I
H. W. McCUNE Miami Valley Laboratories, Procter & Gamble Co., Cincinnati 31, Ohio
Corrosion of Aluminum by Alkaline Sequestering Solutions A description of the corrosion processes is the foundation for building an understanding of inhibitor
S
I OLUTIONS of most alkalies and alkaline salts corrode aluminum because of the amphoteric character of the hydrous oxide film and the high activity of the metal; solutions of silicates and ammonia are often exceptions (4, 5, 77). Corrosion uses hydroxide ions, leading to a decrease in pH, and evolves hydrogen, as would be expected from the reaction: A1 OHHzO = AlOz1.5 H2. Few data have been published concerning corrosion of aluminum by alkaline solutions of sequestering agents; but solutions of sodium citrate, tartrate, and salicylate corrode aluminum with a resulting increase in pH, while the nonsequestering salts, sodium acetate, succinate, and benzoate are essentially noncorrosive (6). Corrosion by sodium hydroxide-sodium citrate solutions causes an increase in pH if the initial pH is below about 10.9 and a decrease in pH if the initial value is higher (7). Sodium triphosphate solutions have been shown to be more corrosive toward aluminum than sodium orthophosphate solution, though both are adjusted initially to the same pH of 7.5 or 9.5 (74). This article compares corrosion of aluminum by alkaline and alkaline sequestering solutions, and describes corrosion by solutions of sodium triphosphate and (ethylenedinitrilo) tetraacetate. Published studies of corrosion by solutions of sequestering agents have included those of lead pipes by polyphosphate-treated water (72) ; and copper, zinc, and brass by polyphosphates that might be used in mechanical dishwashing products ( 3 ) . Heavy-duty detergent products built with condensed phosphates contain sodium silicates, added to prevent such corrosion (8). The temperatures, concentrations of sequestrant, and area of metal to volume of solution ratios discussed are those existing in some of the applications of built detergents. The specific conditions under which the experiments were carried out are given in the captions of the figures.
+
+
+
action and finding
Experimental Materials. The aluminum used was 3 s (nominally 1.270 manganese; new designation 3003) and was cut into specimens 7.62 X 1.90 X 0.104 cm. After being degreased, specimens were cleaned with fine steel wool, followed by wet pumice; then they were scrubbed under running water and rinsed in distilled water. This was followed by alcohol and air drying. Corroded specimens were dipped into concentrated nitric acid (2) and washed and dried as before. Distilled water and reagent grade chemicals were used. Purified sodium triphosphate, sodium trimetaphosphate, and phosphorus-32-tagged sodium triphosphate were prepared as described previously (13). Tetramethylammonium, triethanolammonium, and ethanolammonium triphosphates were prepared' by neutralizing triphosphoric acid, prepared by ion exchange, with the proper bases. Corrosion Tests. Weight loss data were determined with an analytical balance on specimens exposed in stagnant total immersion tests. The weight loss agreed within experimental error, with dissolved aluminum, determined by wet chemical analysis of the solutions. The ratio of area of metal to volume of solution was 145 sq. cm. per liter, except where the effect of volume was being investigated. Aeration was shown not to affect the results appreciably. Test solutions in borosilicate glass flasks were open to air but with water loss limited. The extent of corrosion was no greater when polyethylene containers were used; thus silicate from the glass did not have an inhibitory effect. Temperatures were controlled to rt0.2' C. by a water bath. pH measurements were made with a Beckman pH meter, Model G, using glass and saturated calomel electrodes. The values reported are for room temperature unless, otherwise indicated. Potential measurements were also made with the pH meter and sometimes with
new
inhibitors
a Clippard electronic voltohmmeter Model 405. Complexing of Aluminum by Triphosphate. The reaction was studied by titrating 200 ml. of each of the following solutions with 0.204M sodium hydroxide, using a Precision-Dow Recordomatic Titrometer equipped with glass and calomel electrodes : A. 0.0200M perchloric acid B. 0.0200M perchloric acid aluminum perchlorate C. 0.0200M perchloric acid sodium triphosphate D. 0.0200M perchloric acid aluminum oerchlorate sodium trip'hosphate
+ 0.00167M + 0.00500M + 0.00167M
+ 0.00500M
Curves resulting from titrations A, B, and C were added to obtain a curve for no interaction. The difference between this curve and D was then taken to find the amount of hydroxide liberated or consumed by the aluminum-triphosphate interaction. The ratio of aluminum perchlorate to sodium triphosphate in D was varied so as to give additional mole ratios of 0.5, 1.0, 1.5, 1.7, and 2.0 at a constant total molarity of 0.00667.
