![]() ![]() |
![]() |
UFDC Home | Search all Groups | UF Institutional Repository | UF Institutional Repository | UF Theses & Dissertations | Vendor Digitized Files | | Help |
Material Information
Subjects
Notes
Record Information
|
Full Text |
MECHANISMS AND INHIBITION OF DEALLOYING IN AN ALPHA BRASS By WILLIAM CLARENCE FORT, III A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1975 Dedicated to my wife, Deborah. ACKNOWLEDGEMENTS I would like to express my appreciation to the chairman of my supervisory committee, Dr. Ellis D. Verink, Jr., for his encouragement and inspiration throughout the course of my studies and research. Thanks are extended to Dr. R. T. DeHoff, Dr. R. W. Gould, and Dr. G. M. Schmid for serving as members of the supervisory committee, and to Dr. C. M. Chen for his many helpful suggestions. I wish to acknowledge the financial assistance made available by the College of Engineering of the University of Florida, the NDEA Title IV fellowship Program, the Office of Naval Research, and the International Nickel Company. In addition, certain of the equipment used in these investiga- tions was purchased with funds from the Office of Saline Water. I also extend special appreciation to Mr. P. M. Russell, Mr. M. H. Froning, Mr. J. T. Healey, and Dr. R. G. Connell for their unselfish assistance in the final moments. I am indebted to Mr. W. A. Acree, Mr. E. J. Jenkins, Mr.E.C. Logsdon, Mr. C. J. Minier, and Mr. C. Simmons for their expert technical assistance during the course of the inves- tigation., TABLE OF CONTENTS Page ACKNOWLEDGEMENTS . LIST OF TABLES . . LIST OF FIGURES. . iii . vi . vii . xiii ABSTRACT CHAPTER 1 INTRODUCTION . CHAPTER 2 LITERATURE REVIEW . 2.1. Selective Leaching Mechanism ... .. .. 5 2.2. Dissolution and Replating Mechanism .... 20 2.3. The Role of Arsenic as an Inhibitor of Dezincification. . 29 CHAPTER 3 EXPERIMENTAL PROCEDURE ... .34 3.1. Potentiokinetic Polarization Experiments 34 3.2. Selective Leaching Experiments ... 37 3.3. Controlled Anodic Dissolution Experiments 38 3.4. Artificial Occluded Cell Experiments 39 3.5. Auger Electron Spectroscopic Analyses 42 3.6. X-Ray Identification of Corrosion Products 51 CHAPTER 4 RESULTS AND DISCUSSION . . 4.1. Electrochemical Characterization of Copper, Zinc, and Cu30Zn . . 4.2. Dealloying of Cu30Zn by Selective Leaching 4.3. Dealloying of Cu30Zn by Dissolution and Replating . . . 4.4. Effect of Arsenic Additions on Dealloying of Cu3OZn . . . CHAPTER 5 CONCLUSIONS . . CHAPTER 6 RECOMMENDATIONS FOR FURTHER RESEARCH . S52 S52 66 . 103 S150 . 165 S169 TABLE OF CONTENTS (Continued) Page APPENDICES 1 2 3 4 5 6 7 SAMPLE CHARACTERIZATION AND PREPARATION. . SOLUTION CHARACTERIZATION AND PREPARATION. STANDARD POLARIZATION CELL . LIST OF EQUIPMENT . . AUGER ELECTRON SPECTROSCOPY . CALIBRATION OF THE ARGON SPUTTERING RATE FREE ENERGY DATA USED IN CALCULATED EQUI- LIBRIA . . . BIBLIOGRAPHY . . BIOGRAPHICAL SKETCH . S171 S174 S175 178 179 S185 S187 S188 S193 LIST OF TABLES Table 1 Conditions of Exposure for the Cu30Zn Selective Leaching Samples . 2 Rate of Zinc Dissolution from Cu30Zn Under Selective Leaching Conditions . 3 Thicknesses of the Zinc-Depleted Surface Layers . . . 4 Summary of Diffusion Analysis Data . 5 Plateau Values of Dissolved Copper Con- centrations Measured in Controlled Anodic Dissolution Experiments . Page . 67 . 72 . 93 . 99 .132 LIST OF FIGURES Figure Page 1 Superposition of calculated equilibria for the Cu-O.1M C1--H20 and Zn-H20 systems and experimental potential versus pH diagram for Cu30Zn alloy in 0.1M chloride solutions, after Verink and Heidersbach (18) 7 2 Kink-step-terrace model of a dissolving alloy surface, after Pickering and Wagner. 10 3 Schematic of the equipment set-up for generation of potentiokinetic polarization curves . . 35 4 Artificial occluded cell electrode assembly 40 5 Schematic of the electrical circuit for the artificial occluded cell experiments .. .43 6 Schematic of the AES system used to analyze selective leaching samples . .45 7 Typical Auger spectrum from the surface of a Cu3OZn sample .. . 46 8 Effect of the composition of Cu-Zn alloys on the Cu/Zn ratio obtained by AES analysis. 49 9 Auger composition profile obtained for Cu-15 w/o alloy . 50 10 Experimental potential versus pH diagram for pure copper in nitrogen saturated, 0.1M chloride solutions .. 54 11 Superposition of polarization data for pure copper in 0.1M chloride solutions and the equilibrium potential versus pH diagram calculated for the system Cu-O.lM C1'-H20 at 250C, after Van Muylder et al. (71) 57 vii LIST OF FIGURES (Continued) Figure Page 12 Experimental potential versus pH diagram for pure zinc in nitrogen-saturated 0.1M chloride solutions . ... 59 13 Superposition of polarization data for pure zinc in 0.1M chloride solutions and the equilibrium potential versus pH dia- gram calculated for the system Zn-H20 at 250C, after Pourbaix (72) . 61 14 Experimental potential versus pH diagram for Cu30Zn in nitrogen-saturated 0.1M chlo- ride solutions . ... 62 15 Variation of the corrosion potential of Cu30Zn with time of exposure in 0.1M chloride solutions of pH 2 and pH 4. 64 16 Zinc dissolved from Cu30Zn samples poten- tiostated at -0.700 VSCE in 890C, 0.1M chloride solutions of pH 4 . 70 17 Zinc dissolved from Cu30Zn samplespoten- tiostated at electrode potentials between -0.500 and -0.900 VSCE at 890C . 71 18 Pitting of Cu30Zn sample exposed 4 days at -0.450 VSCE. 3700X . 75 19 Typical surface of Cu30Zn samples poten- tiostated below -0.450 VSCE. From a Cu30Zn sample exposed 4 days at -0.600 VSCE. 3500X . ... 76 20 Auger composition profile obtained from an unexposed Cu30Zn sample . 78 21a Auger composition profile obtained from a Cu3OZn sample potentiostated at -0.450 VSCE for 1 day . 80 21b Auger composition profile obtained from a Cu30Zn sample potentiostated at -0.450 VSCE for 4days . ... 81 viii LIST OF FIGURES (Continued) Figure 21c Auger composition profile obtained from a Cu30Zn sample potentiostated at -0.450 VSCE for 7 days . . 22 Auger composition profile obtained from a Cu3OZn sample potentiostated at -0.500 VSCE for 4 days . . 23 Auger composition profile obtained from a Cu3OZn sample potentiostated at -0.600 VSCE for 4 days . . 24a Auger composition profile obtained from a Cu3OZn sample potentiostated at -0.700 VSCE for 2 days . . 24b Auger composition profile obtained from a Cu3OZn sample potentiostated at -0.700 VSCE for 4 days . . 24c Auger composition profile obtained from a Cu3OZn sample potentiostated at -0.700 VSCE for 7 days . . 24d Auger composition profile obtained from a Cu3OZn sample potentiostated at -0.700 VSCE for 10 days . . 25 Auger composition profile obtained from a Cu30Zn sample potentiostated at -0.800 VSCE for 4 days . . 26 Auger composition profile obtained from a Cu30Zn sample potentiostated at -0.900 VSCE for 4 days . . 27 Infinite diffusion model used in zinc diffusion analysis . . 28 Log-log plot of the zinc diffusion data taken from composition profiles obtained by AES from Cu30Zn samples selectively leached at -0.700 VSCE . . Page . 82 . 84 85 . 86 . 87 . 88 . 90 . 91 S. 95 100 LIST OF FIGURES (Continued) Page Figure 29 Simplified potential versus pH diagram for Cu30Zn in 0.1M chloride solutions outlining conditions of exposure for controlled anodic dissolution experi- ments . . 30 Etched surface of a Cu30Zn sample potentiostated 13 days at -0.200 VSCE in pH 4, 0.1M chloride solution. 1500X . . 31 Curves showing the concentrations of copper and zinc present in solution during dissolution of Cu30Zn at -0.200 VSCE. . . 32a Pitting observed on etched Cu30Zn sam- ple exposed 10 days at -0.100 VSCE. Two copper crystals formed on the sur- face are marked by arrows. 125X . 32b Detail of one of the crystals found on the sample described in figure 32a. The tetrahedral shape is outlined in the accompanying sketch. 300X . 32c Cross section of one of the copper crystals detailing the Cu30Zn grain boundary artifacts visible in the de- zincified structure. 75X . 33 Concentrations of copper and zinc mea- sured in the solution of a cell contain- ing a Cu30Zn sample polarized 10 days at -0.100 VSCE . . 34 Dezincification noted on a Cu30Zn sam- ple exposed 3 days at 0.000 VSCE. 100OX 35 Dezincification noted on a Cu30Zn sam- ple exposed 13 days at 0.000 VSCE. 60X 36 Results of solution analysis for Cu30Zn samples potentiostated at 0.000 VSCE for 7 days and 13 days . . . 104 . 107 . 108 . 111 112 . 113 . 115 117 . 118 119 LIST OF FIGURES (Continued) Figure 37 Dezincification noted on a Cu30Zn sample exposed 10 days at +0.100 VSCE. 100X . 38 Results of solution analysis for a Cu30Zn sample potentiostated 10 days at +0.100 VSCE. . . 39 Detail of sample surface of a Cu30Zn sam- ple exposed 13 days at +0.200 VSCE, cov- ered by a layer of cuprous chloride. 1500X 40 Results of solution analysis for a Cu30Zn sample potentiostated 13 days at +0.200 VSCE . . 41 Results of solution analysis for a Cu3OZn sample potentiostated 10 days at +0.500 VSCE . . 42 Extent of dezincification noted on a Cu30Zn sample exposed 10 days at +0.500 VSCE. 60X . . 43 Equilibria calculated for the Cu-C1--H20 system at pH = 4 showing agreement between equilibrium calculations and constant cop- per concentrations observed in the control- led anodic dissolution experiments . 44 Cu30Zn sample potentiostated 4 days at +0.500 VSCE, showing separation between the brass and the copper sponge structure. 250X . . . 45 Results of artificial occluded cell experi- ments showing occluded cell potential and occluded cell current measured as a func- tion of time . . 46 Results of solution analysis for arsenical Cu30Zn potentiostated at -0.200 VE in pH 4, 0.1M chloride solution . Page . 122 123 . 125 . 126 . 127 . 133 . 138 . 141 . 154 LIST OF FIGURES (Continued) Figure 47 Results of solution analysis for arseni- cal Cu30Zn potentiostated at -0.100 VSCE in pH 4, 0.1M chloride solution . 48 Results of solution analysis for arseni- cal Cu3OZn potentiostated at -0.000 VSCE in pH 4, 0.1M chloride solution . 49 Results of solution analysis for arseni- cal Cu30Zn potentiostated at +0.100 VSCE in pH 4, 0.1M chloride solution . 50 Results of solution analysis for arseni- cal Cu30Zn potentiostated at +0.200 VSCE in pH 4, 0.1M chloride solution . 51 Results of solution analysis of arseni- cal Cu30Zn potentiostated at +0.500 VSCE in pH 4, 0.1M chloride solution . 52 Standard polarization cell . 53 Energy level diagram depicting a KLM Auger transition . . 54 Schematic energy spectrum and experimental AES curve for silver excited in a 1000 eV electron beam . . 55 Composition profile of a 102625A copper film deposited on a zinc substrate, used to calibrate the argon sputtering rate Page . 157 158 . 159 161 . 176 . 180 S 182 . 