CQRROSION OF MILD STEEL IN ALKALINE $EGUESTER3NG AGENT SOLUTIONS Theats for flu Degree 0? 911. D. E‘EECEEGAK STATE Efii‘é’ERS-TY 'l‘homas Richard Mulvaney 1962 This is to certify that the thesis entitled CORROSION OF MILD STEEL IN ALKALINE SEQUESTERING AGENT SOLUTIONS presented by Thomas Richard Mulvaney has been accepted towards fulfillment of the requirements for _Eh-_D_,_degree in Eggd Science Major prggssor U Date August 9’ 1962 0-169 LIBRARY Michigan State University ABSTRACT CORROSION 0F MILD STEEL IN ALKALINE SEQUESTERING AGENT SOLUTIONS By Thomas Richard Mulvaney The problem under investigation was the determination of the corro- sion inhibiting or accelerating effects of organic sequestering agents, one of the hydroxycarboxylic acid type and one of the aminopolycarboxylic acid type, relative to the effects of the more traditionally used inor- ganic phosphates and the other components of the system (water, alkali, and hardness-forming salts in water), under conditions similar to those commonly encountered in mechanical washers cleaning reusable glass con- tainers. Review of the literature has indicated a marked deficiency of information on this specific problem and in this general area. This presentation is divided into two sections. The first deals with the development of experimental procedures and the second with the corrosion investigations. Literature on corrosion testing procedures is reviewed and the results of preliminary tests and tests to obtain suit- able procedures for preparing mild steel specimens before exposure and for removal of corrosion products after exposure are included in the first section. The general method employed was that of laboratory total immersion corrosion tests, with the amount and intensity of corrosion determined by weight loss measurement and visual examinations. The sec- ond section dealing with the corrosion investigations is divided into five phases of study: (1) Materials and methods, (2) Distilled water- sequestering agent systems, (3) Distilled water-sodium hydroxide- sequestering agent systems, (4) Hard water-sodium hydroxide-sequestering agent systems, and (S) Distilled water and hard water studies combined and compared. Also included in some of these studies are the effects of sequestering agent concentration, temperature, surface-active agent, length of exposure, and successive exposures. Studies on materials and methods: Corrosion rates of cold rolled and Thomas Richard Mulvaney hot rolled AISI No. C 1008 steel specimens immersed in alkaline sequester- ing agent solutions did not differ significantly. The experimental pro- cedure developed for these investigations was found satisfactory for use by laboratory personnel having considerable differences in previous train- ing and experience, since results obtained with such personnel did not differ significantly. Studies on distilled water-sequestering agent systems: Distilled water corrosivity was reduced 97.6 percent with either trisodium phos- phate or tetrasodium pyrophosphate and about 75 percent with sodium gluc- onate (hydroxycarboxylate). Tetrasodium ethylenediaminetetraacetate (aminopolycarboxylate) caused a 12 percent increase in corrosion. Studies on distilled water-sodium hydroxideosequestering agent sys- tems: Alkaline solutions of ethylenediaminetetraacetate, gluconate, and pyrophosphate listed in order of decreasing corrosivity were each found significantly more corrosive than the sodium hydroxide control. More corrosion was found at 1700F than at 130°F with the control, gluconate, and ethylenediaminetetraacetate solutions. The addition of a surface- active agent to gluconate solutions did not significantly alter the solutionS’corrosivity. The amount of corrosion found with one, three, five, and seven day exposures did not differ significantly except with ethylenediaminetetraacetate. Studies on hard water-sodium hydroxide—sequestering agent systems: The gluconate solutions were markedly more corrosive than the control, which was itself significantly more corrosive than pyrophosphate, ethyl- enediaminetetraacetate and orthophosphate, which themselves did not differ significantly in corrosivity. Distilled water and hard water studies combined and compared: Dis- tilled water plus ethylenediaminetetraacetate was significantly more corrosive than distilled water, which was itself significantly more corro- sive than 12 grain per gallon hard water. Corrosion rates of alkaline hard water solutions of gluconate and of the control were significantly greater than their distilled water counterparts. Corrosion weight loss was found to increase slightly as the concentration of ethylenediamine- tetraacetate was increased from one-to-two-to-four moles of sequestrant per mole of alkaline earth metal cations in alkaline distilled water Thomas Richard Mulvaney solutions, but not in alkaline hard water solutions. The corrosion weight loss of mild steel specimens was found to increase linearly with gluconate concentration over the same range in alkaline distilled water solutions according to the following calculated predicting equation: ‘3?’ = 3.065 + 2.601X Corrosion weight loss with gluconate in alkaline hard water solutions over the same concentration range was found to decrease linearly as a function of concentration when the logarithm of weight loss was plotted versus the reciprocal of concentration, thus giving the following calcul- ated predicting equation: log i’ = 0.7782 + W X The apparent discrepancy between the effects of sodium gluconate concen- tration on corrosion in these distilled water and hard water solutions may be explained as follows: In the first case gluconate anions are free to sequester ferrous ions of the ferrous hydroxide protective film bf corrosion products, and as the concentration of gluconate is increased more ferrous ions can be sequestered. As the film is diminished in such a manner, less resistance is afforded to the diffusion of oxygen to the cathodes, and corrosion is thereby promoted. In the second case alka- line earth metal cations are present in solution and the gluconate anions would normally be expected to sequester them, thus forming a stable water soluble complex. However it is proposed that possibly the complex so formed is not adequately solubilized at the lower concen- tration and that intermingling of this complex with the corrosion prod- ucts could decrease their protectivity and result in increased corrosion. Addition of more gluconate could enhance the solubility of the complex leaving less to alter the corrosion products protectivity. As a result less corrosion might be found with increasing gluconate concentration. COPYRIGHT By Thomas Richard Mulvaney 1963 CORROSION OP MILD STEEL IN ALKALINE SEQUESTERING AGENT SOLUTIONS by Thomas Richard Hulvaney A THESIS Submitted to ‘Hichigan State university in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Food Science 1962 ACKNOULEDGMENTS The author wishes to express his sincere appreciation to Professor I. J. Pflug, the author's major professor, for providing challenging opportunities for professional development, and for suggestions, criticisms, and support during the author's graduate program. Appreciation is also extended to Professors w. D. Baten, J. L. Fairley, L. J. Bratzler, R. C. Nicholas, and L. L. Quill for constructive criticisms and suggestions during the preparation of this manuscript. Special recognition is given the American Bottlers of Carbonated Beverages Association for thbir interest in supporting research funda- mental to the industry and for the special assistance that made possible obtaining a background prior to the initiation of this research. Recog- nition is also given the united States Steel Corporation for their in- terest in the problem and for providing the steel specimens of known composition and history used in this investigation. Thanks go to Professor B. S. Schweigert, Chairman, Food Science Department, for his interest and support of this program; to Mr. Harris Gitlin for his sincere and most helpful suggestions and advice; to Mr. Gary Gaffield, Hr..Allen Stewart, and Hrs. Grace Mutz for assistance in the laboratory phases of this investigation; and to John and Anne Blaisdell for proofreading the final manuscript. ‘ Special thanks are given my wife, Betsey, for her tireless efforts during the program and in the preparation and typing of the manuscript. The sacrifices and understanding of my wife and children during the graduate program are recognized and appreciated. ******* ii TABLE OF CONTENTS Page INTRODUCTION..................................................... REVIEW OF LITERATURE............................................. Corrosion Costs............................................. Corrosion of Iron and Steel................................. \wawH Corrosion of Some Other Metals.............................. ,‘SEQUESTERING AGENTS AND THEIR CHEMICAL BEHAVIOR.................. l7 EXPERIMENTAL DESIGN AND ANALYSIS................................. 30 DEVELOPMENT OF EXPERIMENTAL PROCEDURE............................ 37 INTRODUCTION................................................ 37 REVIEW OF LITERATURE........................................ 37 General Corrosion Test Procedures...................... 37 Factors Important in Total Immersion Corrosion Tests... 41 Metal factors..................................... 41 Corrosion media................................... 46 Exposure conditions............................... 47 Assessment of corrosion effects on metal and media...........................................» 49 TEST RESULTS AND DISCUSSION................................. '54 Preliminary Total Immersion Corrosion Test Results..... 54 Tests to Develop an Acceptable Pickling Process........ 56 Tests to Develop an Acceptable Corrosion Products Removal ProcessOOOOOOOOOOOOOO000000000000000000000000 58 CORROSION STIJDIESOOOOOOOOOOOOOOOO00000000000000000000000000000000 62 THE EXPERIMENTAL PROCEDUREO00000000000000OOOOOOOOOOOOOOOOOOO 62 metal FactorSo O O O 0 0 O 0 O 0 O 0 O 0 0 0 O O O 0 0 0 0 O 0 0 0 0 0 O O 0 O 0 O 0 0 0 O O 0 0 62 Corrosion Media........................................ 64 Exposure conditionSU 0 0 0 O 0 0 0 0 0 O O O 0 0 O O 0 0 0 0 0 O o O O O 0 O O 0 0 O 0 0 0 66 Assessment of Corrosion Effects on Metal and Media..... 66 RESULTS AND ANALYSIS........................................ 69 Studies on Materials and Methods....................... 69 Studies on Distilled Water~Sequestering Agent Systems.. 79 Studies on Distilled Water-Sodium Hydroxide~ Sequestering Agent Systems........................... 85 iii CORROSION STUDIES continued . Page Studies on Hard WateraSodium Hydroxide-Sequestering Agent SYStemSOOOOOooooooooooooooooooooooooooooooooooo 105 Distilled Water and Hard Water Studies Combined """ and ComparedOOOOO0000000000OOOOOOOOOOOOOOODOOOOOOOOOO 120 DISCUSSION OF RESULTS....................................... 133 Studies on Materials and Methods....................... 133 Studies on Distilled Water-Sequestering Agent Systems.. 135 Studies on Distilled Water-Sodium Hydroxide- Sequestering Agent Systems........................... 142 Studies on Hard Water-Sodium Hydroxide-Sequestering Agent systemSOooooooooooooooooooooooooooooooooooooooo 149 Distilled Water and Hard Water Studies Combined and comparedouoooooOooooooooooooooooooooooooooooooooo 152 SUM-RY AND CONCLUS IONS Q 0 v 0 CI 0 O O O 0 O O 0 U U 0 0 O 0 0 O O U 0 O 0 O 0 0 0 0 0 0 0 O 0 0 0 O 0 0 O 1'59 REFERENCESOOOOOOOOOOOOO0°00.Q000OOUOOO0000000OOOOOOOOOOOOOOOOOOOO 162 Corrosion Studies Related References........................ 162 Experimental Procedure Related References................... 164 iv Table 10. 11. 12. 13. 14. 15. 16. 17. LIST OF TABLES Corrosiveness of some gluconate-caustic systems........... Solubilities of some calcium and magnesium salts in water. Sequestering power of polyphosphates...................... Symbolic representation of a two-way table with t treat- ments, b blocks, and equal numbers of observation (n = 4) in eaCh cell,k:1...onecoo0.0000000000000000...0000000000 Symbolic description of analysis of variance of a two-way table with t treatments, b blocks, and equal numbers Of‘ observations in each cell................................. Average values of mean squares for a randomised Complete- bIOCR anaIYSisOOOO0.000000000°0000000900090.00000000000000 Recommended ratios of solution volume to metal surface"" area.OOOOOOOOOOOOOOOOOOO000000.000.00000000000000000000000 Interpretation of scores according to Darrin (1946)....... Composition of hot rolled and cold rolled AISI No. C 1008 Steel spec1men8000000.000.0000000000COOOOOOOOCOO0.0.0.0... Interpretation of scores according to Darrin (1946)....... Corrosion weight loss data of steel specimens immersed in distilled water for one week (7 days)..................... Corrosion weight loss data of steel specimens immersed in alkaline detergent solutions for one week (7 days)........ Analysis of variance of corrosion weight loss data of cold and hot rolled AISI No. C 1008 steel specimens immersed in distilled waters.cocoa.acoo009000ooeee00000000000000.0000. Analysis of variance of corrosion weight loss data of cold and hot rolled AISI No. C 1008 steel specimens immersed in detergent salutionsi.0000.000.0000...OOOOIOOOOOOOOOOOOOOOO Macroscopic examination of the effects of corrosion on cold rolled AISI No. C 1008 steel specimens. Mean values for three1mersion8dataOOOOIO0.00..OOOOOOOOOOOOOOOOOOOOO Macroscopic examination of the effects of corrosion on hot rolled AISI No. C 1008 steel specimens. Mean values for three immersions data................................. Macroscopic examination of the condition of the solutions after immersion of the cold rolled or hot rolled steel specimens. Mean values for three immersions data......... Page 17 22 32 33 35 47 51 62 67 69 70 71 73 74 75 76 Table Page 18. Over-all condition of the system expressed as a percent- age of the maximum possible score for the sum of the sol- ution and the specimens in the system before removal of corrosion products. Mean values for three immersions dataOOOOOOOOO.IOOOOIOOOOOOOOOOOOOOOOOOOOOOOOOOOOOCOOOOOOOO 77 19. Corrosion weight loss data obtained by three individuals using cold rolled AISI No. C 1008 steel specimens immer- sed in 32 sodium hydroxide plus 0.084% sodium gluconate for one week (7 days)..................................... 78 20. Analysis of variance of corrosion weight loss data ob- tained by three individuals ........... ...,................ 78 21. 'Analysis of variance of corrosion weight loss data ob- tained by three individuals (by immersions)............... 79 22. Corrosion weight loss results of cold rolled AISI No. C 1008 steel specimens immersed in distilled water, sequestering agent solutions for one week (7 days)........ 81 23. Macroscopic examination of the corrosive effects of dis- tilled water, sequestering agent solutions on cold rolled AISI No. C 1008 steel specimens. Mean values for five 1mersions dataOOOOOO0.0.0.000.0.000000000000000IOOOOOOOO. 82 24. Macroscopic examination of the condition of the solutions after immersion of cold rolled AISI No. C 1008 steel specimens. Mean values for five immersions data.......... 83 25. Over-all condition of the system expressed as a percentage of the maximum possible score for the sum of the solution and the specimens in the system before removal of the corrosion products. Mean values for five immersions data. 84 26. Corrosion weight loss results of cold rolled AISI No. C 1008 steel specimens immersed for one week (7 days) in alkaline detergent, distilled water solutions.......... 86 27. Analysis of variance of corrosion weight loss data for steel specimens immersed in alkaline detergent, distilled water salutionBOIOOOOOO0.0.0...0......OOOOOOOOOOOOOOOOOOOO 87 28. IMacroscopic examination of the corrosion effects of alkaline detergent, distilled water solutions on cold rolled.AISI No. C 1008 steel specimens. Mean values for five imeraiona data..0.0..0...OOOOOIOOIOIOOOIOOOOIO0.0... 93 29. Macroscopic examination of the condition of the alkaline detergent, distilled water solutions after immersion of cold rolled AISI No. C 1008 steel specimens. Mean values for five immersions data.................................. 94 30. Over-all condition of the system expressed as a percent- age of the maximum possible score for the sum of the sol- ution and the specimens in the system before removal of the corrosion products. Mean values for five immersions.. 