AI~(SO~)3 IN I-I2804 ELECTROLYTE 429tanned with oxalic acid addition. Glycolic and tartaric acid additions were nearly as good. Chromic and phosphoric acids actually decreased the coating ratio, acting as if more free H2S04 had been added.

Scaling rates and scale compositions of nickel-manganese alloys were determined.All the alloys scaled according to the parabolic rate law between 600 ~ and 1000~ At any given temperature the scaling rate increased at low manganese concentrations, then levelled off at intermediate concentrations, approaching the scaling rate of manganese as the upper limit.Both an external scale and a subscale were found after scaling, the scale composition being a function of alloy composition and temperature. Above a critical concentration of manganese (15% at 600~ to 60% at 1000~ the external scale consisted exclusively of manganese oxides; the subscale was MnO. Below this critical concentration, complex external scales consisting of the oxides of both nickel and manganese were found along with subscales of either NiO or a solid solution of the monoxides (MnO + NiO). The spinel oxide (NiO.Mn203) found in most of the complex scales was not associated with improved oxidation resistance. Schematic isothermal sections of the deduced Ni-Mn-O phase diagram were applied as an aid in interpreting the scaling behavior. It is concluded that none of the current theories of scaling of alloys describes the present case. SCALING BEHAVIOR OF NICKELA summary of the crystal structures of nickel and its oxides and nitrides is given in Table I. The nickel-oxygen phase diagram is available (1).Nickel oxidizes according to the parabolic rate law in the temperature range 600~176 Scaling constants reported by numerous investigators have been plotted on log K vs. lIT coordinates in Fig. 1. All of the data--except those of the three latest studies (18-20)--cluster along a straight line. The scatter about this line has been attributed to the effect of those "impurities most frequently found in commercial nickel, i.e., manganese and iron. Wagner (2) predicts that oxidation rates of metals such as nickel which form metal-deficit oxides" are increased by additions of metals of higher valencies such as manganese. The markedly lower scaling constants obtained in the three latest studies cannot be explained with any definiteness. Moore and Lee (20) used preoxidized samples which may have affected their results. Gulbransen and Andrew (18) claim their results differ because the metal was exceptionally pure; however, it'seems unusual that nickel of 99.9 % purity has ten times the scaling resistance of nickel of 99.8% purity (carbonyl nickel). And, in fact, this is not true if the scaling constant at 750~ determined from their data is included in Fig. 1. (This point was not included in their plot of log K vs. lIT.)This point differs from the older work by a factor of two at the most. It may well be that better agreement could be obtained at still higher temperatures. However, it should be noted that Frederick and Cornet (19) also obtained lower rate constants using high purity carbonyl nickel.Excluding the data of these latest studies, there is little difference in the scaling rate in an atmosphere of pure oxygen (PO2 = 760 mm Hg) as compared with air (PO~ = 152 ...