186 xii Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MECHANISMS AND INHIBITION OF DEALLOYING IN AN ALPHA BRASS By William Clarence Fort, III July, 1975 Chairman: Dr. Ellis D. Verink, Jr. Major Department: Materials Science and Engineering Dealloying of a single-phase binary alpha brass in pH 4, 0.1M chloride solution was investigated in terms of the operative mechanism. Selective leaching of zinc from the brass was produced under conditions of cathodic polari- zation. Auger electron spectroscopy was employed to mea- sure composition versus depth profiles of the resulting zinc-depleted surface layers. The selective leaching mecha- nism was found to be based on enhanced volume diffusion of zinc to the brass surface. Dealloying by dissolution of the brass as a whole, followed by redeposition of the copper, was found to occur exclusively under occluded cell conditions. The mechanism was examined in terms of the copper-chloride equilibria established at the dissolving brass surface. The redeposition of copper was found to proceed by reduction of cuprous chloride near the brass surface. The inhibitive ef- fect of arsenic was determined to be the prevention of the cuprous chloride reduction reaction. xiii CHAPTER 1 INTRODUCTION Since the initial description of the dealloying phe- nomenon in brasses more than 100 years ago (1), the metals literature has witnessed an ongoing debate as to the mecha- nism(s) responsible. Notwithstanding that rather early in the century the practical problem of dezincification of alpha brasses was solved by alloying with arsenic (2), a considerable amount of research effort is even now being directed towards understanding the phenomena involved. This can be at least partly attributed to ongoing discoveries of dealloying behavior in a host of diverse alloy systems (3). In addition, several reports in more re- cent years have linked dealloying behavior with stress cor- rosion cracking in several systems of interest (4-8). It remains the goal of researchers in this area of investigation to apply a basic understanding of the mechanisms of dealloy- ing in brasses to solve the dealloying problems encountered in other systems, and to predict when dealloying might be expected in new applications. Two principal mechanisms for dealloying* of single phase binary alloys in aqueous environments have gained wide acceptance: (1) Selective leaching of the less noble component of the alloy from the metal matrix into solution without cor- responding dissolution of the more noble component; and (2) Total dissolution of the alloy (both components enter- ing solution at rates proportional to their respective concentrations in the alloy) followed by redeposition of the more noble component metal on or adjacent to the dissolving alloy surface. Almost without exception, one or the other of these mechanisms has been assigned to each reported dealloying occurrence. Efforts to resolve the nature of the mechanism operating for any one system have resulted, for the most part, in confusing and often contradictory conclusions. It is the goal of this research to clarify the mecha- nisms responsible for dealloying in a defined system - Cu-30 w/o(weight percent) Zn [hereafter designated as Cu30Zn] exposed in aqueous solutions containing 0.1 molar of chloride ion. The choice of this system is based on several factors. Cu30Zn has been and remains the most important binary brass alloy. For this reason, if none other, it has also received the majority of research attention. Consequently, a large * Dealloying is defined as the corrosion process whereby one constituent of an alloy is preferentially removed from an alloy, leaving an altered residual structure (3). volume of literature concerning its corrosion and physical properties is available. The fact that it is a single phase alloy eliminates complications which might otherwise result from galvanic interactions between dissimilar phases. The possibility of phase changes under corrosion conditions is likewise eliminated. Further, a large proportion of the dezincification failures reported have occurred in saline environments. Thus, the role of the chloride ion in causing or in enhancing de- zincification cannot be overlooked. This investigation is divided intofour main sections: (1) Definition of the polarization behavior of the system in order to establish a firm basis for predicting the conditions which lead to dealloying, and the effects of these conditions on the mechanism; (2) Elucidation of the detailed mechanism involved in se- lective leaching of zinc from the brass matrix through use of surface sensitive techniques; (3) Clarification of the chemical and electrochemical equi- libria which govern dezincification of alpha brass in chloride-containing environments; and (4) Clarification of the role played by arsenic additions to alpha brass in inhibiting dezincification. Special attention is payed to the role of crevice or occluded cell phenomena in the initiation and propaga- tion of the attack. CHAPTER 2 LITERATURE REVIEW Investigations of dealloying mechanisms have been limited almost entirely to the binary copper-zinc and copper- gold systems. The research popularity of non-commercial copper-gold alloys is due to the following factors: (1) the system is isomorphous, hence phase changes with accompanying compositional changes need not be considered; (2) gold ex- hibits a very noble single electrode potential, and does not dissolve under ordinary circumstances (the dissolving compo- nent in this case is copper); (3) there is little overlap in the X-ray spectra of copper and gold; and (4) the melt- ing points of copper and gold are similar, and alloying is relatively simple. Dealuminification of the commercially important alu- minum bronzes has received only limited attention regarding mechanisms, primarily because the problem has been remedied through slight compositional adjustments and improved heat treatment procedures (9-11). The literature concerning de- alloying in other systems is almost entirely limited to re- ports of in-service failures. Several excellent reviews covering the entire dealloying spectrum are available (3, 12-15). 5 2.1. Selective Leaching Mechanism The theoretical aspects of dealloying by a selective leaching mechanism have been described most eloquently by Pickering and Wagner (16). They state that preferential re- moval of the less noble component of a binary alloy under- going anodic dissolution requires that: (1) there be a sufficiently large difference (i.e., several RT/F) between the single electrode potentials of the two constituent metals; and (2) the electrode potential of the alloy be higher than that of the less noble component and significantly lower than that of the more noble component. Such is the case for Cu- Au alloys and beta brass. The electrode potential of alpha brass, however, is nearly the same as that of its more noble component, copper (17). The important condition here is that, at a given alloy electrode potential, there be a much greater tendency for the more active component atoms to dissolve than for the noble component to dissolve, based on the respective single electrode potentials of the component metals. Thus, alpha brass polarized several RT/F below the single electrode po- tential for copper (i.e., cathodically) meets this require- ment. Verink and Hei'dersbach (18) have proposed that the electrochemical conditions outlined above as necessary for dealloying by a selective leaching mechanism can be rather simply predicted by comparing the calculated and experimen- tal potential versus pH diagrams for the system of interest. Thus, referring to figure 1, selective leaching of zinc from a Cu30Zn alloy exposed in neutral and acidic 0.1M chloride solutions free of copper ions would occur at potentials more noble than the calculated Zn/Zn+(10-6M) equilibrium and less noble than the experimentally determined corrosion po- tentials for Cu30Zn (area of small dots). The authors tested this hypothesis by exposing Cu30Zn samples in buffered 0.1M chloride solutions of pH 4 for up to 13 days at several potentials in the predicted range (19). They report that zinc is dissolved into the electrolyte with no discernable accompanying copper disso- lution. However, the formation of a dezincified layer was not confirmed although tarnishing was observed. The iden- tity of tarnish layer could not be determined by any of the then currently available techniques. The results of Verink and Heidersbach are consistent with earlier investigations of selective dissolution of Cu- Zn and Cu-Au alloys. Pickering and Byrne (20) measured par- tial currents (by solution analysis) for the dissolution of copper and zinc from alpha, gamma, and epsilon brass speci- mens held at potentials between -1.000 and +0.500 VSHE in acetate buffered (pH = 5) 1N Na2SO4 alloy where only zinc was dissolved in detectable quantities. At potentials more noble than certain critical potentials, simultaneous 3 4 5 6 7 8 9 10 II 12 13 Figure 1. Superposition of calculated equilibria for the Cu-O.lM C1--HzO and Zn-H20 systems and experi- mental potential versus pH diagram for Cu30Zn alloy in 0.1M chloride solutions, after Verink and Heidersbach (18). -1.4 dissolution of the alloys occurred. Similar results were reported by the same authors for Cu-13a/o (atomic percent) Au and Cu-18a/o Au alloys potentiostated between -0.350 and +0.875 VSHE in IN Na2SO4- 0.01N H2SO4 solutions (21). In this case, however, simul- taneous dissolution does not occur above the critical po- tentials. Instead, severe roughening of the alloy surface was noted. These results are in substantial agreement with those of earlier investigators (22) for the same system. Ionization of the less noble component results in rapid accumulation of the other components' atoms on the alloy surface. If selective dissolution is to continue, some means must be established to provide the atoms of the more active component access to the liquid-solid interface. The two likely possibilities are (1) Aggregation of the more noble component atoms by sur- face diffusion, exposing new alloy surface containing active atoms; and (2) Transport by volume diffusion of active metal atoms to the liquid-solid interface. Simultaneous dissolution is defined by both alloy components entering solution at rates proportional to their respective concentrations in the alloy. tImplicit in this argument is that the ionization of the ac- tive atoms at the liquid-solid interface and their conse- quent diffusion into the bulk electrolyte is fast and not rate-controlling. This assumption has been recently dis- cussed in detail by Rubin (17). Pickering and Wagner (16) favor the latter possibi- lity and have proposed a divacancy diffusion mechanism which counters the classical argument against such a concept (i.e., the sluggishness of transport by simple bulk diffusion at ambient temperatures). The mechanism is based on a kink- step-terrace model of the surface of the dissolving metal alloy A-B, figure 2. Ionization of the less noble component atom A at kink position 1 proceeds by its moving along the step and onto the terrace as an adsorbed atom, from where it passes into solution as an ion. The more noble component atom B initially at step position 2 may then move along the step, and become adsorbed, but not ionized, or it may desorb. As this process continues, the concentration of adsorbed B atom increases, thereby increasing the probability that ad- sorbed B atomsdesorb to kink sites, and consequently de- creasing the probability that active A atoms occupy kink sites. This leads to increased electrochemical polarization and increased driving force for ionization of A atoms. The increased driving force makes possible direct adsorption of A atoms (position 6) from step sites and, eventually, direct adsorption from terrace sites. The step and surface vacan- cies resulting from such events may in some cases be filled by subsurface atoms, thereby introducing excess monovacan- cies and divacancies into the alloy matrix. The buildup of an excess vacancy concentration in the vicinity of the sur- face would enhance chemical diffusion (i.e., diffusion in dissolved ion 6 A 34 A step position 2 B kink position I L / erroce . position 77 absorbed atom Figure 2. Kink-step-terrace model of a dissolving alloy surface, after Pickering and Wagner. the concentration gradient due to enrichment of the surface in component B) of A atoms towards the surface and B atoms into the subsurface layers. The authors have presented calculations of the cur- rent density and effective diffusion zone thickness expected for dissolution of copper from a Cu-10 a/o Au alloy at 250C. Several assumptions have been made which are open to ques- tion (17), among them: (1) the concentration of divacancies near the surface is -2 taken to be as high as 10- a/o; (2) the accuracy of divacancy diffusion