95 vi Table 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. Page Analysis of variance of corrosion weight loss data for steel specimens immersed in distilled water and 3% sodium hydroxide, phosphate solutions............................ 95 Comparison Of agent and immersion over-all mean weight " 10888800090000000000000OOOOOOOOOOOOOOOOOOOOOO000000000000° 96 Corrosion weight loss results of cold rolled AISI No. C 1008 steel specimens immersed in alkaline detergent, distilled water solutions at different temperatures for' one week (7 days)......................................... 97 Analysis of variance of corrosion weight loss data for steel specimens immersed in alkaline detergent, distilled water solutions at different temperatures................. 98 Comparison of the amount of steel specimens immersed in alkaline detergent, distilled water solutiOns at differn‘ ent temperatureSoooootnoononoocoooooooooooooooooooooOooooo 99 Macroscopic examination of the corrosion effects of alka- line detergent, distilled water solutions at different temperatures on cold rolled AISI No. C 1008 steel specié mens. Mean values for five immersions data............... 101 Macroscopic examination of the condition of the solutions used at different temperatures after immersion of cold rolled AISI No. C 1008 steel specimens. Mean values for five immersions data...................................... 102 Over-all conditions of the systems at different temper- atures expressed as a percentage of the maximum possible score for the sum of the solution and the specimens in the system before removal of corrosion products. Mean values for five immersions data........................... 103 Corrosion weight loss results for cold rolled AISI No. C 1008 steel specimens immersed in an alkaline sequester- ing agent solution with and without a wetting agent for one week (7 days). Solutions prepared with distilled waterOOOOOOOOOOOOOO0°000000000000.000000000000000000000000 104 Analysis of variance of corrosion weight loss data for steel specimens immersed in an alkaline sequestering agent solution with and without a wetting agent........... 104 Corrosion of cold rolled AISI No. C 1008 steel specimens immersed for increasing lengths of time in alkaline detergent, distilled water solutions...................... 106 Analysis of variance of corrosion weight loss data for steel specimens immersed for increasing lengths of time in alkaline detergent, distilled water solutions.......... 107 Comparison of the amount of corrosion of steel specimens immersed for increasing lengths of time in alkaline detergent, distilled water solutions...................... 108 vii Table 44. 45. 46. 47. 48. 49. 50. 51. 52. S3. 54. 55. Page Macroscopic examination of cold rolled AISI No. C 1008 steel specimens immersed for different lengths of time in alkaline detergent, distilled water solutions. Mean values for two immersions data............................ 109 Macroscopic examination of the condition of the solutions used for the different lengths of immersion of cold rolled AISI No. C 1008 steel specimens. Mean values for two immersions data....................................... 111 Over-all condition of the systems tested for different lengths of time expressed as a percentage of the maxi- mum possible score for the sum of the solution and the specimens in the systems before removal of corrosion products. Mean values for two immersions data............ 112 Corrosion weight loss results of cold rolled AISI No. C 1008 steel specimens immersed for one week (7 days) in alkaline detergent, hard water solutions of 12 grains per gallon total hardness as calcium carbonate................ 113 Analysis of variance of corrosion weight loss data for steel specimens immersed in alkaline detergent, hard water salutionBOOOOOOOOO00.00.00.004.OOOOOIOOOOOOOOOOOOOOOO 115 Macroscopic examination of the corrosion effects of alkaline detergent, hard water solutions on cold rolled AISI No. C 1008 steel specimens. Mean values for five immersions data.......... ...... ........................... 117 Macroscopic examination of the condition of the alkaline detergent, hard water solutions after immersion of cold rolled AISI No. C 1008 steel specimens. Mean values for five immersions data...................................... 118 Over-all condition of the alkaline detergent, hard water systems expressed as a percentage of the maximum possible score for the sum of the solutions and the specimens in the systems before removal of the corrosion products. Mean values for five immersions data...................... 119 Corrosion weight loss results of cold rolled AISI No. C 1008 steel specimens immersed in water for one week (7 days)........OOOOOCOOOOOOOCOCO-D.IOOOOOOOOOOOOCOOOOOOOO. 121 Analysis of variance of corrosion weight loss data for steel specimens in distilled water and hard water of 12 grains per gallon total hardness as calcium carbonate.. 122 Macroscopic examination of the corrosion effects of dis- tilled water and hard water on rolled AISI No. C 1008 steel specimens. Mean values for five immersions data.... 123 Macroscopic examination of the condition of the dis- tilled water and hard water solutions after immersion of cold rolled AISI No. C 1008 steel specimens. Mean values for five immersions data........................... 123 viii Table 56. 57. S8. 59. 60. 61. 62. 63. Page Over-all condition of the system expressed as a percent- age of the maximum possible score for the sum of the sol- ution and the specimens in the system before removal of the corrosion products. Mean values for five immersions dataOOOOOOOOOO000.000.00.000000000000000000000.00.00.00.00 124 Analysis of variance of corrosion weight loss data for steel specimens immersed in distilled water, 12 grain? per gallon hard water and distilled water plus 0.1561 tetrasodium ethylenediaminetetraacetate................... 124 Comparison of the amount of corrosion of steel specimens immersed in distilled water, 12 grain per gallon hard water, and distilled water + 0.156% tetrasodium ethylene- diaminetetraacetate. ..................................... 124 Analysis of variance of corrosion weight loss data for steel specimens immersed in alkaline detergent solutions made with distilled water and 12 grain per gallon hard water ...... .......... ..... ........... .................... 125 Comparison of the amount of corrosion of steel specimens immersed in alkaline sequestering agent solutions made with distilled water or 12 grain per gallon hard water.... 126 Analysis of variance of corrosion weight loss data for steel specimens immersed in distilled water and 12 grain per gallon hard water solutions of 3% sodium hydroxide plus sodium gluconate and 3% sodium hydroxide plus tetra- sodium ethylenediaminetetraacetate (each sequestering agent used at three concentration 1evels)..... ..... ....... 127 Comparison of the amount of corrosion of steel specimens immersed in alkaline sequestering agent solutions of different concentrations made with distilled water and 12 grain per gallon hard water............................ 128 Analysis of regression of corrosion weight loss (Y) on the concentration of sequestering agent (X)............... 129 ix Figure l. 2. 10. 11. 12. 13. 14. 15. 16.‘ 17. LIST OF FIGURES Effect of pH on sequestering action of polyphosphates.... Effect of water hardness‘on‘sequestering'action‘0£‘p¢1y4" phosphates.OOOQOOOOOOOOOOOOOOOOO00.00.00.000...0.0.0.0.... Reversion rates of various polyphOsphates at 100°C........ Relative calcium chelating capacities of various sequester- 1“ agents as affected by pHOOO.0.0...OOOOOOOOOOOOOSOOOOOOO Relative calcium chelating capacities of various sequest- ering agents as affected by sodium hydroxide concentration. Experimental designOOOOOOOOOOOO..COOOOOOOOO0.00.00.00.000. mrrosion experiment....0OOOOOOOOOOOOOOOOOOO0.0.00.0.COCO. Immersion-times-steel interaction with distilled water as corrosive liqu1d000000.09.00000000000000000000000.00.00... Immersion-times-agent interaction with cold and hot rolled steel spec1men80000000000O0.0...OOOOOOOOOOOOOOOOOCOOO....0 Immersion-times-individual interaction obtained with steel specimens immersed in 32 sodium hydroxide plus 0.0842 sadi‘mgluconateOOOOOOOOO00......OOOOOOOOOOOOOOOOOO Comparison of the amount of corrosion of steel specimens immersed in alkaline sequestering agent solutions made With distilledwaterOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Comparison of the amount of corrosion obtained with dife ferent immersions of steel specimens in alkaline sequest- ering agent solutions made with distilled water........... Imersion-timeB-agent interactions 0 s s s a s o s s s a s a s s s s s s o a s s s Corrosion weight loss of steel specimens immersed in alka- line detergent, distilled water solutions at different temperatures for one week (7 days). Temperature over-all mean weight losses plotted................................ Arrhenius plots of corrosion weight loss of steel speci- mens immersed in alkaline detergent,distilled water sol- 'utions at different temperatures for one week (7 days). Temperature over-all mean weight losses plotted........... Comparison of the amount of corrosion of steel specimens immersed in alkaline sequestering agent solutions made with hardwaterOOOOOOOOOOOOO0.0.0.000...OOOOOOOOOOOOOOOOOO Comparison of the amount of corrosion obtained with dif- ferent immersions of steel specimens in alkaline sequest- ering agent solutions made with hard water................ X Page 23 23 25 25 26 31 65 72 72 80 88 89 90 100 100 114 116 Figure 18. 19. 20. 21. Corrosion weight loss of steel specimens immersed in alkaline solutions made with distilled water and con4 The estimated linear regression line is drawn.............. Corrosion weight loss of steel specimens immersed in alkaline solutions made with 12 grain per gallon hard water and containing different concentrations of sodium ........ gluconate. The estimated linear regression line is draVnOOOOOOO0.0...OOOOOOOOOOOOOOOOOOIOOOOOOOOOIOOOOODOOOO. Corrosion weight loss of steel specimens immersed in alkaline solutions made with distilled water and con- dimnetetraacecateOOO...0....OOOOOOOOOOOOOOOOOOOOO0...... Corrosion weight loss of steel specimens immersed in alkaline solutions made with 12 grain per gallon hard water and containing different concentrations of tetra- sodium ethylenediaminetetraacetate........................ xi Page 130 130 131 132 INTRODUCTION Scale in beverage bottle washers results in interference with normal bottle pickup and release, increased carry over of washing solutions into the rinse sections of the washer, increased wear and drag ofVIoving parts, decreased speed.of machine operation, and impedance of heat transfer due to layers of scale on the heat transfer surfaces. To obtain clean bever- age bottles, and for efficient and effective bottle washer operation, 1 scale formation and accumulation in the washer and on itssmoving parts must be prevented. I The beverage industry usually softens the water used for cleaning reusable containers by chemical treatment. ’Sequestering agents of both the inorganic and organic types are utilized for this purpose. The most widely used sequestering agents are the inorganic phosphates because of their relatively low cost. More recently the organic complexing agents such as the alkali metal salts of the hydroxycarboxylic acids (e. g., sodium gluconate) and the amdnopolycarboxylic acids (e. g., tetrasodium ethylenediaminetetraacetate) have become increasingly more important in beverage bottle washing. With the increased use of sequestering agents, concern relative to the effect of these agents on the washing machine has developed. A difference of opinion exists in the beverage industries as to the effect of these agents on bottle washing equipment. Information on the corrosiveness of these agents is needed and would be of value to the various beverage industries and their‘suppliers. This investigation was designed to examine the corrosion characteristics of basic chemical in- gredients that are used either singly or in formulated compounds so as to better understand the behavior of these compounds-under bottle wash- ing conditions. No formulated bottle washing detergent compounds or mixes were examined since the effect of individual ingredients could not be determined. This investigation is a part of a project sponsored by the American Battlers of Carbonated BeVerhges Association to evaluate the effect of basic bottle washing components on washing equipment and their effectiveness in cleaning soiled bottles, to ultimately ascertain if present standards for cleaning carbonated beverage bottles are satis- l 2 factory or if they should be revised. V 3 This thesis describes the results of.anLinvestigation-of the corro- sion of mild steel in alkaline sequestering agent solutionse .The first part of this thesis describes the development of the experimental proced- ure employed in the investigation. Literature on corrosion testing~pro- cedures is reviewed and the results of preliminary tests to obtaianuit- able procedures for preparing specimens before.exposure and removing---— corrosion products after exposure are included. The second part of this thesis describes the results of the investigation to determine the effect of organic and inorganic sequestering agents on the corrosion of mild steel in distilled water, distilled water plus sodium hydroxide, and hard water plus sodium hydroxide solutions under a variety of test conditions. REVIEW OF LITERATURE Corrosion Costs Estimates have been made of the.economic cost of corrosion. Uhlig (1949) made a careful survey of only the direct losses by corrosion in. . the united States and reported a rate in excess of $5.5 billion annually. Jelinek (1958) states that in the nine year interval since Uhlig's study was published, corrosion control methods have improved considerably,.but material and labor costs have generally risen and thus the order of mag- nitude of the direct costs of corrosion remains about the same.- Schmitt (1960) states that corrosion problems cost near $8 billion annually, while mining metals cost about $1 billion annually. Gegner (1960) says that the cost of corrosion to this country has been estimated by various experts to be upwards of $7.5 billion per year of which industry's share is about $6 billion. Included in these estimates are the direct costs due to corrosion alone; such as replacement and maintenance of corroded equipment, costs of painting, electroplating, and other protective meas- ures including application and labor costs. Among the indirect costs not included are such things as safety hazards, product contamination, shutdown time and loss of products when equipment fails. Equipment over- design is often required where uncertainty relative to corrosion effects exists. Thus, it is readily apparent that strong incentives exist for improvement in decreasing corrosion. Improvement of only one percent in decreased corrosion would mean an annual savings of at least $55 million. Another example of the tremendous cost of corrosion, which in this case should be of particular interest to the American automobile owner follows: It has been reported in a staff feature of Corrosion (1961) that corro- sion damage to automobile exhaust systems cost American car owners over $500 million in 1960. According to this report, 30 million mufflers sold in 1960 at an average retail price of $12.50 for a total of $375 million. About 35 million tailpipes were also sold the same year at an average retail price of $5 for a total of $175 million, giving a grand total of over $500 million. It was pointed out that the loss to corro- 3 4 sion of auto mufflers alone ($375 million) equaled the total cost to the united States for construction of the Panama Canal. Cegner (1960) states in his paper on alkaline environment corrosion. that industry has long been experiencing a steady erosion of its profits. Further, that competition,public opinion and more recently a genuine fear of inflation has prevented most of them from raising prices to—maintain their margin of profit and thus industry has been forced to search out every possible means for reducing its-manufacturing costs. He goes on to point out that because of its size, maintenance cost is one of the~ obvious targets and within this area are the sizeable costs attributable to corrosion. It is this same concern relative to maintenance and replacement costs within the carbonated beverage industry which is the primary reason for this study. The specific problem pertains to the corrosive effect of relatively recently introduced sequestering agents on’the bottle washer, relative to the corrosive effects of the more traditionally used agents and components of the cleaning systems. Corrosion of Iron and Steel Nieland, Maguire, George and Kahler (1950a) have reported that sol- uble salts of hydroxypolycarboxylic acids having less than 7 carbon atoms, such as sodium citrate, tartrate, malate, and mucate form a vis- ibly thick, black,tenacious film over iron surfaces. ‘Concentrations from 0.1 to about 100 ppm reportedly gave maximum protection; higher concen- trations decreased protection and caused pitting. Above 1000 ppm the total weight loss of the exposed specimens exceeded that of the control. Sodium citrate was reported to have penetrated existing tuberculation products and to have formed a film on the base metal underneath, grad- ually restoring the carrying capacity of pipe. They found 2 ppm of sod- ium citrate to give 60% protection in Philadelphia tap water at 1600?, which was noted as being triple the protection of 2 ppm of sodium hexa- metaphosphate, or 100 ppm of sodium silicate solution (28% $102). The film was reported as seeming to be established in all cases in about 5 days. It was pointed out that the film with sodium citrate formed fast- est in low hardness water. The greatest corrosivity was found in inter- mediate hardness waters. 