ResuI ts Effect of pH on Corrosion by Alkaline and Alkaline Sequestering Solutions. Solutions of sequestering agents [sodium (ethylenedinitrilo)tetraacetate, pyrophosphate, triphosphate] corroded aluminum much more rapidly at 60 C. than solutions of nonsequestering salts (sodium trimetaphosphate, sodium sulfate, tetramethylammonium sulfate) at alkaline pH values up to about 10.5 (Figure 1). Above this pH, alkaline corrosion overwhelms sequestering corrosion. These data were obtained at constant pH maintained by addition of sulfuric acid or sodium hydroxide to the sequestering solutions and sodium hydroxide to the nonsequestering solutions. A decrease in pH requiring addition of alkali in these experiments is VOL. 50, NO. 1
JANUARY 1958
67
typical of alkaline corrosion, and the increase of p H is typical of sequestering corrosion, for most conditions, which will be elaborated later. Tetramethylammonium hydroxide was used with tetramethylammonium sulfate, but the absence of sodium ions did not significantly change the amount of corrosion in these particular solutions. A similar picture was presented by potentials attained by cells consisting of an aluminum strip and a saturated calomel electrode dipping into test solutions which were held at constant p H and 30' C. (Figure 2). The European sign convention, commonly used in corrosion studies, was employed. The potentials given are the essentially steady-state values that were usually obtained after 1 to 2 hours. Potentials from sodium sulfate solutions at p H 7 to 9.5 did not become steady in 6 hours and were read after 3 hours when the rate of change was slight. The potentials for aluminum immersed in sodium triphosphate solutions became much less noble between pH 8.0 and 8.5, a little lower than the p H of rapidly increasing weight loss (especially in view of the temperature difference), Potentials for aluminum immersed in sodium sulfate solutions remained more noble than the others u p to a pH of 10.5 where the potential became approximately equal to those from the sodium triphosphate and (ethylenedinitri1o)tetraacetate solutions, and where the weight loss increased rapidly. Variation of p H from 7 to 10.5 did not change the potential obtained from the (ethylenedinitrilo) tetraacetate solutions. Although potential measurements cannot be said a priori to indicate tendency for corrosion, they do in this case appear to indicate correctly anodic depolarization and tendency for corrosion.
Corrosion by Sodium Triphosphate. Corrosion was general over the surface of the metal and (for conditions like 60' C., 3 hours, and 200 ml. of 0.18% sodium triphosphate solution) resulted in etching-a decrease of specular reflectivity of the metal with little or no discoloration. Lower sodium triphosphate concentrations, smaller volumes, and longer times caused a brown discoloration. Pitting was not observed at the p H of sodium triphosphate solutions. The effects of volume, concentration of triphosphate, and temperature on the amount of corrosion taking place in 3 hours are shown in Figures 3, 4, and 5; corrosion and p H as functions of time are shown in Figure 6. Larger volumes (Figure 3) provide more triphosphate, so that reaction neither decreases the triphosphate concentration nor increases the soluble corrosion product concentration so greatly. At 1 liter, the molar ratio of aluminum dissolved to triphosphate initially present is roughly one fifth, and the observation that volume is still having an effect is understandable. Agitation increases the amount of corrosion; therefore diffusion must also be contributing to the shape of the curve of Figure 3. At a volume of solution to area of metal ratio of 14.0 ml. per square centimeter, moving samples lost an average of 1.3 mg. per square centimeter in 3.0 hours while samples in the stagnant test lost 0.7 mg. per square centimeter. The moving specimens were held along their 1.90-cm. ends in a slotted plastic disk rotating at 94 r.p.m.; the surfaces of the specimens were perpendicular to their path, and the radius from the center of rotation to the center line of each sample was 4.1 cm. The extent of corrosion is approximately proportional to triphosphate concentration a t least u p to 0.3% (Figure 4). The presence of a detergent, an alkyl benzene sulfonate,
pH a t 60°C.