coefficient values extrapolated to 25C are realistic; (3) the diffusion coefficient value used is assumed to be composition independent; and (4) contributions to the total diffusion rate by short- circuit diffusion paths (subgrain boundaries and dis- locations) are ignored. Notwithstanding these criticisms, the authors calculate a current density of about 2 x 10-4 amp/cm2 and an effective diffusion zone thickness equal to about 106 cm (0.01 um) after 1000 seconds'exposure. Experimental verification of the model calculations presented above were undertaken by the authors. Cu-10 a/o Au sheets dissolved at 1 ma/cm2 for 320 min (19.2 coulombs/cm2 passed) in 1N H2SO4 or in buffered 1N NaC1 solution developed gold-rich surface layers approximately 8 pm thick. Based on the amount of charge passed in the experiments, 8 pm is very close to the depth from which total dissolution of copper would occur. Calculation of the effective diffusion zone thickness from the divacancy diffusion model gives the maximum thickness as only about 0.7 pm for 320 min exposure. These conflicting results can be reconciled by con- sidering the surface roughening noted to have occurred in the gold-rich layers (23). As the attack proceeds, the surface roughening effect introduces channel-like penetrations and porosity, thereby significantly increasing the surface area of the gold-rich surface layers. Consequently, a much larger surface area is exposed to the solution, and can undergo se- lective dissolution. This mechanism results in the formation of a leached structure significantly thicker than could be formed without roughening. The operation of a selective leaching mechanism does not preclude simultaneous operation of dissolution and replat- ing mechanism. Pickering and Wagner have discounted the possi- bility that significant numbers of gold atoms are ionized on the basis of rotating ring disk electrode measurements on a Cu-10 a/o Au alloy specimen dissolved at polentials up to +0.900 VSHE. The limitations of this method in measuring very small concentrations has been pointed out by Rubin (17). At any rate, the effects of increased overpotentials on the mo- bilities of adsorbed gold atoms and any resultant surface diffusion effects are unknown. Under appropirate polarization conditions, complete dissolution and selective leaching might be expected to occur simultaneously. Such would be the case if a Cu-Au alloy were dissolved at potentials close to the single electrode poten- tial for gold, or if alpha brass were dissolved at very low anodic overpotentials. The consequent effect of the reced- ing metal-solution interface on the interdiffusion zone thick- ness has been treated by Holliday and Pickering (24) for the case where the rate of selective leaching and the rate of total dissolution are similar. After an initial period during which the metal diffusion profiles are established, a station- ary state is reached, wherein simultaneous dissolution from the surface critically enriched in the more noble component occurs. Thereafter, the rate of selective leaching of the more active component is proportional to the total alloy dis- solution rate, and the diffusion profiles adjacent to the surface (i.e., the level of surface enrichment of the more noble component) remain constant as dissolution continues. The role played by surface diffusion in the selec- tive dissolution mechanism has received relatively little attention. Pickering and Wagner (16) discount its importance, pointing out that noble component atoms migrating on the metal surface would agglomerate and form small crystals. As these crystals grow to impingement, blockage of the active atoms' access to the liquid-solid interface occurs, and the leaching process is stifled. The ease with which blockage would occur presumably is related to the alloy composition, i.e., alloys containing larger percentages of the noble element would be more immune to significant leaching than those containing more of the active component. This possible explanation of the fact that high brasses, containing less than about 10 w/o of zinc, have not been observed to dezincify has been hinted at by Feller (25). Feller discounts normal unaided isothermal sur- face diffusion as being too slow to produce measurable ef- fects. Instead he proposes that under appropriate high anodic potentials, surface diffusion is enhanced. This view- point is supported by Tischer and Gerischer (26), who report that surface mobility of gold depends on the electrode po- tential. Rubin (17) has recently proposed that noble and ac- tive atoms might combine via surface diffusion to form patches of alloy having a composition more stable under the prevailing surface condition than the initial alloy compo- sition. As is pointed out, the mechanism could be used to explain the detection of intermediate compositions in de- alloyed Cu-Au alloys, to be discussed subsequently. Direct experimental evidence for the operation of the selective leaching mechanism in dealloying is rare. Furthermore, quantitative data concerning leaching rates, layer thickness, and the detailed diffusion processes in- volved are almost completely lacking. The optical metallographical evidence for selective leaching is primarily speculative in nature, and has in most instances been successfully challenged (3). Aside from the obvious comment, that the extent of attack found in most dealloying experiments cannot be accounted for by even the most optimistic diffusion mechanism, consideration is rarely given to the factors and processes which result in the for- mation of the grossly porous residual metal sponges often found.* Explanation of the formation of such structures by a selective leaching mechanism is difficult. X-ray and electron diffraction examinations of de- alloyed metals have yielded the best evidence to date. The applicability of the method to dealloying measurement rests on the assumption that the lattice parameter of the alloy undergoing selective leaching will evolve in some orderly fashion from its initial value towards that of the more noble component. Thus, the diffraction pattern obtained from a partially leached alloy would contain "bands" of intensity limited by the normal diffraction line positions of the alloy and the noble metal. Furthermore, if dealloy- ing has taken place by a dissolution and replating mechanism, then only the lines corresponding to the initial alloy and to the more noble metal should be detected. *Some experiments by Lucey (27) are the notable exceptions. Graf's early X-ray diffraction experiments (28, 29) indicated the formation of a series of crystals containing from 60 to 80 atomic percent of gold on Cu-10 a/o Au alloy specimens exposed in alcohol-picric acid solutions or in alkaline solutions of sodium tartrate at 25C. He found, however, that Cu-25 a/o Au specimens, dealloyed in more highly oxidizing nitric acid solutions at 70C, formed no regions of intermediate composition;only diffraction spectra for the original alloy and for pure gold were found. Pickering and Wagner (16) refined Graf's method and applied it to Cu-10 a/o Au foils dissolved in room tempera- ture 1N H2SO4 and in 1N NaC1 solution at current densities 2 up to 20 ma/cm. They report the emergence of maxima in diffracted intensity between the Cu-10 a/o Au peaks and the pure copper peaks. These maxima move towards the gold peaks as the amount of dissolution increases. Later electron dif- fraction experiments on a series of copper-rich Cu-Au alloys exposed under the same conditions were reported to confirm the findings for the earlier stages of dissolution (30, 31). The positions of the intensity maxima observed in patterns from dealloyed Cu-Au alloys containing 3, 5, and 10 a/o of gold were observed to be nearly the same, and to correspond to those of an alloy of approximate composition Cu-69 a/o Au. Similar formation of equivalent intermediate compositions by a series of alloys has been described in the Cu-Ni system (17). In this case the stability of an intermediate composition was attributed to altered thermodynamic condi- tions present at the surface of the dissolving metal. Similar X-ray diffraction determinations were car- ried out by Heidersbach and Verink (19) on alpha (Cu-30 w/o Zn) and beta (Cu-48 w/o Zn) brass specimens exposed in 5N HC1 for 20 and 30 days and 2 days, respectively. Their X-ray patterns from the alpha brass show a band of increased intensity between the peaks due to copper and to the Cu30Zn alloy. A similar pattern reported for a beta brass sample shows peaks for beta brass and copper and a peak whose posi- tion corresponds to an alpha brass containing approximately 36 w/o of zinc (32). The diffraction results described above embody the only undisputed evidence for the operation of a selective leaching mechanism in dealloying. The diffraction methods are limited, however, in that they yield very little infor- mation about the distribution in the dealloyed structure of new alloy compositions. In both investigations discussed above, the electron microprobe was employed to obtain profiles of composition across all or part of the dealloyed layers. The gold and copper profiles reported for dealloyed Cu-lO a/o Au speci- mens show a monotonic increase in gold concentration and a corresponding decrease in copper concentration over the width (,10 pm) of the dealloyed layer (30, 31). Zinc composi- tion profiles reported for alpha and beta brass were taken only close to the interface between the bulk metal and the dezincified layer (19, 32). The widths of layers in which the zinc contents drop from 30 w/o to zero are less than 10 pm for alpha brass exposed in 1N NaC1 for 7 days at room temperature and in 5N HC1 for 10 days at 750C, but about 30 pm wide for beta brass exposed in 5N HC1 for 2 days at 75C. It should be pointed out that concentration profiles obtained by spectrographic techniques reflect only the aver- age composition of the material in the volume of measure- ment. Thus, a profile obtained from a binary alloy sample having a uniform gradient of depletion in one element may be indistinguishable from a sample composed of fine particles of each of its constituents and having only a uniform gra- dient in the fraction of particles of each element. For this reason, such profiles by no means represent unequivocal proof of the operation of a selective leaching mechanism. The possibility that selective dissolution might result in the formation of stable phases having compositions intermediate to the starting alloy and the more noble compo- nent has been tested in two investigations of dealloying in high zinc content brasses. Stillwell and Turnipseed (33) used X-ray diffraction to study epsilon brass (. Cu-80 w/o Zn) samples exposed in a series of acid solutions. They found that in acetic acid or dilute HC1, gamma brass (% Cu- 65 w/o Zn), beta brass, and alpha brass or copper were successively formed on the epsilon brass substrate. How- ever, in stronger acids (HNO3, H2SO4, and concentrated HC1), only copper was formed, or complete dissolution took place. Pickering (34) has extended the analysis to epsilon (Cu-86 a/o Zn) and gamma (Cu-65 a/o Zn) brasses dissolved anodically (1 to 5 ma/cm2) in IN H2SO4 and in a series of buffered (pH = 5) solutions. He reports the formation of alpha brass on the gamma samples and alpha and gamma brass on the epsilon brass samples. In addition, the starting brass compositions were observed to shift towards higher copper contents, thus reflecting depletion of zinc. In summation, several important conclusions can be drawn about our knowledge of the selective leaching pheno- menon as it pertains to metal corrosion. (1) The theoretical aspects of dealloying by selective leaching of the less noble component from the alloy matrix are rather well-developed, detailed models having been proposed. (2) Undisputed experimental evidence for the occurrence of selective leaching in the Cu-Au and Cu-Zn alloy systems exists in the form of X-ray diffraction measurements of lattice parameter changes. (3) No direct measurements of the physical scale of deal- loyed layers resulting from the operation of a selec- tive leaching mechanism exist. (4) Consequently, little is known of the operational de- tails of the selective leaching mechanismss. These conclusions clearly point out the need for a new experimental approach in analyzing leached structures. Such an approach is contemplated in Auger electron spectro- scopy, a highly surface-sensitive experimental technique. 