5 Nieland, Maguire, George and Kahler (1950b) have also reported that solutions of 0.1 to 100 ppm of sodium gluconate and gluconic acid act. similarly to other hydroxycarboxylates and.form a hard, dark,corrosion re- sistant protective film over iron surfaces. Above 100 ppm the over-all protection was found to continue without appreciable reduction in weight loss, but the attack localized in pits, and perforations were found in a flow system using Philadelphia tap water of 42 ppm total hardness as CaCO3 at 120°F. The percent protection given for a concentration of 840 ppm of sodium gluconate was about 84%. It was stated that sodium gluc-u- onate penetrates existing corrosion products, forming a film on the metal underneath and allows for their gradual removal. It was found that the dosage could be reduced once the film was formed. Dvorkovitz and Hawley (1952a) have stated that hard adherent scale or etchings formed on glassware from the constituents of water in which it is washed, is inhibited by the use of a mixture of about 90 parts by weight of caustic alkali, such as sodium hydroxide, and 10 parts of sod- ium or potassium gluconate. They noted that this mixture also serves as a better inhibitor of rust, and iron and steel corrosion, than sodium hydroxide solutions alone. Dvorkovitz and Hawley (1952b) have also stated that the hard water used for rinsing glass, metal, rubber, and porcelain articles coming from alkaline washing solutions can be treated with caustic alkali and alkali metal gluconates in the amounts of 0.25 - 20% and 0.025 - 2.0%, respect- ively. This treatment is advocated in order to prevent or reduce scale precipitation of calcium and magnesium salts which interferes with the clearing operation on the surface of the objects. It was pointed out that such solutions have the advantage of better inhibition of rust and corrosion of iron and steel, and are relatively non-foaming. An extract of a 1955 letter from Glyco Products Company forwarded to us by Cooper (1960) of the George J. Meyer Manufacturing Company, states, "it is reported that corrosion of washing equipment occurs with sodium gluconate compositions. In addition, this type of compound does not sequester iron or aluminum. Aluminum is deposited in washing from alum- inum labels frequently used on milk bottles." Shaw and McCallion (1959) have reported results on corrosion by 6 gluconate-caustic systems and other considerations. They stated that, "some time ago corrosion of bottle washing equipment using gluconate- caustic solutions was reported. Actually the complaints were not wide- spread but the fact that they existed at all, was sufficient reason to begin an investigation." The experimental procedure employed will be described in detail in the deve10pment of experimental procedure section~ of this dissertation. The type of metal (carburized and hardened steel), pre-cleaning procedures and procedures for the removal of corrosion prod- ucts differed from those employed in the subsequent studies to be reported here. Duplicate tests were conducted with rotating specimens. Test sol- utions of 3% caustic plus additives (based on total solution weight) were prepared using water of 12 grains per gallon hardness. Each test was conducted for 7 days in a constant temperature bath at 140°F. The re- sults as reported in mils penetration per year (MPY) and recalculated (metal density assumed = 7.87 gm/cm3) to give the results in terms of weight loss (mg/dmZ-week) are given in Table 1. They noted that test solutions containing phosphates were generally not as corrosive as the gluconate-Caustic system, and in fact reduced corrosion to a point much less than the caustic control. When the level of gluconate was reduced from 0.15% (which is three times as much as is theoretically required for 12 grain hardness water) to 0.045% (mole to mole with calcium) the corro- sion rate was reduced 60% to a point essentially the same as that of the caustic control. They stated that, "the corrosion results certainly in- dicate that the higher the level of gluconate, the more corrosion. The D primary lesson, therefore, is that large excesses of gluconate should be avoided. Large excesses are quantities far over and above the levels actually required to do a satisfactory job." Shaw and McCallion present recommendations as to amounts of gluconate to be used, based on the fact that 1 mole of gluconate sequesters 1 mole of calcium or magnesium. The gluconate levels recommended in practice are twice these amounts. It was also pointed out that the gluconate level may need to be still high- er if rusty necks, aluminum labels, bottles containing cement, etc., are common occurrences. Cooper (1960) states that as a result of non-biased laboratory and field investigations on sodium gluconate and other sugar acid materials, 7 Table l. Corrosiveness of some gluconate-caustic systems (Shaw and McCallion (1959)). Test solutions . Corrosion rate Mils Weigh loss, penetration mg/dm -week per year, MPY . 3% Sodium hydroxide 0.40 15.38 3% Sodium hydroxide 1.11 42.67 + 0.15% sodium gluconate 3% Sodium hydroxide 0.47 18.07 + 0.045% sodium gluconate 3% Sodium hydroxide 0.21 8.07 + 0.10% sodium gluconate + 0.10% trisodium phosphate 3% Sodium hydroxide 0.24 9.23 + 0.15% sodium gluconate + 0.05% trisodium phOSphate 3% Sodium hydroxide 0.05 1.92 + 0.10% sodium gluconate + 0.10% sodium tripolyphosphate 3% Sodium hydroxide 0.54 20.76 + 0.15% sodium gluconate +0.0075% sodium tripolyphosphate 3% Sodium hydroxide 0.09 3.46 + 0.10% sodium gluconate + 0.10% sodium hexametaphosphate 3% Sodium hydroxide 1.45 55.74 + 0.10% sodium gluconate + 0.10% tetrasodium pyrophosphate 3% Sodium hydroxide 0.72 27.68 + 0.15% sodium gluconate + 0.05% Renex (surface-active agent) that they do not recommend the use of these materials in Meyer Dumore washers. He points out that it is true that these compounds, if properly used, will inhibit scale formation in the strongly alkaline compartments of the washing machine; however, he also points out that if the materials are used they must be carefully tailored to fit the conditions in each bottling plant. He states that too much of the gluconate will favor rusting and too little will not give the proper amount of protection. He goes on to point out that even if the proper amount of material is added to the freshly prepared solution, there is no method known for the 8 rapid evaluation of the residual gluconate, in order to bring up the per- cent of material to the original working strength. C00per states that they have information from a manufacturer of gluconates stating that as the caustic strength decreases in the various washer compartments, the . efficiency of gluconate drops rapidly to the point where it gives no pro- tection. They advocate the use of some material such as tetrasodium pyro- phosphate in the rinsing compartment of the washer in order to prevent scale build up. White (1962) points out that it is undesirable to go significantly beyond the minimum concentrations and temperatures recommended for wash- ing bottles because hot caustic solutions are very corrosive to the machine and to the glass itself. Putilova, Balezin, Barannik (1960), and Levina (1956) state that the solution of iron in alkalies has received relatively little study. They note that various substances have been used as inhibitors of the solution of iron in acids. Among these are the substituted aliphatic and aromatic amines, sulfonated naphthalene and anthraquinone derivatives. These in- hibitors are reported to be quite ineffective in alkaline solutions, since they are not adsorbed on the iron under these conditions. This has been attributed by Levina to the fact that some are practically insol- uble in alkaline solutions, and also to the fact that iron in alkaline solution is coated by oxides that have very pronounced hydrophilic prop- erties. For these reasons the author stated that the choice of inhib- itors of iron corrosion in alkaline media should differ from those used for acid solution. Levina states further that the addition of substances capable of forming surface compounds with iron and its oxides should prove to be inhibitors of the solution of iron in alkaline solutions. To determine if such was the case, silicic acid, sodium phosphate, and tannin were chosen to be evaluated. These substances were noted as being easily soluble in alkalies and also to form stable chemical compounds with iron and its oxides. Levina states that it was not possible to pre- dict their influence on the solution of iron. The data obtained in the studies showed that silica gel, phosphate, and tannin form surface com- pounds with the iron. It was reported that these compounds retard sol- ution of iron by raising the hydrogen overvoltage and that this is not 9 accompanied by a decrease of the electrochemical activity of the iron surface. Electrodes treated in this manner were shown to have consider-. able electrochemical capacity. It was further noted that when large_quan- tities of these inhibitors were added, gas evolution at the iron ceases and the iron loses its electrochemical activity. Corrosion of Some Other Metals McCune (1958) investigated and reported on corrosion of aluminum by alkaline sequestering solutions. Corrosion of aluminum by solutions of sodium triphosphate and sodium ethylenediaminetetraacetate (sodium ((ethyl- enedinitrilo)) tetraacetate) is treated in detail. Weight loss data were determined with an analytical balance on specimens exposed in stagnant total immersion tests. The ratio of area of metal to volume of solution used was 29 cmzper 200 ml. Aeration was shown not to affect the results appreciably. Distilled water and reagent grade chemicals were used in preparing the test solutions. Solutions of sequestering agent (sodium ethylenediaminetetraacetate, pyrophosphate, triphosphate) corroded alum- inum.much more rapidly at 60°C than solutions of non-sequestering salts (sodium trimetaphosphate, sodium sulphate, tetramethylammonium sulphate) at alkaline pH values up to about 10.5. Above this pH, alkaline corrosion reportedly overwheLmed sequestering corrosion. The data were obtained at constant pH. With sodium triphosphate corrosion was general over the surface of the metal and conditions of 60°C, 3 hours, and 200 m1 of 0.18% sodium triphosphate solution resulted in etching and a decrease of spec- ular reflectivity of the metal with little or no discoloration. Lower sodium triphosphate concentration, smaller volumes, and longer times caused a brown discoloration. Pitting was not observed at the pH of sod- ium triphosphate solutions. In the presence of sufficient sodium tri- phosphate, aluminum perchlorate-perchloric acid solutions were found to form no precipitate when sodium hydroxide was added between pH 3 and 11. The over-all reaction was reported to be that of metallic aluminum*with triphosphate solution to give an aluminum-triphosphate complex and hy- drogen, and in the pH range 7 to 10, hydroxide ion. The following over- all reactions were indicated as illustrative: At about pH 8; -4 -3 .. A1 + 3H20 + HP3010 = (A1(OH)2HP3010) + OH + 1.5H2 10 At about pH 10; -5 —5 A1 + P3010 + 3112 10 _ 2 It was reported also that aluminum immersed in 0.182 phosphorus -32- tag- 0 = A1(OH)3P3O + 1.5 H sed (P32),sodium triphosphate solution for 3 hours at 60°C sorbed phos- phate equivalent to 0.019 mg sodium triphosphate per cm?, according to radioactive count. McCune stated that the action was apparently in the hydrous oxide surface film, since the radioactive triphosphate was removed by the nitric acid rinse which dissolves corrosion products, but very little of the metal itself. The surface film isolated from an aluminum foil exposed to 0.181 of sodium ethylenediaminetetraacetate solution under the same conditions contained nitrogen and carbon indicating the presence of the sequestrant in the film. The over-all corrosion reaction at about pH 8.5 was given as Al + 21120 + HEMA'" = (A1(0H)EDTA)" + 011' + 1.5 H2 It was further pointed out, that this demonstration of the presence of the corrosive agent at the solid surface is pertinent to recent discuss- ion of chelating agents as corrosion inhibitors or as corrosion acceler- ators in cases where the surface complex of the chelating agent and metal ion is sufficiently stable to disrupt the metal or oxide lattice. Add- itions of metal cations to triphosphate solutions were found to reduce corrosion; the order of increasing effectiveness was found to be the same as the order of increasing complexing strength of these cations by the polyphosphate. Lithium, calcium and magnesium, and aluminum and nickel sulfates reduced the corrosiveness of sodium triphosphate, with effect- iveness increasing in that order. Sundararajan and Char (1961) studied the corrosion rates of commer- cial aluminum, containing 4 percent manganese and 3 percent iron, in sodium hydroxide solutions under different conditions. Inhibitor effi- ciencies were calculated for the following substances, the efficiency decreased from 90 to 60 percent in the order: agar-agar, gum-acaciae, dextrin, gelatin, and glue. Corrosion potentials were measured with and without dextrin inhibitor, and cathodic and anodic polarization studies were also made. They reported that the results obtained showed that the dissolution of the metal was electrochemical in character. The corrosion ll process was described as appearing to be under mixed control, with the' predominance of action of dextrin inhibitor on the anodic areas of the metal surface. Hbore and Smith (1942) conducted laboratory studies to determine the influence of metaphosphate on the quantity of lead picked up by water passing through lead wool and through lead pipe previously exposed to cor- rosive water. They reported that in water having a pH value of 7.0 or - less, the addition of metaphosphate materially reduced the amount of lead taken up by the water. Itwas reported that under certain conditions, the addition of metaphosphate to a water having a pH value of 8.8 result- ed in an increased amount of lead taken up by the water. (The authors pointed out that it seemed to them that the gravest question (with re- gard to danger to public health) with respect to the introduction of metaphosphate into water systems containing lead pipe was the effect of this chemical on the deposits of lead salts previously formed in the pipe.) They stated that most recent work indicates that metaphosphate treatment is most effective at low pH values. Jensen and Claybaugh (1951) reported on the corrosive effects of sodium tripolyphosphate and tetrasodium ethylenediaminetetraacetate on tinned steel when used as 90% of detergent in 0.31 distilled water sol- ution and in presence of O, 4, 8, 12, 16, 20, 24, and 361'sodium meta- silicate with a five day exposure period.' Weight losses of 0.1462 g and 0.2884 g were reported for sodium tripolyphosphate and tetrasodium ethyl- enediaminetetraacetate respectively, when no sodium metasilicate was adde ed to the solutions. Addition of 0.41 metasilicate to the tripolyphos- phate solution reduced the weight loss to 0.0084 g. Additional corrosion inhibition was not particularly apparent above this level of metasilicate. The addition of metasilicate in amounts up to 24% did not prevent tetra- sodium ethylenediaminetetraacetatéh heavy corrOsion. . ’- Lewandowski (1952) presents comparative data on some acids used in dairy cleaning. The study included corrosion of tin plate (hot-dip and electro- plated tinned steel strips were both used)“and“stainless steel (18-8 sheet) strips in sodium bisulfate and in phosphoric, gluconic, levulinic, glycolic, citric, acetic and sulfamic acids. All acidic solutions were prepared in distilled water with a final concentration of 0.25 percent 12 by weight. The strips (about 3 x l in.) were precleaned before partial immersion in essentially static test solutions (70 ml). After immersion the cleaning procedure consisted of rinsing the strips three times in 7-, distilled water, with light brushing during the second rinse.