-1.50'
7jo
8D '
'
'
9.0
10.0 '
I
-L- -L' -
'0
260
pH at 3OoC.
Figure 1. Effect of pH on alkaline and sequestering corrosion
Figure 2.
200 MI., 0.18% solution, 1.0 hour, 60' C., pH maintained constant
200 MI., 0.01OM NaSSOa f 0.005M in salt indicated, 30' C., pH maintained constant
6%
at a constant fraction of the triphosphate concentration had a slight effect or no effect. The corrosion under discussion is much less important at room temperature than at hot water temperatures (Figure 5). The rates of corrosion by triphosphate solutions initially at the natural p H of 9.95 and adjusted to 8.0 are shown in Figure 6 ; the p H values (at room temperature) are also shown. The unadjusted p H quickly falls to about 9.7 and tends to remain about there. The adjusted pH quickly rises to about 9.3 and changes little thereafter. Under the conditions employed in this study depletion of triphosphate and the increase in concentration of soluble corrosion product can be an important factor-for example, a 20-mg. weight loss (in the experiments of Figure 6) makes the solution 0.0037M in soluble aluminum, but the initial triphosphate concentration was 0.004.9M. Reaction of Sodium Triphosphate. Stoichiometry of sequestering corrosion can be studied by adjusting the p H to below 9 where alkaline corrosion is insignificant. Molar ratios of alkali formed to aluminum dissolved are shown in Table I. They were calculated from data determined by titration with standard acid and by weight loss of the metal. The results ranged from 1.1 to 1.3 OH-/Al in those experiments where the p H did not go above 9. Measurements of the hydrogen evolved gave 1.5 moles per mole of aluminum dissolved, as would be expected from the equivalence of hydrogen and aluminum. The reaction of aluminum ions with triphosphate was studied by means of potentiometric titrations in order to explain the formation of alkali in the corrosion reaction. In the presence of sufficient sodium triphosphate, aluminum perchlorate-perchloric acid solutions form no precipitate as sodium hydroxide is added between pH 3 and 11.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Effect of pH on potential
Figure 3.
400 600 800 VOLUME, ML.
