wullen-Scholten, W. van, Korrosion u. Metallschutz, 4, 265-72 (1928). RECEIVED April 4 , 1933.