~ The strips were then dried and weighed. Lewandowski concluded that'factorsrcontribut- ing to corrosion of tinned steel by mild acidic solutions were temper- ature, pH, apparent volatility of organic acids, structure of acidic com- pounds and type of tinplate. Weight loss of tinned steel changed or re- mained constant with change in temperature depending on the acid compound and type of tinplate. Under all test conditions, phosphoric acid caused only very slight weight losses as compared to the various organic acids, and sodium bisulfate and sulfamic acid. Lewandowski noted that the corro- sion weight loss figures obtained were all below the so called'herious" weight losses represented by 100 mg/dmz-day, which he points out is an indication of the essentially mild nature of all the tested acidic sol- utions. An apparent relationship between pH of acidic solutions and formation of colored deposits, as well as spangling (seemingly etched patterns) of treated tinned steel strips was observed. Pitting of tinned steel was observed only with volatile organic acids, and was also related to the presence of sulfur in molecules of the inorganic acidic compounds used. Corrosion was noted as more intense on thinly electroplated than on heavily plated hot-dip steel. Stainless steel was not appreciably corroded by any of the acidic solutions tested. Solutions of four polyphosphates (tetrasodium pyrophosphate, sodium tetraphosphate, sodium hexametaphosphate, and sodium tripolyphosphate) were shown by Bacon and Nutting (1952) to dissolve copper, zinc, or brass. Corrosion rate data were presented for several metal-polyphosphate combin- nations at 80°C and for several commercial and experimental dishwashing detergents. The inclusion of sodium metasilicate was shown to afford a practical means of retarding the corrosive action of polyphosphates on dishwashing machine parts. 0f the several polyphosphates investigated, pyrophosphate was the most difficult to inhibit. It was pointed out that corrosion of copper, brass, or bronze pumps, valves, and spray nozzles by polyphosphated detergents may seriously reduce the efficiency of dish- wastuing machines and that many complete failures of machine parts have 13 occurred within 6 months. A brassy tarnishing of silverware washed in certain dishwashing machines was attributed to copper solubilized from machine parts by poorly inhibited polyphosphated detergents that then plated on the silverware by electrolytic action. Getty,‘McCready, and Stericker (1955), have determined the degree of corrosion with some synthetic detergent solutions (wetting agent plus polyphosphate), and the degree of protection obtained by the addition.of» various amounts of various silicates. The tests were made with aluminum, zinc, copper, brass, nickel silver, and several enamels. The addition of certain sodium silicates reduced corrosion or tarnishing of the metals in household detergent mixtures containing polyphosphates. It was point- ed out that the silicate content of the most popular heavy-duty household detergents now on the market is sufficient to prevent the corrosion of aluminum, but an increase would be desirable to reduce corrosion of zinc and its alloys. It was shown that silicated detergents lessen any at- tack and loss of gloss of enamels, and especially with less resistant enamels. Wright (1956) has also reported. that silicates are.effective corrosion inhibitors for cleansers of aluminum dairy equipment (particularly aluminum milk cans). Hunziker, Cordes, and Nissen (1929) studied the resistance to corro- sion of 19 different metals, plated metals, and metallic alloys in sol- utions of washing powders, chemical sterilizers, and refrigerating brines. The metals included in the tests were: manganese-aluminum alloys, alum- inum plated alloy, capper, nickel, monel metal, nickel silver, tin, tin- ned copper, tinned iron, iron, zinc, galvanized iron, ordinary chromiwm steels in the form of Ascoloy and Enduro, and a chromium-nickel steel in the form of Allegheny Metal. The alkalies tested were a special alkali (35%“Na0H, 62% Na2C03), sodium carbonate, wyandotte (492 Na2C03, 361 NaHCOa), sodium hydroxide, trisodium phosphate, and commercially form- ulated tin cleaner, and Diversal. Chemical Sterilizers were sodium hyt pochlorite, Diversol, and ChloraminerT. The test comprisedtpartial immer- sion of the metal strips in solutions‘in closed'Mason jars for 5 hOnrs at 150°F for.alkalies and 5 days at 70°F for chemical sterilizers. Hunziker, Cordes and Nissen used 0.5 percent solutions of alkalies. Aluminum prod- l4 ucts suffered by far the most intense corrosion in all washing solutions used. The tinned copper and tinned iron products ranked next in intensity of corrosion, but showed much greater resistance than the aluminum prod- ucts. 0f the remaining metals, the corrosion of copper, iron, galvanized, iron, and zinc was decidedly greater, but not as severe as that ofthe tin- plated products. They found that the addition of 0.025 percent sodium chromate to 0.5 percent solution of trisodium phosphate in water wery greatly reduced corrosion. ‘With 0.5 percent solutions of alkalies, the addition of sodium silicate in 0.05 percent quantities did not reduce the attack of NaOH and trisodium phosphate, but completely eliminated corro- sion by soda ash. The use of trisodium phosphate solution (0.16 to 0.5 percent) combined with 0,025 percent sodium chromate was described as best from the corrosion standpoint for tin-coated equipment. Parker (1942) reported a laboratory study of the corrosive action of an acid cleansing product known as Mikro-San (described as being a non- toxic mixture of certain organic acids, specific wetting agents, corro- sion inhibitor, and a microstatic agent), and acid sterilizing product known as Nflkro—Puer (content not given), and Chicago tap water on nine different metals normally used in the construction of can washers and dairy processing equipment, as well as that used in the construction of milk cans. The results obtained were compared with those of Hunziker, Cordes, and Nissen (1929) described above. Parker concluded that the corrosive effects of Mikro-San, Mikro-Puer and Chicago tap water were relatively mild compared to the various mineral and organic acids, and particularly to the action of the various washing powder solutions used by Hunziker, Cordes and Nissenr deVleeschauwer, Hendrickx, and Moulaert (1951) investigated the in- fluence of several detergents on aluminum, tinned iron, tinned copper, Yellow and red copper, galvanized iron, and chromium-plated red copper. Cleaned metal plates were weighed and immersed for three consecutive 24 hour periods in 70°C detergent or immersed for three consecutive 5 day Periods in 20°C disinfectant. The plates were rinsed, visually examined, dried, weighed and placed in new solution after each immersion. The wEight loss data were used to determine the resistance to corrosion of the metals. They stated that aluminum was generally the most affected. 15 Corrosion of the tinned iron and tinned copper were noted as not parti-. cularly important, but corrosion was observed with each solution tested. Weight losses from the yellow and red copper specimens were noted as being small, but they were stained and made dark by most of the deter- gents. It was observed that some products did not affect galvanized iron while others affected it strongly. Chromiumpplated red copper was only affected by Trosilin, and stainless steel was not affected at all by any of the detergents. A 0.51 solution of a quaternary ammonium compound was reported to have affected only the tinned metals, yellow and red copper, and galvanized iron. Sodium hypochlorite solution of 120 ppm affected aluminum; a higher concentration affected tinned metals, but to a lesser extent. Solutions of only 15 ppm were said to have stained copper and galvanized iron. The disinfectant solutions were found not to affect chromium-plated red copper and stainless steel. ' deVleeschauwer, Hendrickx, and Wallez (1953) studied the corrosive effect of detergents on milk-can metals. Aluminum and tinned iron were tested in solutions containing sodium hydroxide, sodium carbonate, sodium phosphate, sodium silicate and sodium hexametaphosphate. In the inner- sion test method used, each test panel was given 180 immersions of l min- ute each, rinsed, dried, and weighed. They noted that weight losses of light metals in milk-can cleaners should not exceed 20 g/mz-day and 60 g/mg-day for heavy metals. Haller, Grant, and Babcock (1941) studied the corrosive action of what they described as typical chlorine disinfectants for use on dairy equipment under various conditions. It was determined that sodium hypochlorite and Chloramine-T solutions containing approximately 200 ppm of available chlorine would corrode most metals used in the dairy industry. .Other studies on the effects of washing media on metals other than iron and steel, are those of Bablik and Belohlavy (1957), Stupel and Rock (1959), and Bukowiecki (1958). Bablik and Belohlavy investigated and determined that certain detergents containing polyphosphates will attack hot galvanized surfaces. Stupel and Rock determined the corrosive ef- fects of some soaps and washing compounds on stainless steel, copper, brass, bronz, tin, zinc and aluminum. Bukowiecki investigated the effects of various detergents on metals used in the construction of washing ma- l6 chines. The materials tested included galvanized steel, copper.with and without a solder or tin deposit, and aluminum alloy, and brass. Sheet metal samples were suspended in solution at 90°C and the weight loss .~ determined every three hours. The effects of agitation, water hardness, and contact between dissimilar metals were investigated. The most inten- sive attack was found to be produced by the simultaneous action of poly- merized phosphates and sodium perborate. Soap had little effect and it was stated that it may even act as an inhibitor. Galvanized steel and the aluminum alloy were found most susceptible to corrosion by detergents. Stainless steel and nickel-plated copped alloys were described as virtually corrosion resistant. It was also observed in some instances that inter- action of two metals led to anodic inhibition. It is apparent from the literature that many investigations have been conducted in the general area of detergent corrosion of metals and alloys. However the effect of alkaline sequestering agents on the corro- sion of mild steel have not been determined. Further there is a need for a comprehensive evaluation of the corrosive effects of organic sequest- ering agents, relative to the effects of the more traditionally used in- organic phosphates and other components of the systems, under conditions similar to and varied in accordance with what is commonly encountered in their practical use. General aspects pertaining to sequestering agents and their chemical behavior are elaborated in the next section. SEQUESTERING AGENTS AND THEIR CHEMICAL BEHAVIOR .Sequestering agents are used to chemically soften hard water used in cleaning reusable glass containers. The basic bottle washing recommend- ation of'the "American Bottlers of Carbonated Beverage Association" (1958) is: "unclean bottles shall be exposed to a 3 percent alkali solution of which not less than 60 percent is caustic (sodium hydroxide), for a period of not less than 5 minutes at a temperature of.not less than 1300?, or to an equivalent cleansing and sterilizing process." The bottles are then rinsed free of washing solution. 0f the 43 states having require- ments governing type and strength of solution, 26 of them follow this recommendation very closely, allowing only one solution concentration, a minimum temperature, and a minimum contact time. In addition to the al- kali used, other agents may be added to either improve solution deter- gency or for chemical water-softening purposes, since the sodium hydroxv ide bottle-washing solutions are adversely affected by hard water salts. The high hydrOxyl ion concentration tends to precipitate metal ions as slightly soluble hydroxides. Chaberek and Martell (1959) demonstrate the low solubilities of the calcium and magnesium salts of the anions present in these washing solutions by the following table: 'Table 2. Solubilities of some calcium and magnesium salts in water. I Anion - ‘ Solubility.g11 Calcium Magnesium Carbonate 0.014 0.1 Hydroxide 1.85 0.009 Phosphate 0.02 0.2 Silicate 0.09 --- They point out that this leads to serious precipitation problems, which result in both imprOper cleaning of the glass containers and acceleration of scale formation on the washing machine. Skaggs and Miller (1958) concur with this as pointed out in their studies of the influence of hard water on dairy detergents. ‘ Chemical water treatment is most commonly followed in dealing with 17 18 the water-hardness problem in bottle washing solutions. The softening agent, hereafter referred to as a sequestering agent, is added when un-. formulated mixes are used at the time of making up the solutions or when they are added to the washer. It is appropriate at this point to define sequestration and other terms in this area. Smith (1959) summarizes what 1; meant by sequestration and defines what function a material must perform to be regarded as a sequestering agent. He states,"sequestratiOn consists in the suppression of a partic- ular property or properties of a metal in solution, without the removal of that metal either into another phase nor its concentration into a particular portion of the original phase, while at the same time the agent used for the purpose of this sequestration must not introduce any new factor, reaction, or characteristic which makes the system unsuit- able for the original purpose." Chaberek and Martell (1959) state that a metal complex or metal co- ordination compound is the resulting substance when water molecules sur- rounding a metal ion are replaced by other molecules or ions. The group which combines with the metal ion is called a ligand. A metal chelate compound (or metal chelate) is defined simply as a complex in which the donor atoms (the atoms directly attached to the metal) are attached to each other as well as to the metal and the metal becomes part of a heter- ocylic ring. When a ligand forms a stable, water-soluble metal chelate or metal complex, the ligand is said to be a sequestering agent and the metal is said to be sequestered. The ligand (donor molecule or ion) of a stable, non-water-soluble metal chelate or metal complex may be refer- red to as a precipitating agent and the metal is said to be precipitated. Agents used for chemical water treatment may be classed as inorganic and organic agents, examples of which follow: Trisodium phosphate is an inorganic water-softening agent, softening by precipitation. Tetrasodium- pyrophosphate is also an inorganic water-softening agent, softening by sequesteration. Two groups of organic sequestering agents of importance are the hydroxycarboxylic