1030
Effect of volume
0.1 8 % NajP~010,3.0 hours, 60" C., initial pH 9.5, 29 sq. cm. AI
ALUMINUM CORROSION
0
9 was 1.1 to 1.9, and was somewhat higher than that liberated in corrosion, 1.1 to 1.3. Perhaps this alkali was used in alkaline corrosion. The curve in Figure 6 shows that the p H of an unadjusted sodium triphosphate solution starting at p H 9.95 drops very slightly during corrosion. In agreement, the deviation from the ideal in the titration data at p H 9.5 to 10 is slight (Table
11)*
CONCENTRATION W I ~ P ~ O%~ ~
(> = Alkyl Bz. Sulfonate Present, 33% of NaP ,O ,,
Figure 4. tration
Effect of
NabP3010
concen-
200 MI.,3.0 hours, 60' C., initial pH 9.5
Near the middle of the range the observed pH values are higher than those calculated from the composition of the solutions assuming that no complexing occurs. These observations, which hold a t least over the range 0.5 to 3 triphosphate per aluminum, indicate reaction between the aluminum and triphosphate species. A little hydroxide is consumed (Table 11) a t pH values above 9.5 to 10. The mole ratio of hydroxide liberated per aluminum in the p H range 7.5 to
The over-all reaction i s clearly metallic aluminum with triphosphate solution to give an aluminum-triphosphate complex, hydrogen, and in the p H range 7 to 10, hydroxide. Attempts to identify the aluminum-triphosphate complexes using the potentiometric titration data did not lead to clear-cut results and indicated only that several species were involved. The species of hydrogentriphosphate ions present could be calculated from published dissociation constants (78) except that their variation going from 25' to 60' C. is unknown. The equilibria existing in alkaline solutions of aluminum ions have not been well characterized (70). Without making any claims that these reactions are a correct representation of the detailed process and realizing that a t either pH there must be others involved, the following can be written as illustrative. Formulas written for the aluminum triphosphate complexes are consistent with the titration data but are not established as correct. At about pH 8,
TEMPERATURE,
Figure 5.
OC.
Effect of temperature
200 MI., 0.1 8 % NasPsOla solution, 3.0 hours initial pH 9.5
At about p H 10,
Al(OH),-.+
,
Pa010-6 =
.-
[Al(OH)aPaOio] + OH+ P3010-~+Al(OH),PaOio-6 3Hz0 = + 1.5 Hz -6
A1
Table I. Formation of Hydroxyl in Corrosion by Sodium Triphosphate (200 ml., 0.18% NarPsO~o,60' C.) .A1 DisTime, PH solved, OH-/ Hours Initial Final Mmole Al 1.0 2.0 4.0 4.8 3.0 3.0
5.0
8.13 8.13 8.13 8.13 8.00"
7.50 7.50
8.54 8.94 9.32 9.34 8.00" 8.62 9.20
0.148 0.222 0.475 0.538 0.312 0.255 0.452
1.10 1.25 1.38 1.38 1.12 1.20 1.27
a pH, measured at 60° C., maintained constant with 0.1N HzS04.
Table II. PH 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0
Hydroxide Liberated by Complexing OH- Formed/Al 1.0 1.1 1.5 1.9 1.8 0.06 -0.4 -0.2 -0.5
0.333 Mmole of A,l(ClOa)8 and 1,000 mmole of NabPsOlo in 200 ml.
Figure 6.
Effect of time
200 MI., 0.1 8% NasPsOlo solution, 60' C. VOL, 50, NO. 1
JANUARY 1958
69
MOLES ADDED CATiON /MOLE P300-’
Figure 7.
Effect of cation-P30lo ratio
200 MI., 0.1 8% NasPaOlo or [(CH&N]jP30m, 3.0 hours, 60’ C., initial pH 9.5
The reactions of aluminum hydroxide with both triphosphate and hydroxide are given for the p H 10 case to indicate that a t this p H both are probably important. Corrosion by an unbuffered sodium triphosphate solution starting in the p H range 7 to 9 is an autocatalytic reaction. I t is, however, self-limiting above about p H 9, where the net liberation of hydroxide changes to a net consumption. As would be expected from the increase in corrosion with p H and the production of hydroxide in sequestering corrosion below p H 9, buffering the p H does reduce the extent of sequestering corrosion.