2.2. Dissolution and Replating Mechanism In contrast to the Cu-Au alloys which undergo selec- tive dissolution rather easily, the alloys of the Cu-Zn sys- tem normally dissolve simultaneously in aqueous environments. This observation led early investigators to conclude that the porous layers of copper remaining on brass after dezinci- fication must certainly be the result of redeposition (at favorable sites on the brass surface) of copper which had previously passed into solution. Pickering and Wagner (16) have analyzed the condi- tions for simultaneous dissolution and replating of a binary alloy for the case where no completing reactions are possible and no solid products are formed. Consider dissolution of a binary Cu-Zn alloy by the reactions Cu[Cu-Zn] Cu+n + ne-, n = 1, 2 (1) Zn[Cu-Zn] Zn++ + 2e (2) If the reactions are independent, then the single electrode potentials for dissolution of each component are given by a +n E Eo + RT In Cu (la) Cu[Cu-Zn] Cu nF a u[Cu- aCu[Cu-Zn] and by a+2 EZC- = Eo + T n Zn (2a) Zn[Cu-Zn] Zn 2F aIn 2F aZn[Cu-Zn] Subsequent redeposition of dissolved copper ions at copper- rich surface sites, by the reaction Cu+n + ne- Cu[Cu sites], (3) occurs at an electrode potential given by +n E Eo RT n Cu (3a) Cu[Cu sites] u nF au[Cu sites] The activity of copper on the brass surface,aCu[Cu-Zn], is less than unity. Consideration of equations (la) and (3a) for these conditions leads to the conclusion that the single electrode potential for dissolution of copper from the brass, ECu[Cu-Zn], is more noble than the potential for deposition of copper, ECu[Cu sites]. Consequently, redeposition does not occur. If, however, the copper and zinc dissolution reac- tions are not independent, then copper atoms may tend to remain adsorbed on the surface as the zinc dissolves. These adsorbed copper atoms tend to aggregate on the surface. An isolated adsorbed copper atom not surrounded by other copper atoms possesses a higher activity than a copper atom which has aggregated and been surrounded by other copper atoms. Due to this activity reversal, the single electrode poten- tial for deposition of copper at copper-rich sites, equa- tion (3a), becomes more noble than the single electrode potential for dissolution of copper from the Cu-Zn surface, equation (la). Under these circumstances, copper dissolving from the brass surface may be plated out on copper-rich areas on the surface. The necessity of copper-rich sites as areas for ini- tiation of replating is intuitively attractive. Formation of such cathodic sites has been attributed to selective leaching of zinc from zinc-rich impurities at grain bound- aries (35), traces of beta brass, or compositional inhomo- genieties (36, 37). This rather simple argument has a basic flaw in that it fails to explain why dezincification of high brasses (i.e., brasses containing less than 15 w/o of zinc) has not been observed. Surely for such dilute alloys, the differ- ence in single electrode potentials for dissolution and replating of copper, equations (la) and (3a), respectively, is much smaller than in the case of brasses with higher zinc content, which do dezincify. Additionally, Pickering and coworkers have shown, both through analysis of metal ion pickup in solution (20) and by a soft X-ray diffraction technique (24), that simul- taneous dissolution of Cu-30 a/o Zn in acid sulfate solu- tion is preceded by an enrichment of the brass surface in copper. Similar results were obtained by Bumbulis and Graydon (38) for dissolution of a Cu-15 w/o Zn alloy. In each of these cases, no tendency towards further dezinci- fication by redeposition of copper was observed. Thus the existence of copper-rich sites on a dissolving brass sur- face does not necessarily represent a sufficient condition for the occurrence of copper redeposition. The situation discussed above is critically altered if the concentration of dissolved copper ions adjacent to a portion of the surface is increased above the equilibrium value corresponding to the average surface potential. Then, ECu[Cu sites] may become larger than ECu[Cu-Zn], and depo- sition of copper will occur. In connection with dealloying failures in service, copper concentrations higher than the equilibrium concentrations have been attributed to: (1) stagnant conditions (15); (2) decreased flow rates (19, 36); (3) porous surface films or deposits (39, 40); and (4) cre- vices or pits (41-43). It should be noted that the porous copper deposits resulting from dezincification produce stag- nant conditions in the solutions adjacent to the corroding brass surfaces beneath. Deposits of copper (27, 44, 45) and of nickel (46) have been employed to initiate dezinci- fication on brasses. The traditional, extensive use of brasses in saline environments has made the effects of chloride ions in pro- moting dezincification well known. Several early investi- gators recognized the importance of the complex salts of copper and chlorine which are commonly observed on dezinci- fication failures. Bengough, Jones, and Pirret (2) observed that brass immersed in aerated sodium chloride solution becomes covered by a film of sparingly soluble cuprous chloride Cu[Cu-Zn] + Cl- 1 CuCl + e- (4) which eventually is oxidized to cuprite and cupric chloride by the reaction 4CuC1 + 02 Cu20 + 2CuC12 (5) The authors hypothesized that the cupric chloride formed by this reaction attacks the brass according to the reactions CuC12 + Cu 2CuC1 (6a) and CuC12 + Zn ZnC12 + 2Cu. (6b) Of these reaction products, copper is deposited at the brass surface, soluble ZnCl2 passes into solution, and sparingly soluble CuCl is available to continue the reaction cycle. Abrams (39) demonstrated in a classic series of experiments the ability of cuprous chloride reaction pro- duct membranes as well as artificial membranes to pro- duce dezincification. He was also able to show that condi- tions which retard the formation of dense CuCl films, e.g., low pH and agitation, also retard preferential dissolution. More recently, Lucey (27) noted the physical simi- larities between pits in copper and dezincified areas in brasses formed in chloride containing environments. He reasoned that most of the copper dissolved from a pit is precipitated in the pit as cuprous chloride, some of which is subsequently oxidized to cuprite. A more detailed dis- cussion of pit formation and morphology has been presented by Van Muylder et al. (47). In the case of dezincification, however, the cuprous chloride formed through dissolution of the brass is reduced to copper. Further, beta brass is capable of direct cathodic reduction of cuprous ions. Alpha brass, having a much higher electrode potential, can reduce only cupric ions, which are formed by disproportionation of cuprous chloride 2CuC1 Cu + CuC12 (7) Therefore, according to this reasoning, dezincification of alpha brass is dependent upon the formation of cupric ions. Unfortunately, Lucey's rather low experimental value for the potential of a film-free alpha brass surface is unsubstan- tiated, making the validity of this mechanism questionable. Heidersbach and Verink (18, 19) have also recognized the importance of the copper-cuprous chloride reaction in producing dealloying in copper alloys. They observed the deposition of copper on Cu30Zn and CulONi samples polarized below a certain critical electrode potential during cyclic polarization in acid O.1M chloride solutions. This critical potential, +0.200 VSHE' also observed by Efird (48), corre- sponds closely to the equilibrium potential of the copper- cuprous chloride reaction Cu + Cl- CuCl + e- (8) On this basis, the authors have predicted the conditions of electrode potential for which copper alloys having polariza- tion behavior similar to that of pure copper may undergo dealloying by a dissolution and replating mechanism in neu- tral and acid chloride environments. Referring to figure 1, Cu3OZn undergoes simultaneous dissolution at electrode po- tentials more noble than its corrosion potential, the area of large dots. However, at potentials less noble than this critical potential, cuprous chloride formed on the brass surface as a result of dissolution may be reduced to copper. Thus, in the potential range bounded by the corrosion poten- tial of the brass and the critical potential, the area of crosshatching, dealloying by dissolution and replating may take place. Substantiation of this prediction was undertaken by potentiostating alpha and beta brass samples in deaerated 0.1M chloride solutions of pH 4 at several potentials in the predicted potential range. All beta brass samples were dezincified after a few days' exposure. The alpha brass samples dezincified only if their solutions were unstirred or after several weeks in stirred solutions. This result is consistent with earlier observations of agitation on the formation of adherent cuprous chloride layers (39). Essen- tially identical potential dependence has been reported by Marshakov and Bogdanov (49) for dealloying of several alpha, beta, and duplex brasses polarized anodically in 0.5M NaCl solutions. Comparable results have been obtained by Robinson and Shalit (50) and by Sugawara and Ebiko (51) for similar series of brasses potentiostated in 31% NaCl solutions. Additionally, Heidersbach and Verink exposed alpha and beta brass samples in solutions known to produce dezinci- fication. The steady state electrode potentials were mea- sured, and in each case were found to fall within the pre- dicted potential range for dealloying. The uncertainties involved in specifying electrode potentials required to produce, or to inhibit, dealloying have been pointed out by several investigators (19, 21, 48, 51). Potential drops due to anodic films, dealloyed layers, and crevices can alter the electrode potential at the attack interface. For example, as dealloying attack proceeds, the formation of a thick, porous residual metal sponge over the sample surface changes significantly the conditions adjacent to the dissolving alloy surface. This process is in many ways analogous to the development of pitting or crevice attack (52). By the same token, as the initiation and the propagation stages of pitting attack can be treated as separate processes (53), so ought the initiation and propagation stages of dealloying attack. A case in point is the initiation of dezincification in brass by deposition of copper onto the brass surface. Early investigators (54) referred to such a deposit as "apparent dezincification." The resulting dezincification which occurs beneath the deposit is true dezincification, and the condi- tions which sustain the attack are very probably quite dif- ferent from those initiating it. The determination of condi- tions at the dealloying attack interface is the prerequisite necessary for understanding how the attack proceeds and, ultimately, how to stop it. Unfortunately, heretofore very little work has been directed towards this area of investi- gation. 