acids such as gluconic acid and the aminopoly- carboxylic acids such as ethylenediaminetetraacetic acid. Chaberek and Martell (1959) point out that the most widely used sequestering agents are the inorganic polyphosphates (pyrophosphates, tripolyphosphates, and 19 polymetaphosphates),.because of their relatively low cost. In recent years the organic sequestering agents have become increasingly important. The organic sequestering agents and their compounds have higher thermal stabilities and tend to resist the effects of prolonged exposure to high temperature and alkalinity, while the polyphosphates tend to hydrolyze back to trisodium phosphate under these conditions. The sequestering agents to be used in this study are trisodium phos- phate (not technically a sequestering agent), tetrasodium pyrophosphate,- sodium gluconate, and tetrasodium ethylenediaminetetraacetate. Trisodium phosphate is an inorganic water-softening agent, functioning by precip- itation of metal cations, and is the end product of hydrolysis often called reversion of perphosphates, tripolyphosphates, and the polymeta- phosphates in aqueous solution. The reversion rate increases with temp- erature, is lower in alkaline solutions than in acid solutions, and metal- 1ic cations in the system (such as calcium and magnesium ions which are. . sequestered as complex anions) tend to decrease the stability, particular- ly at high pH levels (Schwartz, Perry, and Berch (1958)). Of the three condensed phosphates (pyro-, tripoly- and polymeta-), pyrophosphate hy- drolyzes more slowly than tripolyphosphate, which in turn hydrolyzes more slowly than tripolyphosphate, which in turn hydrolyzes more slowly than polymetaphosphates (glassy phosphates). The reversion rates of these three classes of phosphates are each separated by an order of mag- nitude. For this reason pyrophosphate is often recommended rather than tripolyphosphate or the polymetaphosphates for use in chemical water softening of solutions used in cleaning reusable glass containers, des- pite the fact that it is not the most effective sequestering agent of the three for calcium ions and the initially soluble complex which is formed may break down, particularly at higher temperatures, and foam an insol- uble calcium pyrophosphate. Pyrophosphate, tripolyphosphate, and the polymetaphosphates (glassy phosphates) can sequester iron and other heavy metals in addition to calcium and magnesium (Schwartz, Perry, and Berch ((1958)). The sequestering power of all condensed phosphates is influ- enced by pH, temperature, and the presence of other anions in the solution. Mehltretter, Alexander, and Rist (1953) have investigated the sequest- ering action of various hydroxycarboxylic acid salts toward calcium, iron, 20 and copper. They found the calcium sequestering ability to be relatively low over the pH range 4 to 10 at 25°C. A marked increase in sequestering capacity was noted with the 6-carbon acids and tartaric acid when the solutions were made strongly alkaline (2 to 51 sodium hydroxide). In the case of gluconic and mucic acids, maximum sequestration of calcium was not obtained in caustic up to 52 concentration. 0n the other hand, the sequestering capacity of saccharic acid was markedly reduced in concen- trations of alkali higher than 4%. The optimum sequestration with sac- charic acid was found in 2 to 42 sodium hydroxide. It was pointed out that when the concentrations of saccharic and gluconic acids were in- creased in 31 sodium hydroxide, their sequestering power toward calcium increased in direct pr0portion to their concentration. 0f the hydroxy- carboxylic acids tested they found that potassium sodium saccharate and sodium mucate gave the best sequestration of calcium and sodium lactate the poorest. With the exception of sodium citrate, it was found that the hydroxycarboxylic acid salts failed to sequester iron or copper to any great extent at pH 7. In acid solution (pH 4) the hydroxycarboxylic acids were found relatively ineffective as sequestrants for these metals ions. However the longer chained hydroxycarboxylic acids sequestered large amounts of iron and copper in the presence of alkali compared with the shorter chained acids. The authors point out that of the hydroxyc- boxylic acid salts tested only gluconic is used in considerable quantity in acid and alkaline cleaning compositions. It was also noted, that in general the hydroxycarboxylic acids examined exhibited maximum sequest- ration in 2 to 52 sodium hydroxide solutions. They state that on the basis of their results, saccharic acid is the best general acid sequest- rant in alkaline solution. Mehltretter and Watson (1959) state that the sodium salts of gluconic, glucoheptonic, and saccharic acids are being used in increasing quantities as sequestering agents. When dissolved in caustic solutions they prevent the deposition of lime scale on bottles and bottle washing equipment. They point out that caustic solutions of these sugar acids are also of considerable importance for removing rust stains and aluminum foil labels from bottles. They state that mixtures of organic acids produced by oxidizing dextrose hydrate and corn starch with nitric acid are rich in 21 gluconic and saccharic acids and.may contain other chelating substances. The oxidation liquors from various mole ratios of these reactants were evaluated for their sequestering action toward calcium, aluminum, and ferric ions in alkaline solutions. They reported good sequestration of calcium and noted that liquors produced by oxidation of one mole of the -. carbohydrates with two to four moles of nitric acid gave the best results. This was attributed to a higher proportion of saccharic acid formed under these conditions. The sequestration of aluminum and of iron was also reported as quite satisfactory with alkaline solutions, but was noted to be somewhat lower than that found with sodium gluconate. They point out that such oxidation liquors have the advantage of lower cost and on this basis might be the more economical sequestrant to use. . Chaberek and Martell (1959) have discussed the use of aminopolycar- boxylic acids in metal ion deactivation. They stated that these compounds form exceedingly stable alkaline earth chelates, as well as stable che- lates with many transition and heavy-metal ions. They pointed out that, "these chelating agents have a number of advantages over the inorganic sequestering agents: (1) The compounds are thermally stable in aqueous solution. (2) The alkaline earth sequestering capacity remains constant over a wide pH range, in contrast to the behavior of the polyphosphates and the hydroxycarboxylic acids.(3) The stabilities of the metal chelates of the aminopolycarboxylic acids are considerably higher than those of the inorganic chelating agents, and greater reduction in the free metal ion concentration occurs. (4) The organic sequestering agents are po- tentially more compatible with organic systems than inorganic sequester- ing agents." They stated that it is evident that these compounds are at least as efficient as, and generally more efficient than, the inorganic reagents represented by the condensed phosphates. Their primary limit- ation at present was noted to be their relatively higher price. An attempt is made in the following paragraphs to summarize some of the available information on the theoretical and practical behavior of the sequestering agents to be studied or of similar sequestering agents under conditions like or similar to those to be used in these investi- gations, in order to gather information to be used to predict or explain results: 22 Clements and Kennedy (1952) have presented the following comparison of the theoretical and practical sequestering power of the polyphosphates: Table 3. Sequestering power of polyphosphates. Sequestering agent Total amount of polyphosphate to soft- en 100 ppm of calcium carbonate water Theoretical,4ppm Practical, ppm Tetrasodium pyrophosphate, 133 800 Na4P207 Sodium tripolyphosphate, 148 417 N“51’3010 Sodium tetraphosphate, 154 384 ua(34°13 Sodium hexametaphosphate, 204 306 Na6P6018 Sodium heptaphosphate, 174 293 Na9P7022 The theoretical quantity of sequestering agent was calculated assuming all the sodium was replaced by calcium. However it was noted that all the sodium is not so replaced and that there may be quite a big discrep- ancy between the theoretical and practical figures obtained, as shown above. They also observed that the sequestering power increases as the number of phosphates in the complex increases. Chaberek and Martell (1959) have reviewed, summarized, and discussed the literature on the behavior of the polyphosphates, and the hydroxycar- boxylic and aminopolycarboxylic acids. The effects of a number of fac- tors on the sequestering properties of these agents as summarized by Chaberek and Martell are presented in the following figures: The effect of pH on alkaline earth metalion sequestration by sodium pyrophosphate, sodium tripolyphosphate and sodium hexametaphosphate is illustrated graph- ically in Fig. 1. They noted that increasing chain length increases the effectiveness of these polyphosphates and that the shapes of the curves were similar. In the pH range of about 7 to 9, and above 11 an increase in pH produced relatively little change in sequestration. In the inter- mediate pH range from about 9 to 11 the efficiency of all three phosphates 23 13‘5 12- l 11_: ”GPO l 10 +1 - pH 1 7 J I l l l i L L 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Grams of sequestering agent per liter of 1120 havi'ng unit hardness of 17.9 ppm Fig. 1. Effect of pH on sequestering action! of polyphosphates. Eros/'45P“??? s 0.7 ' - U a § 0.6 - J E . _ u ,3 0.5 o s - a: 0.4 P Ffrap‘asf/wt' d s 0 0'9. ' g ' . p / - O / , s E 0.1 ~ - I 0 5 10 15 20 water hardness, grains per gal. as Ca003 Fig. 2. Effect of water hardness on sequestering action of poly- phosphates. ‘ 24 decreases rapidly. The relative effectiveness of the polyphosphates is also dependent on water hardness as illustrated in Figure 2. Chaberek and Harrell pointed out that the amount of a particular polyphosphate required to soften water of varying hardness at 75 to 700C is not a linear function of water hardness. An increase in water hardness from 10 to 15 grains per gallon required doubling the amount of tripoly-. phosphate, while an increase from 10 to 20 grain hardness required about three times the amount according to the authors. A serious disadvantage of the polyphosphates, namely their tendency to hydrolyze to the orthophosphate anion (mentioned previously) is illustrated in Figure 3 for polyphosphates of different chain lengths. This "reversion" is dis- advantageous since it results in loss of sequestration ability, and in the formation of an anion having a strong tendency to precipitate metal ions. It was pointed out that reversion rates are dependent on solution pH, temperature, degree of dehydration of the polyphosphates, poly- phosphate chain length; metallic cations themselves tend to increase reversion. The relative chelating capacities of citrate, gluconate, tetraphosphate, and EDTA for calcium ions has been calculated from titration data obtained with an oxalate indicator. Chaberek and Martell have reviewed, summarized and discussed these data. The re- sults are shown in Figures 4 and 5. They noted in the pH range of 2 to 11 a low affinity of gluconic acid for the metal, which is shown by the relatively high ratio of sequestering agent to calcium required to maintain the metal in soluble form. Citric acid was considerably more efficient in the pH range 7 to 11, requiring on the average two moles of reagent per mole of metal. Tetraphosphate was described as the most effective ligand in the pH range of about 4 to 6, and between 6 and 7.5 it was comparable to EDTA which exhibited a relatively constant ligandzmetal ratio of unity in the pH range 6 to 11. It was pointed out from Figure 5 by the authors that contrary to sodium gluconate's behavior in the normal pH range,it is the most effective sequestering agent in the presence of free sodium hydroxide over a wide range of concentration. It was also noted that the citrate ion is considerably better than the inorganic polyphosphate anion. Organic sequestering agents are becoming more important and being 25 50 I I Polar} q/gqu Ira/f7: 3/355‘ 8 4° ‘ V! g Eggs: ,PIWI’A’E § 30- a D/ggs 1.2 s ‘6: 3 20 q H 0 ha 10 - o ‘ i‘ 0 1 2 3 4 Fig. 3. Reversion rates of various polyphosphates at 100°C. 12 I I i I r’ T’ 1' 1 11 - -___.9.l:’s.9.ea.§;____ - 10 ‘ ‘ 9 — cl 8 ‘ ‘ 7 - 4 Cthaai J Holes of sequestering agsgt required per mole of as I S P ‘ 4 r “ 3 - ' EDWA 2 ‘8 5., 1 1 Eta}, 43-." x‘ ”/ 0 I I I I l l l l 2 3 4 5 6 7 8 9 10 11 pH Fig. 4. Relative calcium chelating capacities of varioua.sequester- ing agents as affected by‘pH. 9 I I I u .4 8 8 ~ ~ g . g. Efrapdmfdafe 'o 7 — I I} u . ~43 a 6 b . m I... u 31 5 ” 7 $8 g? 4 - - u 3 3‘ 3 - .. :I 3' Q 2 __ . q "a (jib-ate. fi 1 _ - § 0 l J J 0 S 10 15 20 Z NaOH Fig. 5. Relative calcium chelating capacities of various sequest- ering agents as affected by sodium hydroxide concentration. ,used in ever increasing quantities as water-softening agents. Such is ’ indeed the case in the carbonated beverage industry where these agents are used in cleaning bottles. Concern has existed for some time as to what effect these agents might have on the washer itself. It is chiefly this concern, which has prompted this comparative study of the corrosive- ness of these organic sequestering agents relative to that of the more commonly used inorganic phosphates and other components of the system (water, alkali, and hardness-forming salts in water). An indication of’the influence sequestering agents might have on washing equipment, in addition to their effective alkaline earth metal sequestration, is apparent from discussions of their use in metal clean- ing. Chaberek and Martell (1959) have noted that alkali metal salts of aminopolycarboxylic acids such as EDTA, used with effective ferric- sequestering agents such as gluconate, citrate, and triethanolamine, should be effective in formulations for scale removal in alkaline sol- 27 utions. They also pointed out.that in neutral or alkaline solutions, rust may be removed by the use of an iron-sequestering agent in conjunc- tion with a suitable reducing agent. They stated that EDTA is especially suitable for this purpose since it forms stable chelates with ferrous ion as well as with ferric and other metal ions. Aiken and Garnett (1957) have also noted that one of the outstanding properties of the EDTA group of chelating agents is their ability to "dissolve” water-insoluble metal oxides and hydroxides, carbonates, sulphates and phosphates which are formed when metals corrode. It was stated that EDTA is valuable in special-purpose cleaning compositions for metals and also as an additive to conventional alkaline cleaners. An effective industrial process using EDTA.was described. Smith (1959) has pointed out that the use of sequestering agents can be extremely valuable in obtaining a clean metal surface which is it- self a primary requisite for any successful metal finishing process. He noted that both meta-silicate and polyphosphates have traditionally been used in metal cleaning processes and that EETA can be used to advantage along with them. Further, it was noted that EDTA.may be effective as a sole agent, particularly in conjunction with caustic soda. He also noted that there are hydroxycarboxylic acids which.sequester iron in strongly alkaline solution and in acid conditions, and that the commercially import- ant acid has been gluconic which is used in industrial processes in strongly alkaline media. Gluconic and tartaric acid are also used in aluminum etching solutions to prevent the formation of hard and adherent scale. Smith stated that the uses of citric, gluconic, oxalic and tar- taric acids are well established for cleaning, polishing and plating of metals, and a high proportion of these uses are based on the sequester- ing properties of these acids and their salts. The fact that sequestering agents are used in metal cleaning to pre- vent and/or remove scale would suggest that they might accelerate corro- sion, since corrosion products often function as an important corrosion controlling (inhibiting) factor. The predominate cell reaction in oxygen-type corrosion