The adsorption isotherms for chloride on a corroding aluminium surface have been measured by using 36Cl as a radioactive tracer. The amount of chloride adsorbed, wCl, was a function of chloride concentration, [Cl], and time, t, according to ������������ �����������log WCl = 0.64(log[Cl]+log t)-7.8 where wCl is expressed as g cm-2, [Cl] as mol l-1 and t as minutes. Superimposed on the general curve was a fine structure which correlated with the stages of development of a corroding aluminium surface. The adsorption was localized to corroding pit sites. Addition of nitrate or sulphate delayed but did not prevent the uptake of chloride; the hydroxide ion was considered to have a similar effect. It was concluded that a corroding aluminium surface has a variety of adsorption sites with different adsorption properties; only a minority of these sites are potential or active sites for pitting corrosion.

The corrosion rate of aluminium in flowing neutral waters at 50�C has been determined as a function of pH, oxygen concentration and chloride concentration. The corrosion rate, At, as total aluminium lost between the 4th and 80th day was observed to be logarithmic according to At = B log t+C with a minimum rate in the pH range 5-6, and with B c. 3 x 10-5g cm2, C c. 20 x 10-5 g cm-2 and t in days for oxygen-saturated water. Saturating the water with nitrogen or adding up to 15 mg Cl- l-1 increased the corrosion rate. It was concluded that this was due to competitive action on the oxide surface, between dissolved oxygen and chloride ions.

Adsorption of Mn, Zn, Co, Ba and Ra from solutions in the range 10-5-10-10 mol-1 was measured in varying degrees of detail on α-alumina. It was found that (i) below a limiting concentration of about 1 × 10-1 moll-1, the logarithmic adsorption isotherm was linear whereas, above this limit, the logarithmic adsorption isotherm was curvilinear for Ba, Ra, Mn, and Co but linear for Zn with a slope of 0.51 ; (ii) below this limiting concentration, the pH at 50% adsorption was independent of solution concentration; and (iii) this limiting concentration was independent of the adsorbing metals studied. An adsorption model for the concentration-independent regime was developed based on adsorption at high energy, negatively charged adsorption sites by M(OH)n+ ions without accompanying proton release from either the surface or the adsorber.

Measurements are reported for the variation of the open-circuit potential, Er, of aluminium in oxygen-saturated sodium salt solutions. The value of Er was independent of SO42- and NO3- concentrations and similar to the value obtained for water (0.04 (s.h.e.)). Er was a function of chloride concentration given by ������������������� Er = -0.475-0.060log[Cl-] V (s.h.e.) at 25�C. There was a less well defined relationship between Er and NO2-, I- and Br-, and a complex relationship with F-. ��� The potentiodynamic characteristics are reported for aluminium in 1-0.01 mol l-1 Cl- oxygen-saturated solutions. Functional relationships were found for E0, Ep, Es and E0' with chloride activity at 5, 25, 50 and 75°C. Hysteresis effects are reported. ��� The experimental results are interpreted in terms of a thermodynamic equilibrium condition between the surface oxide and soluble aluminium chloride. As the system oscillates across the equilibrium conditions the surface will passivate or pit. A critical bulk solution chloride concentration is necessary to maintain the growth of the pit; the experimental value was 1.6 mol l-1 Cl- and the corresponding open-circuit potential was Ecrit = -0.48 V (s.h.e.). The pitting potential, Ev, was interpreted as an overpotential, ηp, given by η = Ep,- Ep-Ecrit.

The potential-pH-temperature relationships for the aluminium-water system have been calculated by the methods of de Bethune, Khodakovskiy, Criss and Cobble, and Helgeson and a critical comparison made. The Criss and Cobble method produced the most consistent results and was used to construct a corrosion diagram for the range 25-300�C. The effect of the hydrolysed ions Al(OH)2+, Al(OH)2+ and Al(OH)30 was also calculated. The work has shown up a temperature effect which should be taken into account when designing and operating aluminium circuits in which different parts are operated at different temperatures.