2.3. The Role of Arsenic as an Inhibitor of Dezincification Extensive research early in this century concerning the interplay of compositional variables and dezincification of brass condenser tubes resulted in the discovery of arsenic's inhibitive properties (2). Since that time, the addition of 0.02 to 0.04 w/o of arsenic to alpha brasses has become the standard means of preventing their dezincification (3). Re- search into the causes) of the effect, however, has proceeded haltingly, and no consensus has been reached concerning the identity of the mechanisms) involved. The several theories which have been proposed to explain the effect can be loosely classified into two groups: (1) Arsenic acts as an anodic inhibitor, interfering with dissolution of the brass or its zinc component; and (2) Arsenic acts as a cathodic inhibitor, interfering with copper deposition. Fink (36) first proposed the former mechanism, stat- ing that arsenic dissolved from the brass combined with cop- per to form copper arsenide films over the most "active" or zinc-rich points on the brass surface. The protective action of the arsenic films prevented any subsequent enrichment in copper of the active spots by dealloying, thereby eliminating potential sites for deposition of copper from solution. The small amount of arsenic found in commercial brasses precludes the formation of a continuous arsenical film over the entire surface (12, 55), hence, arsenic affects only dezincifica- tion attack, and has little influence on general dissolu- tion rates. More recently, West (56) has proposed that the effect of arsenic under certain conditions of pH may be to render the brass surface passive. According to this theory, arsenic in solution is stable as arsenite, As(OH)4 under the local slightly alkaline conditions present during oxy- gen reduction. These arsenite ions readily adsorb onto the positively charged metal surface, forming strong bonds with the copper atoms, and particularly with the zinc atoms, at the surface. The effect of this bonding is to stabilize the surface by rendering the surface atoms in an essentially passive state. West points out, however, that in the acid environments expected within pits, some other mechanism must be operative. Under these conditions the arsenyl ion, AsO+, is the stable dissolved arsenic species. This ion would possess little tendency to adsorb onto a positively charged metal surface. The theory that arsenic prevents redeposition of copper at the brass surface has been proposed in many forms over the years. Masing (57) found that the overpotential for hydrogen reduction on brass surfaces was less in solu- tions containing dissolved arsenic than in arsenic-free solutions. On the basis of his observation he proposed that for arsenical brasses undergoing corrosion, the cathodic copper reduction reaction was replaced by the hydrogen reduc- tion reaction, hence, no dezincification (i.e., no copper redeposition) occurred. More recent polarization data re- ported by Sugawara and Shimodaira (58) point out a possible flaw in Masing's experiments. They found that anodic and cathodic polarization of copper and copper-zinc alloys is influenced by additions of sodium arsenite to the test solu- tion (3% NaCl). However, arsenical brasses exposed in solu- tions without arsenic additions exhibited polarization data identical to those of the non-arsenical brasses. The dis- agreement between the two works is by no means conclusive. The detection of the influences of trace element additions to alloys by electrochemical polarization curves can be insensitive to very subtle effects (59). Nevertheless, similar indications that the source of arsenic, whether from the alloy or from the bulk solution, influences its inhib- itive effects have been reported by other investigators (27, 60, 61). Hollomon and Wulff (60) have presented the only experimental data concerning the identity of the black or gray films normally found on the surfaces of corroded ar- senical brasses. Their film-structure experiments consisted of immersing samples of copper and Cu30Zn containing up to about 0.5 w/o of arsenic in 2N and 5N nitric acid and in 2N and 5N hydrochloric acid, followed by examination of the corroded surfaces by electron diffraction. Arsenious oxide, As203, films were observed to form on arsenical copper and Cu30Zn in all solutions tested. Significantly, elemental copper was also found on the Cu30Zn samples. These results were interpreted in terms of a dissolution and replating mechanism. As the brass dissolves, redeposition of copper and, later, of arsenic takes place. The arsenic deposits are subsequently oxidized to arsenious oxide. The detection of copper on the arsenical brass sur- faces in this investigation has significant implications, namely: either arsenic additions have limited effects in inhibiting redeposition of copper on Cu30Zn (contrary to experience); or dezincification of brass by a selective leaching mechanism may proceed even when arsenic is present. Hollomon and Wulff later extended their investiga- tions, exposing arsenical and non-arsenical Cu30Zn samples in a series of environments designed to produce dissolution of the copper component of the brass to either cuprous or to cupric ions. They concluded that non-arsenical alpha brass undergoes simultaneous dissolution and that dezincifi- cation occurs by the reduction of cuprous ions to copper on the brass surface. In the case of arsenical brass, the arsenic dissolved from the alloy is subsequently deposited on the brass surface. The arsenical film thus formed increases the effective electrode potential of the brass surface such that reduction of cuprous ions to copper is impossible. Lucey (27) has also stated that arsenic interferes with the copper deposition step of dezincification. Accord- ing to this author's mechanism, copper deposition at the advancing corrosion interface proceeds by the disproportion- ation of cuprous ions to form cupric ions and copper, 2Cu Cu + Cu. (9) When arsenic is present at the interface, the dispropor- tionation reaction is suppressed by the following pair of reactions: 3Cu+2 + As 3Cu+ + As+3 (10) +3 + 3Cu + As+3 3Cu+ + As (10a) A noteworthy feature of this mechanism is that the arsenic +3 is recycled between the As and metallic states. Conse- quently, only a very small amount of arsenic is required for protection. This is consistent with the fact that only traces of arsenic are required to protect alpha brasses against dezincification. In summary, it can be said that the mechanisms pro- posed to explain arsenic's role in the inhibition of dezinci- fication have been influenced by the investigators' choices for the detailed dezincification mechanism. Thus, it is clear that complete clarification of the inhibition phenom- enon awaits final determination of the mechanisms) of dezincification. CHAPTER 3 EXPERIMENTAL PROCEDURE Characterization of the metals used in this investi- gation and the details of sample preparation are discussed in Appendix 1. Details of solution compositions and prepa- ration are given in Appendix 2. The standard polarization cell as used in potentiokinetic testing, and its modifica- tion for subsequent tests, are described in Appendix 3. A list of the equipment used in these investigations is given in Appendix 4. 3.1. Potentiokinetic Polarization Experiments Potentiokinetic polarization tests on copper, zinc, and Cu30Zn were conducted in the standard polarization cell, described in Appendix 3. Details of the potentiokinetic technique and theory have been presented elsewhere (62, 63). A schematic of the experimental set-up is presented in fig- ure 3. Each test was conducted in a fresh solution of about 500 ml volume. Samples were introduced into the cell only after the solution was thoroughly purged with nitrogen. CIII ENT NS ITY Figure 3. Schematic of the equipment set-up for generation of potentiokinetic polarization curves. Equilibration of the sample corrosion potential was monitored with an electrometer. Zinc samples achieved stable poten- tials within a few minutes. Copper and Cu30Zn samples, how- ever, were sluggish, exhibiting changes in potential after several hours. It was soon discovered that a prepolarization treatment, consisting of a 5-minute cathodic polarization to -1.000 VSCE, resulted in much faster attainment of steady corrosion potentials. A sample was considered to have achieved a steady corrosion potential when its potential changed less than about 30 mV in 30 minutes. Polarization curves were obtained by scanning the sample's electrode potential from low cathodic overpotentials to the appropriate anodic overpotential, both in the forward (i.e, active-to-noble potential) and in the reverse (noble- to-active) directions. The curves were plotted on potential versus log current coordinates on an X-Y recorder. The effect of scan rate on the polarization curves was investi- gated for copper by comparing curves obtained at scan rates varying from 10 to 70 mV/min with step scans made in steps of 50 mV every 5 min (64). As a result of these comparisons, 1 V/hour (16.67 mV/min) was chosen as the optimum scan rate consistent with reproducible characteristic potential fea- tures on the polarization curves. 3.2. Selective Leaching Experiments Potentiostatic selective leaching experiments were carried out in the modified standard polarization cell, as described in Appendix 3. All experiments were on polished Cu30Zn samples immersed in 0.1M chloride solutions of pH = 4.0+0.1. Cells containing 500 ml of solution were brought up to temperature and purged with purified nitrogen for 12 hours prior to introduction of the samples. The potentiostat was switched on simultaneously with the intro- duction of the sample. Cell temperature was monitored throughout the tests. A constant temperature bath was employed to maintain cell temperature at 89+0C throughout the tests. Cell potential and cell current were noted frequently. Solution aliquots of 25 ml volume were pipetted from each of the 890C cells before introduction of the sample and at approximately 24-hour intervals during the test. The solution drawn off was replaced with 25 ml aliquots of fresh, nitrogen saturated solution. Solution samples were submitted to atomic absorption spectrophometric analysis for dis- solved copper and zinc. Details of the analytical procedure have been presented elsewhere (32). The experimental limits of detection were 0.10 ppm for copper and 0.05 ppm for zinc. At the conclusion of each test, the potentiostat was switched off as the sample was removed from the cell. Each sample was rinsed in a stream of distilled and deionized water and dried with a blast of compressed dusting gas. The sample connecting wire was ripped away, and the stop- off lacquer was peeled off. Each sample was inspected scrupulously for evidence of leaks in the stop-off lacquer. These were usually easily recognized as patches of deposited copper on the sides or back of the sample. Any indication of leaks was cause for rejection of the experiment. The exposed (projected) surface area of each sample was measured by direct comparison with a celluloid sheet inscribed with a 15 mm x 15 mm grid. The accuracy of these 2 measurements was about +0.005 cm Most samples were care- fully sectioned with a jeweler's saw across a diameter. Half of the sample was submitted to Auger analysis; the other half examined by scanning electron microscopy (SEM). Solution pH was checked before and after each ex- periment, and was found in every case not to have changed. Final solution volume was measured in order to correct atomic absorption data for evaporation losses. 