of iron is 2Fe + 02 + 2820 = 2Fe(0H)2$' The ferrous hydroxide formed on the surface of the metal is insoluble 28 and functions as the main corrosion rate controlling factor, since it limits the amount of oxidizing agent available at the cathodes for reduc- tion (the oxygen diffusion rate is controlled by the ferrous hydroxide film). Sequestration (solubilization) of the ferrous hydroxide film could be expected to increase corrosion. This sequestration of corrosion products can be expressed by the following general reaction (used by . Chaberek and Martell (1959) in their disucssions) which for purposes of simplicity and convenience does not show the valances involved, and musé'mna where B is a precipitating agent and A is a sequestering agent, which combine with the metal to give the precipitate MB and soluble chelate MA respectively. Chaberek and Hartell note that the exchange constant Xx is related to KMA: the metal chelate formation constant, and Run: the l - FHA K MB . that the tendency toward precipitation of the insoluble metal salt, solubility product of M3, by the equation Rx It is seen measured by Rx is a function of both the solubility product and the metal chelate formation constant, and if K.x is sufficiently small, the precipitation of MB may be prevented. It was also pointed out that the equilibrium constant, Xx,, measures the tendency of the chelating agent to solubilize slightly soluble MB and is equal to the reciprocal of Xx. Aiken and Garnett (1957) point out that this is a simplified picture of the actual situation and ignores not only the effect of conditions such as pH and concentration, but also the particle size and crystal form of the insoluble compounds. They note that it is generally easier to pre- vent precipitation than to dissolve a precipitate that has had time to consolidate and crystallize. Nevertheless it was noted that these re- lationships may sometimes be used to predict the possible usefulness of a sequestering agent on the basis of the magnitudes of the solubility product and the metal chelate formation constant. In general, for the systems and conditions used in this investiga- tion the relationship for predicting solubilization of MB (predominately Fe(0H)2) can notbe used since quantitative information on the solubility product and chelate formation constant is either not available or that which is available can not generally be adjusted to take into account 29 the deviations the systems and conditions impose. This problem has been expressed another way in a bulletin entitled "Keys to Chelation" (1959)- in which it was stated that, "most industrial chemical systems which bene- fit by the addition of metal-control agents are not simple systems of water and the metal ions. If they were, the selection of the—proper agent and the most efficient use concentration would be a clear-cut de- cision based on the stability constants. It would then be possible to state definitely that so much of an agent would do a particular job. Un- fortunately, complicating factors due to other materials in the system- profoundly affect the metal-chelation reaction." The following.environ- mental factors were noted to disturb the simple metal-chelate equilibrium: (1) Temperature, (2) Other metal ions, (3) Other anions, (4) Ionic strength effect, and (5) pH. It was pointed out that of all these fac- tors, and aside from the stability constant, the most important factor to be considered is the pH of the system. Since it seems reasonable to expect that water softening agents would affect the nature and distribution of corrosion products this in- vestigation is conducted to determine if such is the case and to what relative extent agents (sequestering agents) either aCcelerate or in- hibit corrosion of mild steel, in systems commonly employed in the cleaning of reusable glass containers. EXPERIMENTAL DESIGN AND ANALYSIS 1 The purpose of this investigation is to assess the corrosion inhib- iting or accelerating effects of organic sequestering agents, relative to the effects of the more traditionally used inorganic phosphates and other components of the system under conditions similar to and varied in accordance with those commonly encountered in practice. The general experimental design to be employed in this investigation is shown in Figure 6. The treatments consist of different agents, agent combinations and agent concentrations. The blocks or outcome groups consist of five consecutiVe seven-day immersions. Each cell or experi- mental unit consists of four observations (or replications). A symbolic representation of such a two~way table with t treatments, b blocks, and equal numbers of observations in each cell is presented in Table 4. Table 5 gives a symbolic description of analysis of vari- ance of a two-way table with t treatments, b blocks and equal numbers of observations in each cell. The notation employed is rather general in statistics. A dot in a subscript indicates summation. The k th ob- servation made in the j th block on the i th treatment (1 = l...t, j = 1...b, and k’z 1...n (n = 4)) is denoted by xijk' Additional con- venient notation employed here follows: xij = the sum of the observations in the cell which is located in the i th treatment and the j th block. x1. is the sum of all observations in the i th treatment and x.j the sum in the j th block. X is the sum of all the observations. For the preper evaluation of experimental data the analysis of vari- ance model employed must be specifically stated. Steel and Torrie (1960) and Snedecor (1956) discuss the various models in considerable detail. Discussions appropriate to this investigation are presented by Steel and Torrie in their chapter on analysis of variance of multi- way classifications (Section 8.8 Linear models and the analysis of vari- ance.) and by Snedecor in his chapter on factorial arrangements of treatments (Section 12.5 The two-factor experiment.). Both references present component analyses (variance components) and procedures for test- 30 31 llll [HI x? llll llll llll llll llll llll Illl llll In NH llll llll llll llll |||| llll I I I I I I I I N III II II. III III .I... II .II A a 3m «Um Hon 3N NON . .nioN . .— I o flunowmusgfi. - assuaga ouaausuuu. , «300.3 . 2.3.3 . now—sun: succession: 330353 .5221 message 538 an 3 a 2%:238923 3333a .533 it «399. «Sue—e3 838 an 3 mouauogufiuufiaflfiaflafio 133833 «3.3+ «Excess 538 an 303 Imousuoussnuouogounuahfiu Ianwooflwwug .336... saucepan 833m fin A83 mousuuuasuuuuongv -833“... 530833 585 + «3.3.63 .538 .3 sec "3.283% 338 «.819. «383.3 52.8 .3 God «3.2303» .538 .586 + «3383 538 an 308 mongoose» 3:3. .2399. «3.383 .538 an 203 3336.2 .588 an 3 £32. consumes 8v “noses at? 832:. 232585 .o .3; 32 Table 4. Symbolic representation of a two-way table with t treatments, b blocks, and equal numbers of observation (n = 4) in 3.“ C211, k g 1000“. ‘ T— Treat- Treat- Treat- ments ---——Bln§kfl+—1—= 1°'°b ment ment M _l__ . . . J . . . b sums means 1 11111 :11 1 inn :112 x112 11le 113 x133 x11:3 x114 134 1b4 . 1 13 b 1 . - ”an“ ‘11 111 71b x1. 1 £111 £111 §1b1 12 x112 Xibz 13 R133 x11:3 14 134 1b4 Sums X X X X 11 -13 -11: 1 . - Means 7111 X13 31b 111 . t §t11 £131 :tbl :12 x‘ 32 xth :‘13 x$.13 xtb3 :14 :34 cb4 Sums X X X x -tl tj .-tb t . - Henna x“ 2‘1 1‘tb x: .3 Block sums X 1 x j x‘b X .- Block means 2'1 2'1 le X in; main effects and interaction effects for the various models. Steel and Torrie's presentation is more recent and easier to follow. Common models of analysis of variance (Steel and Torrie (1960) and Snedecor (1956)) are the fixed effects model or Model I, the ran- dom effects model or Model 11, and the Mixed model. The Mixed model calls for at least one criterion of classification to involve fixed effects and another to involve random effects. Steel and Torrie (1960) state that, "in randomized complete-block 33 Table 5. Symbolic description of analysis‘of variance of a two-way table with t treatments, b blocks, and equal numbers of ob- servations in each cell. — Source o} Degrees of Sum of Computational'formula variation ' freedom ‘ sguares for sum of sgusres Total ntb-l §§;O(1,1"X) Zzéxij" 5% Treatments t-l 01.4;(1-5-9‘ -— mix/t- mil- Blocks b-l mi: (2" 'X) #ZX?’ -- €1- ‘ 1 Interaction (t-l) (b-l) mfZ.( 533.4 1.7)" 1:4??? _;‘2x‘ :Lz Inga Error Cb (ti-1) Z ’2 (X - . .. J. " - EH X1‘ X1!) 22:2:ZX":‘ ’” 4 fix”! design problems where the fixed effects model is appropriate, two or more treatments are selected for testing. These are not randomly drawn from a population of possible treatments but are selected, per- haps as those which show the most promise or those most readily available. All treatments about which inferences are to be drawn are included in the experiment. Block effects are also fixed and inferences about treatments or blocks are not intended to apply to blocks not included in the experiment." Steel and Torrie point out that the distinction between the models is that for the fixed model, a repetition of the experiment would bring the same set of treatments (or components for the populations involved) into the new experiment and our attention is concentrated on these treatments. For the random model, a repetition would bring in a new set of treatments but from the same population and we would be interested in variability of the treatments since we are not in a position to continue our interest in a specific set. With the fixed model we draw an inference about the population of treatments. Steel and Torrie (1960) discuss the testing of main effects for these models. They point out that when the sources of variation are assumed to be random, the experimental error (residual error) is the apprOpriate error for testing hypotheses concerning main effects. They go on to point out that in the case where main effects are assumed to be fixed, one does not necessarily assume that the ex- 34 perimental error is random. Further, when nonrandomness is assumed, essentially there are fixed effects for each block-treatment combin- ation (where blocks and treatments are the main effects) over and above the treatment and block contributions. This can be stated another way by saying that the differences in responses to the var- ious treatments are not of the same order of magnitude from block to block and this cannot be called a random effect. Steel and Torrie point out that when the contribution referred to as experimental error is not random, they relabel it as interaction and that the sampling error is an appropriate error for testing hypotheses. concerning interaction and hypotheses concerning main effects. Steel and Torrie present a table of average values of mean squares for a randomized complete-block analysis to show what sort of conclusions can be drawn from tests of significance, in the form of ratios of mean squares. Included in this table are the linear models for the fixed effects model with and without interaction, and with and without sampling ( , an observation from the j th x 1.1 block on the i th treatment, 1 = l,..., t treatments and j = 1..., r blocks; ethe R th observation made in the j th block on the x1 jk’ a 1th treatment, i=1’ses’ t, 1:1,000’ I" andk=1’oos, 80b88r' vations; /u, the population mean; 71, a treatment contribution; Qj’ a block contribution; (7%)“, an interaction contribution; 11 The portion dealing with the fixed effects model or Mbdel I follows in Table 6. Steel and Torrie state that, "for the fixed model with no inter- (ij’ an experimental contribution; (F. k’ a sample contribution). action, both block and treatment effects can be tested by residual or error mean square. Even when we have only one sample, S =\l, per experimental unit, valid tests are possible. In this case, it is not possible to estimate 5:" However, when there are fixed inter- action effects, it is not possible to make any valid F test when there is no sampling. When s observations are made in each cell 37%, sampling error is a valid error for testing interaction, treatment, and block mean squares." ‘ For the experimental design employed in this investigation the 35 lb [1| no Aanmv up uouuo newsmamm “a... No. N4. w 3 \Vxé\dx\ $35» Neflb Q-V\$-s§\.w®¢N*alb 3.32.: 2323 fitment spurge E>fii 3;: 3.5..3é3E s.§%€§ mewanaom I IlllllllwmmHmmmn oz aoooouw mo coauoououog .aoooa voxwm «common .oousom axlkv Afiunv uu nouns mawagaum dunks c. i .75 . No 3-33-: 3.3.3 \\ JNV\.NL..N~\H+ leflrbhw NIB \\ \VQ\..N~. No\$.~lw..b Tu eucuauooua saiber hurl :éxbflit 7... .3. .HfiWXY‘I‘HWs‘.&&.§.Nwlr1\fiu\KWVA .Www.+m\& $.NV+.3\u\JvA wwwdnasm .Illllllludwaeaon oz aovooum no sowuoououaw on qaovoa vexam «common sousom r vow Assam aoumV nwmhauco xuoHAIouonEoo ounwaopcwu w you noum:Vn news .A .e «Haas «accede «autos no woman. owwuo>< .9 «Hana 36 fixed effects mode1(Model 1) seems most appropriate for the shale yses. The treatments and blocks were not randomly drawn from a pop- ulation of possible treatments or from a population of possible blocks, but were selected for testing. Treatments and blocks about which inferences are to be drawn are included in the experiment.- Block and treatment effects are therefore considered to be fixed. The experiments conducted in this investigation are of the two-way classification (or two-factor) type, with agents as the treatments and immersions as the blocks. The linear equation re- - presenting a single observation (or replicate corrosioneweight loss) is, xijk 8* + rr1.+ 61+ (1%)“ + ijk' The i th agent‘over:' all mean can be represented by X1. = l“ + ‘I’1 + .+ (3)1. + J1. and the j th immersion over-all mean can be represented by the X.j = /u+7'. +61 + (EL-1+ J‘j' These equations show the components of an agent or immersion over-all mean and what is in- volved in agent and immersion over-all mean comparisons. When fac- tor (agent or immersion) over-all means are tested for significance and the model is conformed tojthe components of the over-all mean that are deviation means, contribute essentially a zero effect to the difference between the factor (agent or immersion) over-all means being compared. DEVELOPMENT OF EXPERIMENTAL PROCEDURE INTRODUCTION Corrosion consists of reactions between a metal and its environment, thus, corrosion behavior is a property both of the metal and environment to which it is exposed. Factors associated with both phases must be con- sidered and controlled to establish satisfactory exposure conditions dur- ing a corrosion test. There is, therefore, no universal standard test procedure. Literature pertaining to the development of a satisfactory proced- ure for the evaluation of corrosion of mild steel in alkaline sequester- ing agent solutions follows. REVIEW OF LITERATURE There are certain characteristics of this investigation that may yield to an established general method of corrosion testing. The followa ing literature provides a review of the generally accepted corrosion test- ing methods and is presented to provide a basis for the methods employed. Jelinek (1959) and Borgmann (1948) summarize the major methods for measuring the amount and intensity of corrosion as well as certain qual- itative aids in determining the amount of corrosion. They both point out that the use of two or more methods will remove many of the criti- cisms given for the individual methods. A brief description of the in- dividual methods outlining the fields of usefulness and the major advant- ages and limitations follows. Visual observation may be useful to detect occurrence of attack and identify its general nature; it is a simple method and is often valuable in conjunction with other methods; however it is subject to human error and is qualitative. Loss in weight is useful to determine the extent and rate of uniform corrosion; it is simple, quantitative and direct. However, the method is subject to error due to incomplete removal of corrosion products and loss of uncorroded metal, special types of attack are not measured, and mul- tiple specimens are necessary.