Triethanolammonium and ethanolammonium triphosphate solutions are less corrosive (Table 111) than sodium and tetramethylammonium triphosphate solutions a t the same molar phosphate concentration and adjusted to an initial p H value of 8. Cations which are complexed by triphosphate reduce the extent of corrosion. I n Figure 7, adding sodium sulfate to tetramethylammonium triphosphate is shown to decrease corrosion slightly. Lithium, calcium and magnesium, and aluminum and nickel sulfates reduced the corrosiveness of sodium triphosphate, with effectiveness increasing in that order. The order of increasing effectiveness is the same as the order of increasing strength of complexing of these cations by a polyphosphate (17). Reaction of Sodium (Ethylenedinitril0)tetraacetate. The amount of hydroxide produced per aluminum dissolved by a sodium (ethylenedinitri1o)tetraacetate solution, determined in the same fashion as with sodium triphosphate, is shown in Table IV. From the dissociation constants of (ethylenedinitri1o)tetraacetic acid (75),the reported formula for the aluminum complex ion (76), and the assumption that these are not greatly changed from room temperature to 60’ C., the over-all corrosion reaction at about p H 8.5 can be written:
Table 111. Effect of Buffers (200 nil., 0.0049.71 triphosphate, 3.0 hours, 60’ C., initial pH 8.0) -4verage
Unbuffered
boss, Mg.
Weight
Final pH
NasPaOlo [ (CHa)rN 16P3010
9.0 12.0
9.2 9.3
1.9 1.8
8.0 8.2
Buffered (TEAm)6P3010 (MEAm)jPaOla
Table IV. Formation of OH- in Corrosion by Sodium (Ethy1enedinitrilo)tetraacetate (200 ml., 0.18% NarEDTA, 60’ C., initial pH 8.5) A1
Time, Final pH Min. 20 30 30 20 30 60
9.75 9.80 10.2 8.P 8.5” 8.5“
Dissolved, R‘lmole 0.218 0.419 0.404 0.272 0.326 0.530
OH-/Al Moles 0.93 0.93 0.94 1.04 0.93 1.26
a pH, measured at 60’ C., maintained constant nith 0 . l N HnSOh
70
INDUSTRIAL AND ENGINEERING CHEMISTRY
AI
+ 2Hz0 + HEDTA--- = [Al(OH)EDTA]-- + OH-
f 1.5 Hz
Sorption of Sequestrant. Aluminum immersed in an 0.18% phosphorus-32tagged sodium triphosphate solution for 3 hours at 60’ C. sorbed phosphate equivalent to 0.019 mg. sodium triphosphate per square centimeter, according to radioactive count. The usual rinse in concentrated nitric acid, to remove corrosion products, removed all of the radioactive phosphate. The weight of corrosion products was 0.05 mg. per square centimeter. This demonstration of the presence of the corrosive agent at the solid surface is pertinent to recent discussion ( 9 ) of chelating agents as corrosion inhibitors or as corrosion accelerators in cases where the surface complex of the chelating agent and metal ion is sufficiently stable to disrupt the metal or oxide lattice. I n the case of aluminum, reaction is apparently in the hydrous oxide surface film, for the radioactive triphosphate is removed by the nitric acid rinse which dissolves corrosion products but very little of the metal itself. An independent demonstration showed, b>- chemical analysis, 0.01 6 mg. of sodium triphosphate per square centimeter in a surface film isolated by dissolving the thin 2s foil on which it was formed with a bromine-methanol solution according to a published procedure ( I ) . The weight of the film was 0.04 mg. per
square centimeter. While part of the triphosphate may be reacting with soluble aluminum species it is also attacking the hydrous oxide film or aluminum ions which are diffusing outward through the film. Determination of the rate law for the reaclion of aluminum with alkaline triphosphate solutions should give valuable insight into the mechanism of this reaction. The surface film isolated from an aluminum foil exposed to 0.28YG of sodium (ethl lenedinitrilo) tetraacetate solution under the same conditions contained nitrogen and carbon indicating the presence of the sequestrant in the film. Anal) tical results were too variable to state how much. Much more surface film was formed in this case (0.7 mg. per square centimeter) than with sodium triphosphate. Acknowledgment The author wishes to thank W. E. Cooley for the titration data concerning aluminum triphosphate complexes. Literature Cited (1) Altenpohl, D., Aluminium 29, No. 9, 361 (1953). (2)