3.3. Controlled Anodic Dissolution Experiments Potentiostatic anodic dissolution experiments were conducted in the standard polarization cell, modified as described in Appendix 3. Experiments were carried out on Cu30Zn and Arsenical Cu30Zn specimens in pH 4, 0.1M chloride solutions of 600 ml volume. Cell temperature was maintained by a constant temperature bath at 30+1"C. Details of cell operation are as described for the selective leaching experiments, section 3.2. At the conclusion of tests, samples were sectioned across diameters. One half was mounted and polished for metallographic examination; the other half was examined by SEM. Analysis of solution samples was accomplished through atomic absorption analysis. The experimental lower limit of detection for arsenic was 0.01 ppm. 3.4. Artificial Occluded Cell Experiments The artificial occluded cell was designed to simu- late the restricted flow and electrochemical conditions ad- jacent to the internal dissolving brass surface of a partial- ly dezincified sample. The electrode assembly, detailed in figure 4, con- sisted of a Cu30Zn sample positioned horizontally to, and shielded from the bulk solution by, a porous copper screen. The sample holder was constructed from three pieces of 0.0625 inch thick polycarbonate sheet, bonded together with epoxy. A length of 0.7 mm inside diameter "spaghetti" tube, inserted through a slot cut in the middle sheet, connected the 1 cm2 hole between the Cu30Zn and the copper screen with the top of the assembly. The entire sample assembly was in- serted in a slot in the rubber stopper. The end of the CONNECTING WIRES Cu 30 Zn SAMPLE -- STANDARD CALOMEL ELECTRODE CAP -RUBBER STOPPER COPPER SCREEN SPAGHETTI TUBE Figure 4. Artificial occluded cell electrode assembly. spaghetti tube was inserted into the bottom of a standard calomel electrode storage cap, which was bonded to the top of the assembly. The copper screen was fabricated from a 18 mm x 18 mm, number 100 mesh bronze, copper-plated in a saturated CuSO4 bath to a total thickness of about 0.6 mm. The plati- num auxilliary electrode, Haber-Luggin probe, gas disperser, and gas outlet, all similar to those used on the standard polarization cell, were inserted through holes in the rub- ber stopper to complete the assembly. The cell container consisted of a glass jar. A saturated calomel reference electrode,SCE 2,was connected to the Haber-Luggin probe through a salt bridge of the test solution. Before each experiment, a fresh Cu30Zn specimen and copper screen were carefully attached to opposite sides of the 1 cm2 hole with a quantity of Miccroflex stop-off lacquer. The connecting wires were fed through glass tubes set in the stopper. The volume of the occluded cell cavity between the two electrodes was approximately 0.154 ml. All experiments were conducted potentiostatically at room temperature in approximately 300 ml of stirred, nitrogen- saturated, 0.lM chloride solutions of pH = 4. The assembled cell was purged with nitrogen prior to introduction of the solution and during the test. A saturated calomel electrode, SCE 1, was inserted into the electrode cap at the top of the cell. The occluded cell cavity, spaghetti tube, and electrode cap were then filled by drawing solution up from the cell with a hypodermic syringe and needle inserted through the side of the electrode cap. The Cu30Zn sample and copper screen were connected through a zero-resistance ammeter (53). The electrode assembly was connected to the working electrode terminal of the potentiostat. The bulk cell potential, EB, was monitored by an electrometer connected between the working electrode assembly and SCE 2. The occuded cell potential, EC, was similarly monitored by a second electrometer con- nected between the working electrode assembly and SCE 2. The occluded cell current, i passing between the Cu30Zn sample and the copper screen was measured by the zero- resistance ammeter. A schematic diagram of the electrical circuit is given in figure 5. When the potentiostat was switched on, the occluded cell current and occluded cell potential were recorded on identical strip-chart recorders. 3.5. Auger Electron Spectroscopic Analyses Auger electron spectroscopy (AES), coupled with argon sputtering,was employed to determine the thickness of the dezincified layers formed on the Cu30Zn samples subjected to selective leaching as described in section 3.2. The theory of AES analysis and of the argon sputtering process is discussed in detail in Appendix 5. AMMETER Cu30Zn Cu SCREEN Figure 5. Schematic of the electrical circuit for the artificial occluded cell experiments. The AES system in the Department of Materials Science and Engineering of the University of Florida is employed routinely in the determination of glass surface compositions and in the measurement of near-surface com- positional profiles of metals, glasses, and composites. The system consists of a high vacuum chamber and accompanying pumping systems, an electron gun, a cylindrical-mirror energy analyzer, a sputter-ion gun, and accompanying control and data treatment and acquisition systems. A schematic diagram of the AES system is presented in figure 6. As many as six samples were attached (by copper clips) to the externally manipulated carrousel sample holder. The vacuum chamber was evacuated to a base pressure no higher -8 than 5x10 atm. Auger surface composition measurements were made with the electron gun operating at 3 keV. The electron beam diameter was about 0.5 mm, and the beam current about 25 pa. The angle of incidence was about 45 degrees from normals to the sample surface. Energies from zero to 1800 eV were normally scanned, however, only the portions of interest were scanned as generation of the composition pro- files proceeded. A typical spectrum obtained from the sur- face of a Cu30Zn sample is shown in figure 7. Composition profiles were obtained by successive removal of layers of metal by argon sputtering and recording the Auger spectra of freshly exposed surfaces. Argon was leaked into the evacuated chamber until a pressure of 45 X-Y Recorder or Oscilloscop, Lock-In Electron p er Gun S Sweep f ____ Supply Carrousel Target / S Holder Electron \ Electron Gun n Multiplier y-,-, SMagnetic Shield Sputter Ion Gun Figure 6. Schematic of the AES system used to analyze selective leaching samples. oo E 0 -o o 4- 0 O o a 0 ru 0 t-" O Z 0 0 '4- LiL 01 -, U a) o - lo CM. I.- o ^L 0 - -5 5-6x10 atm was reached. The sputter-ion gun was then switched on and the Argon sputtering begun. The area of the surface affected by the sputtering was about 5 mm in diameter. The electron gun and the sputter ion gun were aligned so that the incident electron beam was concentric with the sputtering crater. The sputter-ion gun was switched off while the Auger spectra were measured. The peak-to-peak heights measured for each element of interest were plotted as a function of sputtering time to produce the composition profiles. Calibration of the Auger peak-to-peak heights with absolute concentrations was not attempted in this investi- gation. The primary reason for this was the uncertainty in measurement introduced by the argon sputtering process. In particular, the effects of differential sputtering rates of copper and zinc from the brass surface on the measured peak-to-peak heights are unknown. Also, smearing or averag- ing of the surface due to sputtering is likely. Secondly, the introduction of excess vacancies or porosity into the near surface layers of brass samples by the selective leach- ing process could lead to false results. Consequently, the ratio of the peak-to-peak heights for copper and zinc (Cu/Zn ratio) was introduced as a qualitative indicator of the level of selective leaching determined at each depth from the sample surface. An indication of the peak-to-peak ratios to be ex- pected from surface layers leached of part of their zinc contents can be seen in figure 8. The data points on this plot represent the Cu/Zn ratios measured from binary copper- zinc alloys having 90, 85, 80, and 70 w/o of copper. Two important points are to be noted from the curve. (1) The Cu/Zn ratio measured for the Cu30Zn alloy has a value of about 5. (2) A small change in alloy composition away from 30 w/o of zinc towards pure copper is accompanied by a signi- ficant increase in the measured Cu/Zn ratio. Referring to figure 9, the composition profile re- corded for the Cu-15 w/o Zn alloy, one can see that a steady Cu/Zn ratio is reached after about four minutes of sputter- ing. The initial period of sputtering, before the ratio becomes constant, corresponds to a "cleaning" of the surface of adsorbed gases, fingerprints, etc. which are the inevita- ble results of handling samples outside the vacuum chamber. This "steady state" Cu/Zn ratio reflects any effects which may be present due to differential sputtering rates. Calibration of the argon sputtering rate was under- taken in order to assign numerical values to the depth axes of the composition profiles obtained from selectively leached Cu30Zn samples. Details of the calibration procedure are given in Appendix 6. 20 15 * 10 0 N 5 0 WEIGHT PERCENT COPPER Figure 8. Effect of the composition of Cu-Zn alloys on the Cu/Zn ratio obtained by AES analysis. I I OlIv O uz/no -) 0 o 1HO13H >iV3d-o:-AV3d 3.6. X-Ray Identification of Corrosion Products Solid corrosion products were examined by the X-ray powder diffraction method (65). Standard Debye-Scherrer powder cameras of 57.3 mm of 114.6 mm diameter were employed. Nickel-filtered copper radiation or manganese-filtered iron radiation was used in all cases. Product identification was made by the comparison of d-spacings derived from the X-ray films with those listed in the ASTM Powder Diffrac- tion Files. Identification of some complex hydroxy-chlorides of copper was simplified considerably by reference to the d-spacings published for these compounds by Feitknecht and Maget (66). CHAPTER 4 RESULTS AND DISCUSSION 4.1. Electrochemical Characterization of Copper, Zinc, and Cu30Zn Electrochemical characterization of copper, zinc, and the Cu3OZn alloy in deaerated O.1M chloride solutions was undertaken in order to establish a firm basis for pre- dicting the experimental conditions which lead to dealloy- ing behavior in the system. Further, the effects of these experimental conditions on the choice of dealloying mecha- nisms were explored. Potentiokinetic polarization experiments described in section 3.1., were employed throughout this phase of the investigation. Wherever possible, experimental polari- zation results were compared with calculated equilibria. Identification of corrosion products was accomplished through X-ray powder diffraction experiments. 4.1.1. Polarization of Copper in 0.1M Chloride Solutions Cyclic polarization curves were generated for pure copper samples immersed in a series of room temperature (2530C), nitrogen-saturated, 0.1M chloride solutions having pH values between 1 and 13. The characteristic features of the polarization curves are summarized in figure 10 as an experimental potential versus pH diagram, after the method of Pourbaix (67). The general regions of immunity, passi- vity, and corrosion are so marked. A similar diagram has been published (59). The diagram can be divided into two parts according to the change in the corrosion behavior of the system from general corrosion to active-passive at pH 8. In solutions of pH between 3 and 8, copper exhibited pH-independent polarization behavior. Its corrosion potential was about -0.260 VSCE in this pH range. General corrosion behavior was noted at more noble (higher) potentials. A maximum in anodic current density (- 5 ma/cm ) occurred at about 0.060 VSCE. The formation of a film of white cuprous chlo- ride, CuCl, on the sample surface preceded the current maximum (68, 69). At electrode potentials more noble than about 0.100 VSCE, increases in corrosion current with po- tential were again noted. If the polarization scan was reversed at this point, the current fell with decreasing electrode potential until the protection potential (-0.060 VSCE) was reached, whereupon the current changed polarity to cathodic. As the potential was decreased Strictly speaking, the protection potential has no signifi- cance under active corrosion conditions. LEGEND o corrosion potential lAa4passivation potential a rupture potential +protection potential *A -a A N A 4. 