‘ Gain in weight is useful to determine the extent and rate of uni- 37 38 form corrosion where no loss of corrosion products occurs. It is partic- ularly applicable to indoor atmospheric corrosion and high temperature oxidation studies, and error due to improper removal of corrosion prod- ucts is eliminated. However, an analysis of corrosion products must be made in order to determine loss of metal, moisture in corrosion products may vary in amount and thus influence the results, and special types of attack are not measured and accidental loss of corrosion products would introduce an error. Electrical resistance change can be used in gaseous or poorly con- ducting environments or in other environments if the specimen is removed for measurement. In addition to its usefulness in evaluating environ- ment corrosivity it can be used to follow certain property and composit- ion changes, and the method is non-destructive and adaptable to con- tinuous measurement. However, the method is indirect, requires cal- ibration, does not distinguish between types of attack, is subject to surface-to-volume errors, and if small wires are employed, the attack may be different in amount from that on a more massive specimen. Hydrogen evolution may be followed for tests in which corrosion takes place solely with hydrogen evolution. The method is adaptable to rate measurement, but does not determine the distribution of attack. Oxygen absorption is useful in tests where corrosion takes place mainly with the absorption of oxygen. The method is adaptable to rate measurement, but does not determine the distribution of attack, and an analysis of corrosion products is required for metals capable of exist- ing in more than one ionic state. Depth of pitting measurements (other than microscopic) are useful for tests made to determine the serviceability of metals as containers of fluids. The method is especially adapted for use with methods of determining total attack, and it gives a correct measure of penetration of a metal by corrosion except when the attack is intergranular. However, many specimens are required to determine time-penetration curves, and difficulty occurs in obtaining accurate measurement. Microscopic methods are useful to determine the kind of attack, to measure the depth of pits, and to determine the constituents of the metal that are specially capable of initiating attack. It is an excellent tool 39 to supplement other measures, but is not generally useful in making quantitative measurements. Physical property changes of material can be used to evaluate their deterioration; methods that can be used for this purpose are tensile strength, ductility, impact, resistance, and hardness. This method allows a direct measure of the changes in physical properties and hence is of practical value to the structural engineer; however, a resultant of the several possible attacks is measured and it is not possible to evaluate the damage resulting from each type of attack separately. The electrochemical methods that may be useful are: measurements of single electrode potential, potential difference between unlike metals, shorted-cell current, anode and cathode polarization, and film resistance. Single electrode potential methods are useful to study film formation and breakdown at a metal surface. The method distinguishes between anodic and cathodic control, and measures the electrochemical driving force. It does not measure extent or rate of attack and careful interpretation of results is required. The measurement of the potential difference between dissimilar metals is a useful method to study galvanic effects; it measures the relative tendency to corrode, but is qualitative only. Shorted-cell current measurement is useful to measure the extent of corrosion relative to a standard noble metal. The measurement is sim- ple, but an arbitrary choice of a cathode metal may distort the normal influence of such areas on the metal under study. The anode and cathode are separated by a much greater distance than usual, hence resistance of solution and the unnatural formation of corrosion products are sources of error. Anode and cathode polarization methods are useful to study galvanic and concentration cell corrosion and to determine the total polarization current. The method yields a semi-quantitative estimate of the corro- sion rate, but does not measure the distribution of attack. Film resistance measurements may be used to determine the penetras bility of surface films by various anions. A qualitative measure of the influence of the anion on the probability of breakdown of the film is ob» tained. However, an arbitrary standard voltage is usually employed, and 40 care must be taken that other reactions do not interfere. One may measure the decomposition voltage of an electrolyte~meta1 system without reference to a surface film. The electrometric method is useful to measure the thickness of sur- face films. The technique is simple and quite exact, but only useful on adherent, thin, surface films of some metals. Optical methods are useful to study the growth of tarnish and other surface films. Formation and growth of films can be studied with~ out disturbance of the system. Relatively complicated apparatus is re- quired. Results of simpler relectivity methods are difficult to intera pret quantitatively. Environmental analysis is useful to study product contamination from corrosion and to follow relative corrosion of metal components. The analysis may be necessary for product specifications. The method is often applicable to measurement of trace quantities and is useful in liquid metal corrosion studies. It does not show the distribution of. attack on the metal and a material balance is required to evaluate liquid metal corrosion. It was decided from the above information that the weight loss and visual observation methods would be used for this investigation. The weight loss method facilitates a simple, direct, quantitative measure of the extent and rate of corrosion. The visual observation method which is simple and of value in conjunction with other methods is useful to detect the occurrence of attack and identify its general nature. The principal limitations of the two methods are that multiple specimens are required and that caution need be exercised in the selection or develop- ment of a treatment to remove corrosion products. Evans (1960) points out that in wet corrosion, the measure of the progress of attack is often the loss of weight after exposure. Fontana (1960) states that, "change in weight of the specimens is most often used as a measure of corrosion or the basis for calculation of the corro» sion rate." Champion (1952) also points out that the gravimetric method is normally used for determining the metal remaining after corrosion. Corrosion being a complex phenomenon, influenced primarily by fac- tors pertaining to the metal and the environment, makes it generally 41 impossible to have any one detailed corrosion test procedure that will fit a number of different corrosion problems. Once the particular method has been chosen, it is best to acquire a general understanding of the basic principles underlying the detailed procedures commonly employed in the particular method. With this understanding the investigator is ~ then in a position to develop, with the aid of the available literature, a detailed experimental procedure best suited to the particular problem. In some cases the recommendations available in the literature pertaining to a particular detailed procedure are adequate, in others modifications are required, and in still others new detailed procedures have to be developed in order to obtain the best over-all experimental procedure. It is the purpose of the literature to follow, and the results pertain- ing to the development of the experimental procedure presented later, to provide a background and basis for the utilization of particular detailed proceduresto be followed in this study. The value and reliability of the data obtained are dependent upon the details involved. Champion (1952), Borgmann (1948), and Speller (1935) present a com- prehensive discussion of general corrosion testing procedures based on the literature in this field. They discuss the general procedures to be followed in total immersion corrosion tests. The American Society for“ Testing Materials (ASTM) presents specific detailed test procedures for use to evaluate a number of corrosion problems. A review of factors per- taining to the metal, corrosion media, exposure conditions, and assess~ ment of effects on the metal and media important in the development of the experimental procedure follows: Factors pertaining to the metal are selection of the specimens, size, and shape of specimens, replication and number of immersion periods, prep- aration of specimens, and cleaning of the specimens after corrosion. Champion (1952) points out that test specimens should be representative of the material under test. He states further, "that rolled products are most widely used because they provide the following advantages: (1) They are most extensively used in service. (2) A wide range of thicknesses can be obtained. (3) Major variations in corrosion resistance over the surface of 8 rolled sheet are unusual. (4) Sheets of large surface area can be obtained thus permitting the preparation of a large number of 42 replicate specimens and also analytical samples from the same sheet. (5) Relatively simple methods of preparation of the specimens from the sheet can be employed." The ASTM (1952) in discussing test specimens and their preparation states that, "the size and shape of specimens will vary with the purpose of the test, the nature of the materials to be tested, and the testing apparatus to be used. The size may also be limited by the necessity of preserving a proper ratio between the area of the specimen and the volume of the testing solution when the latter must be limited. In general, an effort should be made to have the ratio of surface to mass large and that of edge area to total area small. The shape and dimensions of the specimens shall be such as to permit weighing on an accurate balance and to facilitate accurate measurement and calculation of the area of each specimen. Such measurements of dimensions shall be made accurately to the nearest 1/64 in. (0.5 mm) for length, width or diameter; and to the nearest 0.001 in. for thickness." Champion (1952) and Knapp (1948) point out that the size and area adopted is often determined by experimental convenience, but that it is desirable to adopt a standard size as far as possible to insure that comparable results will be obtained. They suggest the use of simple shapes as being most desirable. In addition, it is advantageous to adopt a size and shape of specimen giving a low edge to surface area ratio and a reasonably large area. Champion (1952) states that, "a thickness of 0.036 in. is convenient for many purposes and that the French Aeronautical Committee have adopted 0.039 in. (1 mm) for sheet thickness." He also states that, "the German specification on corrosion tests in general recommends that the dimensions should be not less than 5 x 2 cm." Fontana (1960) points out that the specimen should be carefully measured to permit calculation of the surface area, since area enters in the formula for calculating the corrosion rate. He states that, "the original area is used to calculate the corrosion rate throughout the test. If the dimensions of the specimen change appreciably during the test, the error introduced is not important because the material is probably cor- roding at too fast a rate for its practical use in the intended application. 43 Borgmann (1948) points out that even when the greatest pains are taken in the corrosion test it is not to be expected that absolute reproducibil- ity will be achieved, since there are factors beyond the control of the experimenter. However, statistical control is often achieved. The ab- solute error of measurement can be reduced only be increasing the number of specimens tested. The type of attack that occurs can greatly influence the reproducibility of the results. The reproducibility will in general be much greater if the attack is uniform than if the attack occurs at a few discrete points. Evans (1960) points out that in general, it is well to perform all experiments in duplicate or triplicate. He notes that in the case of localized corrosion this will not be enough replicates for good reprod- ucibility. The ASTM (1958) standard method of total immersion corrosion test for soak tank metal cleansers recommends that at least two, and prefer- ably four, replicates shall be tested in each cleaner solution. Champion (1952) states that, "a minimum of three replicates has been recommended as a general rule." He also points out that when it is known that the corrosion rate is constant from the beginning of exposure, a single replicate set of specimens will suffice. With non-linear corro- sion time curves further replicates are necessary, and where the shape of the corrosion time curve is not known, it is recommended that at least three replicate sets should be exposed for withdrawal after'increasing periods of time. However, the degree of replication to be adopted will depend on the accuracy required in the mean result and the scatter which is to be expected in the individual results. An ASTM (1952) method, which describes procedures for making total immersion corrosions tests, discusses the problem.of determining the number of specimens to be used from a statistical approach. They state that, "observations on some typical metals immersed under controlled conditions in typical corrosive media have indicated that the coefficient of variation may reasonably be expected to be less than 7 percent." With such a coefficient of variation, the expected limits of error of the average of groups of various sizes can be estimated for a desired statistical probability. They refer to the work of Humes, Passano, and 44 Hayes (1930)'when presenting their statistical procedures.. They state I that, "E, error of average in percent = -§!' where: Z = 1.96 for a- statistical probability, P, of 0.95; V' a universe coefficient of vari- ation of 7 percent; and N = number of repititions or specimens._ There- fore, the expected errors of the average of different numbers of the specimens would be N = 8, E ”.1 4.8 percent; N = 4, E =‘1 6.9 percent; and N = 2, E ”.i.9-7 percent." It was stated that, "running total immer- sion tests in quadruplicate would make it highly probable that the aver- age would be within 7 percent of the true average. This should be satis- factory for most corrosion research programs." Care must be taken to properly clean and prepare specimens for tests. The surface preparation will vary with the type of test. Surface preparation usually consists of a pre-cleaning or degreasing operation to remove surface contamination. A pickling process (pickling is the term applied to the chemical removal of surface oxides from metal by immersion in an acid solution) may be required to remove scale and pro- vide a more uniform, reproducible surface condition. Fontana (1960) states that ideally the surface of the test specimen should be identical with the surface of the actual equipment to be used in the plant. However, this is usually an impossible condition, because the surface of commercial metal and alloys vary as produced and as fab- ricated. The degree of scaling or amounts of oxide on the equipment varies and also the condition of other surface contaminants. Because of this situation and because the determination of the corrosion resist- ance of the metal or alloy itself is of primary importance in most cases, a clean metal surface is usually used. A standard surface condition is also desirable and necessary in order to facilitate comparison with re- sults of others. Champion (1952) points out that extremely thin, invisible films of grease can interfere with the corrosion process. Evans (1960), Jelinek (1959), ASTM (1952), Knapp (1948), and Meyer (1948) also note the import- ance of removal of grease and other surface contaminants. Champion (1952) and Meyer (1948) discuss in considerable detail the general methods for metal precleaning and degreasing. The solvent and vapor cleaning methods are: simple solvent cleaning, emulsifiable sol- ,45 vent cleaning, and vapor degreasing. The ASTM (1958) and Champion (1952) describe in detail various apparatus, cleaning solutions, and laboratory procedures for degreasing of corrosion test specimens. Evans (1960), Champion (1952), ASTM (l952)and McCallam and Warrick (1948) suggest that pickling of the specimens may be required to remove surface scale and provide a uniform, reproducible surface. Wesley (1943) found it necessary to remove 0.0003 in. thickness from