0__ 0 , 4 _-- ----- ---,-- --+,-- -- ---L -, * A * -.2 - -.3 N IMMUNITY -7' ,111 ~1-s ' 0 I I I I I I 2 3 4 5 6 7 8 9 10 II 12 13 14 pH Experimental potential versus pH diagram for pure copper in nitrogen saturated, 0.1M chloride solutions. L 2 (.) 0 U S .1 _1J I- 0 _ --.i Figure 10. - "'a -.5 -.6- further, the CuCl on the copper surface was reduced to copper. In solutions of pH > 8, copper exhibited pH-dependent corrosion potentials and active-passive polarization behav- ior (70). When polarized anodically, the copper became covered by a tarnish film, identified (by X-ray diffraction) as Cu20, and the current density began to decrease at poten- tials about 90 mV above the corrosion potential. One or more current maximum features was observed as the electrode potential was subsequently increased. Such maxima are label- led in figure 10 as passivation potentials. Breakdown of passivity occurred when the rupture potential was exceeded. Initial breakdown of the passive film occurred at small, apparently random, spots on the surface. The spots quickly became covered with voluminous corrosion products identified variously as different forms of copper-hydroxy-chloride com- bined with CuCl. The action of stirring sometimes dislodged the products, causing them to smear over the sample face. This action resulted in the breakdown of passivity on the copper surface beneath the product. The likely mechanism involved is acidification of the solution held between the hydroxy-chloride corrosion product and the passive metal sur- face resulting from copper dissolution, Cu + 2H20 Cu(OH)2 + 2H + 2e" (1a) The resulting pH decrease, in turn promotes dissolution of the Cu20 passive film, Cu20 + 2H+ + 4C1- 2CuC12 + H20 + 2e- (llb) In figure 11, the polarization data for copper in 0.1M chloride solutions are superimposed over the equilib- rium potential versus pH diagram calculated for the Cu-O.1M Cl- H20 system at 250C. Similar calculations have been presented elsewhere (18, 71). A list of the free energy data employed in these and later equilibrium calcu- lations can be found in Appendix 7. Excellent agreement is noted between the experimentally measured corrosion poten- tials and the calculated lines corresponding to the equili- brium between copper and the CuCl2 ion (ion activity = 106) and between copper and cuprite, Cu20. The measured value of the protection potential coincides well with the calculated equilibrium between copper and cuprous chloride. The locus of the rupture potentials for solutions of pH < 9 indicates that the current increases observed at more noble potentials are due to oxidation of cuprous ions at the sample surface to cupric ions (69). It is significant that the protection potential has the same value in both acid and alkaline solutions up to pH 11. This indicates that the Cu-CuCl equilibrium is es- tablished under localized corrosion conditions (47). The Uj T- I- 0i a. z* >- *_ UJ 2 3 4 5 6 7 8 9 10 I1 12 13 pH Figure 11. Superposition of polarization data for pure copper in 0.1M chloride solutions and the equilibrium potential versus pH diagram calculated for the system Cu-O.1M C1--H20 at 25C, after Van Muylder et al. (71). equilibrium was at least partially confirmed by the detection (by X-ray diffraction) of CuC1 in the corrosion products. 4.1.2. Polarization of Zinc in 0.1M Chloride Solutions Cyclic polarization curves were generated for pure zinc at room temperature in stirred, nitrogen-saturated 0.1M chloride solutions having pH values between 2 and 14. The data are plotted on potential versus pH coordinates in figure 12. Apparent from the shape of the diagram is the fact that zinc exhibits amphoteric behavior, undergoing general corrosion in both acid solutions and in highly basic solutions. Equally obvious is that zinc corrodes freely in deoxygenated solutions, its corrosion mixed potentials fall- ing well below the domain of stability for water (with re- spect to the H20/H2 equilibrium, line a). The region of passivity between pH values of about 7.5 and 13 is virtually insignificant since the pitting potentials also fall almost entirely below the lower limit of stability for water. When zinc samples are polarized above these potentials, severe pitting occurs. No regions of complex ion formation were suggested by the polarization data. The voluminous white corrosion products resulting from pitting attack could not be identified by X-ray dif- fraction but are assumed to be amorphous zinc hydroxides as predicted by previous calculations (72). -.7 Lj ui1 0 a- i- (n, .0 I.I -1.2 -1.3 1.4 -1.7' I LEGEND So corrosion potentlol . AD passivation potential rupujre po;enterriol \ t protection potentrol \ PITTING NN* * IMMUNITY PTTNG IM UNITY ot 0x I I I 2 3 4 5 6 7 8 9 10 11 12 13 14 pH Figure 12. Experimental potential versus pH diagram for pure zinc in nitrogen-saturated 0.1M chloride solutions. 3 -------- ---- -- --- ----- -. -- -------1...- .. ---- ----- __ -- --~,,. ,ii- .,~.,,i,,,.,. ,-,, -- --- -- -1.5 In figure 13, the experimental polarization data for pure zinc in 0.1M chloride solutions are superimposed over the stable equilibria calculated for the Zn-H20 sys- tem at 25'C, after Pourbaix (72). Again, excellent agree- ment is found between calculated and experimental behavior. It can be concluded from these data that there is a very large tendency for zinc atoms to pass into solution at all electrode potentials more noble than about -1.100 VSCE. 4.1.3. Polarization of Cu30Zn in 0.1M Chloride Solutions Cyclic polarization curves were generated for the Cu30Zn alloy in nitrogen-saturated 0.1M chloride solutions at room temperature and at 890C. The characteristic elec- trode potential features of the polarization curves gener- ated at room temperature are presented in figure 14. With a few exceptions, these results are very simi- lar to those obtained for pure copper. The locus of corro- sion potentials between pH 4 and pH 8 is -0.260 VSCE, as it was for pure copper. There is, however, a sharp drop in measured corrosion potentials between pH 3 and pH 4 to a steady value of about -0.450 VSCE at lower pH's. The drop- off in corrosion potential was confirmed by exposing two Cu30Zn samples in nitrogen-saturated pH 2 and pH 4 solutions for 34 days and 50 days, respectively. The variation in measured corrosion potential was recorded as a function of -1.8 3 Figure 13. 4 5 6 7 8 9 10 11 12 13 14 15 pH Superposition of polarization data for pure zinc in 0.1M chloride solutions and the equilibrium potential versus pH diagram calculated for the system Zn-H20 at 250C, after Pourbaix (72). LEGEND DN ocorrosion potential \ ADspoassivation potential rupture potential N +protection potential CORROSION 0 00 A : S .----a--- -- a- 4 *+ *- , --- + + o B. 6 N. 8 0 I N\ S *O * N1 N * N N N< PASSIVITY a 0 0 o IMMUNITY N I I I 2 3 4 5 6 7 pH 8 9 10 11 12 13 Experimental potential versus pH diagram for Cu30Zn in nitrogen-saturated 0.1M chloride solutions. Li .1 -J z i~ Figure 14. j L -.5 -.6 -.7L time, figure 15. The value of the corrosion potential mea- sured in solutions of pH < 3 can be explained by considering the equilibrium between copper and the CuC13 ion: -2 - Cu + 3C1 CuC13 + e- (12) Using the free energy data compiled in Appendix 7, the equi- librium potential for reaction (12) at 25C is given by: a ucl -2 aCuC1d2 3 E = -.011 + .0591 log 3 .(13) (aCl-) Assuming aul-2 = 106, the equilibrium potential of the -2 Cu-CuCl1 reaction is -0.427 VSE. This value is in fair 3 SCE agreement with the experimental corrosion potential value. Indications of the establishment of this equilibrium on cop- per in acid chloride solutions have been reported elsewhere (73). The locus of the protection potential determined by cyclic polarization of Cu30Zn samples is -0.070 VSCE, vir- tually the same as that found for copper. The protection potential takes on a new significance in the case of the brass (48), since when the electrode potential of the cuprous chloride-covered brass sample was brought below the protec- tion potential, deposition of copper occurred. This was found to be the case not only in acid and neutral solutions, but also in basic solutions when the rupture potential was 64 -.20- Cu30Zn S CI-3 = 10- -.30 - ui UU c pH =4 u0 o ii > o o^ t1 20 30 40 50 _E -.40 day I- W -.60 10 20 30 40 50 TIME, days Figure 15. Variation of the corrosion potential of Cu30Zn with time of exposure in 0.1M chloride solutions of pH 2 and pH 4. exceeded and copper-hydroxy-chloride corrosion products were formed over part of the surface. Upon passing through the protection potential, deposition of copper took place on the brass surface beneath the corrosion products, but not on the areas of the surface still remaining passive. If the rup- ture potential was not exceeded during the polarization scan, then no deposition of copper was detected when the sample electrode potential was brought below the protection poten- tial. In the context of producing dezincification of the Cu30Zn alloy, the breakdown of passive behavior can be thought of as an activation step. The concept of activation has received considerable attention in connection with the initiation of pitting on Fe-Cr alloys (53, 62, 74). Van Muylder et al. (47) have described the initiation of pitting of copper in aqueous chloride solutions by the introduction of oxygen at local sites on copper surface. This action of the oxygen in destroying passivity was attributed to its ability to increase the electrode potential of the copper surface above its critical potential. A limited number of cyclic polarization experiments were conducted on Cu30Zn in 0.1M chloride solutions at 890C. Through these experiments the corrosion potential of Cu30Zn at 890C in solutions of pH 4 was established at -0.580 VSCE. 4.2. Dealloying of Cu30Zn by Selective Leaching 4.2.1. Conditions of Exposure The electrochemical requirements for operation of a selective leaching dealloying mechanism were outlined in section 2.1. According to the proposal of Verink and Hei- dersbach (18), selective leaching of an alloy's active com- ponent may be expected when the alloy is polarized to elec- trode potentials between its corrosion potential and the corrosion potential of its active component. In the previous section (section 4.1.), the elec- trochemical characterization of copper, zinc, and Cu30Zn was undertaken in order to establish the domain of electrode potentials in which selective leaching of the Cu30Zn alloy could be studied. Referring to figures 12 and 14, this re- gion is bounded in room temperature neutral and weakly acidic solutions by the corrosion potential of Cu30Zn, -0.380 VSCE, and by the corrosion mixed potential of zinc, -1.100 VSCE. At 890C, the upper boundary is displaced to a somewhat lower value, -0.480 VSCE. Accordingly, potentiostatic selective leaching experi- ments, described in section 3.2., were conducted on samples of Cu30Zn exposed in pH 4, 0.1M chloride solutions at 89i0iC. Table 1 is a list of the applied electrode potentials and times of exposure for the samples reported. Note that three samples Table 1 Conditions of Exposure for the Cu30Zn Selective Leaching Samples Electrode Potential Time volts (S.C.E.) days -0.450 44 -0.450 1 -0.450 7 -0.500 4 -0.600 4 -0.700 2 -0.700 4 -0.700 (2 samples) 7 -0.700 10 -0.800 4 -0.900 4 were exposed at -0.450 VSCE, slightly above the measured corrosion potential of the Cu30Zn alloy. It is expected that these samples would show both the effects of anodic (simultaneous) dissolution at low rates and of selective leaching. 4.2.2. Results of Solution Analysis Atomic absorption spectroscopy was employed to ana- lyze the copper and zinc contents of solution samples taken from the cells before the start of and during each experi- ment. As stated previously, the experimental limits of detection were approximately 0.10 ppm for copper and 0.05 ppm for zinc. Smaller amounts, or traces of copper and zinc could be detected, but the accuracy of measurement of con- centrations below the limits of detection is questionable. For the conditions of exposure outlined in table 1, no detectable amounts of copper are expected to enter solu- tion. Such was found to be the case for all experiments reported. Whenever significant quantities of copper were detected in solution, failure of the stop-off lacquer to protect unexposed portions of the sample was determined to be the cause, and the experiment was rejected. Calculation of the amount of zinc leached from each Cu30Zn sample was made from measurements, by atomic absorp- tion analysis, of the zinc content of the solution during the course of the experiment. Corrections were made for changes in the solution volume due to evaporation and for the quantities of zinc removed from the cells in aliquots taken for analysis. The total amount of zinc dissolved at any time during an experiment is reported in micrograms per square centimeter of sample surface (projected) area (pg/cm2). Cal- culations were completed for all samples listed in table 1, except those exposed at -0.450 VSCE. Figure 16 summarizes the zinc dissolution data cal- culated for the five samples exposed at -0.700 VSCE. Con- siderable variation in the rates of zinc dissolution from sample to sample is evident. However, the data denote a marked tendency towards linear zinc dissolution kinetics. Figure 17 summarizes the zinc dissolution data from samples exposed at potentials between -0.500 and -0.900 VSCE The data points plotted for -0.700 VSCE at each time of ex- posure represent the average of the data points shown in figure 15 for the five -0.700 VSCE samples. Again, linearity is evident in each of the plots, at least for four days of exposure. The rates of zinc dissolution measured at each electrode potential are summarized in Table 2. The third column gives the partial current density for the dissolution of zinc to its divalent ion from Cu30Zn, as calculated from the zinc dissolution rates. Pickering and Byrne (20) have presented data for dissolution currents of zinc from Cu30Zn exposed potentio- statically at 23C in an acetate-buffered 1N Na2SO4 solution 180- 160- NE 140 =120- w 100 - o m80 S60 40- 20 0 Figure 16. Cu 30 Zn T=89C E= -0.200 VSE/-0.455SHE TIME, days Zinc dissolved from Cu30Zn samples potentio- stated at -0.700 VSCE in 890C, 0.1M chloride solutions of pH 4. 100 90 Cu 30Zn, T=89C ; E= -0.500 VCE SCE 0 E =-0.600 A E=-0.700 0 E = -0.800 0 E=-0.900 E S 60 0 S50- A O >, o 40 5 30- z 20 0 0 0 I 2 3 4 TIME, days Figure 17. Zinc dissolved from Cu30Zn samples potentio- stated at electrode potentials between -0.500 and -0.900 VSCE at 890C. Table 2 Rate of Zinc Dissolution from Cu30Zn Under Selective Leaching Conditions Electrode Potential volts (S.C.E.) Rate of Zinc Dissolution g/cm2/day Zinc Dissolution Current Density a/cm2 -0.500 3.5 1.2 x 10-7 -0.600 14 4.8 x 10-7 -7 -0.700 22 7.5 x 107 -7 -0.800 7.0 2.4 x 107 -0.900 3.3 1.1 x 10-7 -0.900 3.3 I.I x I0 of pH 5. In the region of electrode potential between -0.600 and -0.200 VSHE, they report zinc dissolution current values 8 7 2 between 7 x 10 and 2 x 10 a/cm. These current values are in qualitative agreement with those measured in the pres- ent investigation for selective leaching of zinc from Cu30Zn at 890C. 4.2.3. Surface Morphology Visual examination of the Cu30Zn samples after expo- sure under selective leaching conditions indicated that de- alloying had occurred in all cases. The unmasked portions of the sample surfaces exhibited the pink or reddish color characteristic of dezincified brass. For the two series of samples exposed for different periods of time at -0.450 and -0.700 VSCE, a definite deepening of the surface color with increasing times of exposure was noted. In all cases, the masked surfaces of the samples showed no detectable change in color from the characteristic yellow color of the Cu30Zn alloy. Further examination of the surfaces of the selec- tively leached Cu30Zn samples was accomplished by scanning electron microscopy. Cu30Zn samples were polished through 1 micron diamond paste before being subjected to selective leaching. In this way, any gross surface effects resulting from the selective leaching process could be easily recog- nized. Figure 18 shows a typical area on the exposed sur- face a Cu30Zn sample potentiostated 4 days at -0.450 VSCE. It is apparent from this photograph that the sample has undergone anodic dissolution, as was expected for samples potentiostated above the corrosion potential measured for the Cu30Zn alloy. The anodic dissolution appears to have taken place primarily in pits, which tend to lie on polish- ing scratches. Similar surface appearance was noted for the Cu30Zn sample exposed 7 days at the same electrode poten- tial. Figure 19 shows an area from the surface of a Cu3OZn sample exposed 4 days at -0.600 VSCE. The surface was fea- tureless and could not be distinguished from the unexposed surface, except by the fact that it developed a pinkish color. Figure 19 exemplifies the appearance of Cu30Zn surfaces after selective leaching at electrode potentials below -0.450 VSCE. It should be noted that the resolution in figure 19 is better than 0.2 vm. Surface roughening has been cited as a factor con- tributing to selective leaching in copper-gold alloys (16, 21, 31), and predicted as a possible factor in the case of copper-zinc alloys (16). In agreement with the results of Pickering and Byrne (20), no evidence for surface roughen- ing of Cu30Zn exposed under conditions expected to produce selective leaching has been found in this investigation. Only in the case of the samples exposed under slightly anodic -.., i- f I ,i"r'^ < - -' i- q-"-- "* '- W w . r- OI I5 b U" - -0.450 V C 3700X -0.450 V SCE 3700X I.i H. r~i Figure 19. Typical surface of Cu30Zn samples potentio- stated below -0.450 VSCE. From a Cu30Zn sample exposed 4 days at -0.600 VSCE. 3500X I ,: ,, ,:,, ..!ii t * A f conditions (at -0.450 VSCE) were any changes in surface appearance noted. 4.2.4. Auger Composition Profiles Composition profiles of the near-surface layers of the dealloyed samples listed in table 1 were obtained through Auger electron spectroscopic analysis coupled with argon sputtering, described in section 3.5. The present study describes the first reported use of Auger electron spectroscopy in the characterization of dealloyed structures. Figure 20 is a composition profile of an area on the unexposed surface of a Cu30Zn sample which has undergone selective leaching. The dimensions on the abscissa of the profile are based on the measured argon sputtering rate of O 25 Angstroms/min, determined as in Appendix 6. The sharp 0 increase in the copper peak-to-peak heights below 100 Ang- stroms is characteristic of all the profiles measured during the investigation. This initial period of sputtering can be assigned to cleaning of the metal surface of gases and oil adsorbed on the surface during handling. After the first few minutes of sputtering the "cleaned" metal surface is un- covered and Auger electron spectra characteristic of the metal structure are measured. As stated previously, the ratio of the peak-to-peak heights of copper and zinc serves as an indication of the relative level of enrichment of the measured surface in these UNEXPOSED Cu 30 Zn 0 200 400 600 I- -10 E:o 5 N Cu/Zn -5 Zn 0 I 0 800 1000 1200 DEPTH FROM SURFACE, Angstroms Figure 20. Auger composition profile obtained from an unexposed Cu30Zn sample. 100 80 60 40 20. two components. The relationship between the composition of a brass surface uncovered during sputtering and its measured peak-to-peak ratio was presented in figure 8, in section 3.5. A measured Cu/Zn ratio % 5 is indicative of a brass surface containing 30 w/o zinc. A higher measured Cu/Zn ratio sig- nifies decreased zinc content, and a lower ratio would indi- cate enrichment of the surface in zinc. Referring again to figure 20, a constant Cu/Zn ratio % 5 is reached after sput- 0 tering about 5 minutes (125 A). If the initial four minutes of sputtering is assigned to cleaning of the surface, then it is evident that no detectable enrichment of the Cu30Zn surface in either component has taken place. Figures 21a, b, and c are the composition profiles measured for Cu30Zn samples potentiostated at -0.450 VSCE for 1, 41, and 7 days, respectively. In each case, deple- tion of the zinc content of the near-surface layers is evi- denced by initially low peak-to-peak heights for zinc. With increasing depth from the sample surface, the amount of zinc measured increases until a constant value is reached. The peak-to-peak heights measured for copper remain relatively constant (after the initial "cleaning" period) with depth. In some cases, e.g., figure 21c, the copper peak-to-peak heights show a decrease after reaching maximum values. The cause of this effect is unclear, but is thought to be a re- sult of the argon-sputtering process. 0 0 E = 0.450 VSCE I day 200 400 600 800 DEPTH FROM SURFACE, Angstroms 100 80 I- I S60 w I 0 0 v 40 Li. a- 20 0 Figure 21a. Auger composition profile obtained from a Cu30Zn sample potentiostated at -0.450 VSCE for 1 day. 100 1000 15 o 0 I- N Ks 0 OILv uzI/n 0 O 0 I- 0 0 0 0 0 0 0 o9 1O O IH913R )lVt3d--01- )V3d C m. S- E 0 04- S- C o Lo .- 0 O 0 4) c4- o -o 4-U CUC 00 r-- 0 (A .- 4-) 0 <:. )c O E0 .c1 f0 OIIVUj uZ/"3 tn 0 0 N( N 00 - * O 0 0 "a E S N 4- N o 0 - 0 0 S4-) UJ U 3 (0 ( L .O E > 0 0 ( 0 S (A- "- E 00 0 S4- OWO Ni 0 - o s -c 0 0 0 o o o > u- r *~ u. I -E c * T (* 0 ( L Q. *r lU U 4-1 - 0 Q .,- 0 I- s 0 0 0 0 0 0 0 0 o 0 o I IH1D3H )iV3d -o )V3d The Cu/Zn ratio curves plotted on each profile indi- cate the level of dealloying at each point in the composi- tion versus depth profiles. Comparison of these curves in figures 21a, b, and c shows that the level of zinc depletion at the surface increases with increasing time of exposure. Similarly, the depths of the zinc-depleted surface layers, indicated by the attainment of constant values % 5 by the Cu/Zn ratio curves, increase with increasing time of expo- sure. Figure 22 and 23 are the composition profiles ob- tained from Cu30Zn samples exposed 4 days at -0.500 VSCE and -0.600 VSCE, respectively. Depletion of the zinc con- tent of the surface is evident, but appears to be less drastic than in the -0.450 VSCE sample exposed 44 days. Figures 24a, b, c, and d are the composition profiles obtained from Cu30Zn samples potentiostated at -0.700 VSCE for 2, 4, 7 and 10 days, respectively. Comparison of the Cu/Zn ratio curves for this series shows that the levels of zinc depletion at the surface and the depths of zinc-depleted layers fall roughly in order according to the times of expo- sure. Figure 25 and 26 are the composition profiles obtained from Cu30Zn samples potentiostated 4 days at -0.800 VSCE and -0.900 VSCE. The sample at -0.800 VSCE gives a profile similar to those measured for the previous samples. The -0.900 VSCE sample exhibits a radically different profile. 84 OIllVi uZ/n3 0 w 0 o 0 I, w .- it o o 0 0 0 0 S 3d 3d 1H913H >IV3d- 0 0 a4 0 E <0 t-r C Or 4-o o (0 E c '-o .au 0 .r-- .- a) o 0 a) o o Sr- O o Lg E, oe C, *i- cu i o * S- Q. C - 0 ( c0 0 CjN C oI.Lvj uz/no 0 0 CL E N O C) 0 CO O V) QJ oc r--o 4- -O .- 4- r-o 0 S.- 0O OC 4- 0 + 00 r 0 0 O C0 (_) *I +->, IH913H >4V3d-ol-)IV3d 0 0 0 0 0 co N 0 Olivaj uZ/In E N o 011 0 CS 4- c. 0 r-- 44 0 4-) LAJ 0 iC) 0) L. 0 .- Dr- O- S*r S-C Q. ) E +n 0 0 s.- a o 0 .t Q" og cj LL 1H913H NV3d -(-oIV3d |