the surface of some commercial alloy specimens to obtain adequately reproducible results. Hoar (1948) obtained reproducible results from the corrosion of steel in acid when the specimens were pickled. McCollan and Warrick (1948) point out that wide variations are poss- ible in the type, strength and temperature of the acid solutions used in pickling. In addition, pickling inhibitors may be added to acid pickling solutions. They are agents which diminish the attack of the acid on the metal areas from.which the scale has been removed, without appreciably retarding the rate at which the acid removes scale or rust. The pro- cedure used depends on the material to be pickled, the character of the scale involved and the surface desired after pickling. Evans (1960), Jelinek (1959), Champion (1952), ASTM (1952), Knapp (1948) and Speller (1935) all point out that in tests where the corrosion is assessed by weight loss, it is necessary to remove the corrosion products from the specimen after exposure. It is essential that the specimens be thoroughly cleaned of all corrosion products without loss of any base metal. The literature points out that there are many satis- factory methods of cleaning specimens but, whatever the method, its effect in removing base metal should be determined for each material. The ap- praisal of the amount of attack should determine the cleaning method. Drastic methods should not be used when there are indications of only a small weight loss resulting from the corrosion test. Champion (1952) and Knapp (1948) discuss in detail the general meth- ods for corrosion products removal from metal specimens. The various methods of corrosion products removal may be classffied as: Mechanical treatment by such means as wet scrubbing with a bristle brush, abrasive and detergent (a satisfactory method for removing light non-adherent 46 films). Chemical treatments using various chemical reagents are often employed, but are generally specific for certain materials. Electrochem- ical treatment, with the specimen as cathode,has been found useful with a large number of metals and alloys, but the possibility exists that de- position of metal from dissolved corrosion products or contaminants in the solution may occur. Champion (1952) and Knapp (1948) present specific procedures taken from or based on available literature for corrosion products removal from iron and steel. Champion describes the use of such solutions as sulphuric acid, Clarke's solution (composed of hydrochloric acid, antimonious oxide, and stannous chloride), ammonium citrate solution,- and sodium hydroxide solution for simple chemical treatment of iron and steel corrosion products. The ASTM (1958) and Speller (1935) suggest immersion of the specimens for 2 to 3 minutes in a boiling solution of ammonium citrate (10 percent by weight) for removal of corrosion products from iron and steel. Evans (1960), ASTM (1958) and Speller (1935) suggest the use of a warm dilute acid solution containing sufficient inhibitor to prevent attack of the base metal. Evans (1960) states that, "for the removal of the corrosion-product, inhibited acid is generally recommended; preliminary experiments are necessary to find the lowest concentration which will remove all the products without seriously attacking the metal." The corrosion media or environment is another important factor in corrosion. Champion (1952) presents a general discussion of corrosion media, based on the literature. He points out that the choice of corro- sion media depends upon the test objectives and that in fundamental re- search it is often important to avoid complications as far as possible, so that a simple pure solution or liquid may be required. With the aid of the literature he lists the following characteristics of the environ- ment which should be considered and points out that their relative import- ance varies with different systems: (1) Composition. (2) Concentration. (3) Concentration gradients. (4) Particle size of any solids constituting the whole or part of the environment. (It was pointed out that the in- tensity of corrosion tends to increase to a maximum with increasing part- 47 icle size, while the area of attack is more likely to decrease.) ~(S). Humidity of gases and vapor. (6) Frequency of renewal or replenishment of the corrosive. The ASTM (1952) also points out that test solutions should be made up accurately, with the composition controlled to the fullest extent possible and that the composition of the test solution should be deter- mined by chemical analysis. A review of factors pertaining to exposure conditions follows: Wesley (1948) states that, "perhaps the most common danger of mis- taken interpretation of laboratory immersion test results arises from ex- haustion of ingredients in the original solution which control the rate or type of corrosion, or from accumulation of products which are corro- sion inhibitors or accelerators." This emphasizes the importance of vol- ume of solution per area of specimen and adequate replenishment of sol- utions. Listed in Table 7 are some of the recommended ratios for volume of solution per square centimeter of metal surface. Table 7. Recommended ratios of solution volume to metal surface area. Source Volume per Surface area per square centimeter 500 milliliter "Corrosion" (1956) 50.0 ml 10.0 cm2 Wesley (1948) 40.0 ml 4 12:5'cm2 Champion (1952) 33.3 ml 15.0 cm2 Calcott, Whetzel, and 8.3 ml 60.0 cm2 Whittaker (1923) ~ as'm (1'958) 6.7 m1 75.0 m2 Champion (1952) also pointed out that the volume of solution recum- mended per square centimeter of metal surface has varied from 6 ml to 40 ml. The important point is that the volume of the testing solution should be large enough to avoid any appreciable change in its corrosive- ness either through the exhaustion of corrosive constituents, or the accumulation of corrosion products that might affect further corrosion. Wesley (1948) and ASTM (1952) point out that each specimen should be tested in a separate container, since it has been found that the practice of testing several specimens of the same material in a single 4s container gives results Which do not measure the true variability of the test as determined by repeated separate tests. In—addition, containers and specimen supports should be used which do not affect the corrosion process. Temperature control of the corrosion media is another important exs posure condition to be considered. Champion (1952) suggests that temper- ature control to‘i’0.5°C (1 0.9OF) is often regarded as adequate for many corrosion tests. Wesley (1948) and ASTM (1952) point out that thermostatic control at the desired temperature,‘1_loc Qt 1.80F) is us- ually considered satisfactory. An important consideration in any corrosion test is its duration and the number and length of exposure periods (or immersion periods) to be employed. Speller (1935) presents the following discussion of corrosion test duration. He points out that in general the initial rate of corrosion is much greater than after the action has prOceeded for some time. He states that, "one good example of this is found in alkaline solutions where the initial rate may be several times greater than the rate after a few hours." Calcott and Whetzel (1923) state that in many cases of chemical corrosion the rate becomes essentially constant after 48 hours. Speller mentions further that in any case, the retardation that follows the initial rate of corrosion is dependent to a large extent upon the character of the corrosion product deposited on the metallic surface, and that this should be determined for a specific case and allowed for, as otherwise the results of short-time tests may be very misleading. Speller summarizes by pointing out that in all cases the length of test should be determined by the length of time required to produce a reason- able amount of corrosion. Wesley (1948) states that, "a constant rate of corrosion is en- countered much more often in total immersion than in other types of corrosion." In addition, he points out that if the test is too short, some materials which build up protective corrosion product films slowly may be ruled out as unsatisfactory and if the test is too long, the ef- fects of exhaustion of ingredients or accumulation of corrosion products may be pronounced. 49 The ASTM (1952) points out that the duration of the corrosion test will be determined by its nature and purpose.- They mention further that in some cases it will be desirable to expose a number of specimens so“. that certain of them can be removed after definite time intervalsusogas to provide a measure of change of corrosion rates with time. In add- ition, they indicate that any procedure that requires removal of solid corrosion products between periods of exposure of the same specimens » will not measure accurately normal changes of corrosion with time since the effect of corrosion products on subsequent corrosion is not evalu- ated. They suggest also that the higher the rate of corrosion, the shorter may be the testing period. Fontana (1960) also points out that proper selection of exposure time and number of exposure periods are important, and misleading results may be obtained if they are not considered. He states that in most 'cases at least two periods should be used, since this procedure provides information on changes in corrosion rate time and may uncover weighing errors (if a test consists only of an original and a final weighing, an error in either case might go undetected.and be reflected directly in the results.) He points out that the corrosion rate may increase, de- crease, or remain constant with time, but quite often the initial rate of attack is high and then decreases. He notes further that a widely used procedure in the laboratory is to use five 48-hour periods with fresh solution for each period. He suggests that the test time should be reported particularly if exposure time is short. The next important phase is assessment of the corrosion., Champion (1952) has reviewed and discussed the assessment of corrosion effects on metal and media. He includes considerable information on quantitative _and qualitative assessment by gravimetric and macroscopic methods,-re- spectively. Champion notes that the frequency of inspection or examin- ations depends on the metal, its environment, the object of the test, and the method of assessment. He points out that the analytical bal- ances available in most laboratories provide adequate accuracy for many corrosion tests, and that this general availability of suitable apparatus is no doubt largely responsible for the extensive use of gravimetric methods for the quantitative assessment of corrosion. He mentions that 50 determinations of loss of weight can be used over a much wider range of conditions than gain in weight, although the accuracy may be lower when cleaning of the specimen after exposure is necessary. -The—ASTM (1952) states that after the corroded specimens have been cleaned, they should be reweighed with the same accuracy as the original weighing of'i 0.0005 g. _Fontana (1960) points out that in many cases, visual observation of the specimens on removal from the test solution provides valuable infor- mation concerning the causes or mechanism of the corrosion involved. Champion (l952)_points out that in macroscopic examination it is often useful to examine the corroded metal in two stages: first with the corrosion products still adhering to the metal, and again after re- moval of the corrosion products. Darrin (1946), from experience with ferrous and non-ferrous metals has devised a system in which a score is allocated for the uncleaned specimen on the following basis: Discoloration--none (3), slight (2), moderate (1), bad (0). Roughening--none (4), slight (3), moderate (2), bad (0). Local Corrosion--none (9), slight (6), moderate (3), bad (0). Depth of Pits-~0.001 in. (12), 0.001 in to 0.004 in. (9), 0.005 in. to 0.014 in. (6), 0.015 in. to 0.030 in. (0). General corrosion-~none (12), slight (9), moderate (6), bad (3), very bad (0). Maximum total score--40. The condition of the corrosive liquid is similarly scored: Cloudiness--none (4), slight (3), moderate (2), bad (0). Precipitate--none (8), slight (5), moderate (2), bad (0). General appearance--good (8), fair (6), poor (4), bad (2), very bad (0). Maximum total score--20. In addition to the individual assessment of the particular features of corrosion, the over-all condition of a given specimen is expressed as a percentage of the possible maximum (40), while the over-all condition of the system is expressed as a percentage of the possible maximum for the sum of the liquid and the specimens-in the system. The percentages are interpreted as shown in Table 8. After the specimen is cleaned the rel- ative order of severity of the different types of corrosion, the maximum pit depth, and the number of pits per unit area are noted. Apart from the depth of pitting, no guide is given as to the meaning of the terms of severity of the corrosion involved in the system. 51 Table 8. Interpretation of scores according to Darrin (1946). o ScoreI Z Designation Degree of Corrosion 100 Perfect No indication 95-99. Excellent Minor, but very satisfactory. 85-94 Good Definite, but probably satisfactory 75-84 Fair Questionable 65-74 Poor Probably unsatisfactory 64 Bad Severe corrosion Two laboratory procedures which have been employed or recommended for investigations of this;nature follow: The ASTM (1958) has published a standard method of total immersion corrosion testing for soak tank metal cleaners. Described is the scope of the test, apparatus, test specimens, precleaning test specimens (de- greasing only), test conditions, procedure for quantitative weight loss test (with corrosion products removal methods for different metals listed in appendixes), procedure for qualitative surface corrosion test, pro- cedure for residual-cleaner corrosion test, and reporting the results. A brief description of the various aspects of the test method fol- lows. The method of testing is intended as a means of determining the corrosive effects of soak tank metal cleaners on all metals other than aluminum and its alloys, under conditions of total immersion, by quant- itative measurement of weight change and/or by qualitative visual deter- mination of change. The test is stated as being designed for the deter- mination of the effects of the cleaner on metals being cleaned, and is not for determination of the life of the cleaner or of the containing equipment. The apparatus consists of test tubes, Allihn-type condensers, with stoppers or joints and a constant temperature device. The representative test specimens should have an area between 0.300 to 0.375 dmz. At least two, and preferably four, replicates are recommended to be tested in each concentration of cleaner solution. The precleaning procedure recommended is as follows: (1) Immerse the test specimens in a beaker of carbon tetra- chlorice or trichloroethylene and swab the surfaces of the individual specimens thoroughly using clean forceps to hold both the cotton swab 52 and test specimen.. (2) Shake off excess solvent and transfer to the vapor degreasing bath long enough to observe the vapor completely cover- ing and condensing on the specimen. (3) Swab the specimen separately in a beaker of alcohol at 50°C (122°F) and shake free from excess alcohol. w (4) Transfer the specimens to a beaker of distilled water at 50°C (lZZPF), swab and shake free of excess water. (5) Immerse the specimens separate-. 1y several times in the beaker of acetone, and shake free of acetone. Dry in a vacuum desiccator or in a low temperature oven at 37°C (99°F) and weigh. . The recommended test conditions are as follows: The ratio of metal area to solution volume should not be less than 1.5 dmzll. The solution concentrations to be tested when the manufacturer's recommendations are available are the following relative concentrations; one-half the concen- tration recommended, the concentration recommended, and twice the con- centration recommended. It was recommended further that a blank test of either two or four replicates in freshly boiled distilled water be in- cluded. Freshly boiled distilled water was recommended for making the other water-soluble cleaner solutions. With the available temperature recommendations it was recommended that the tests be conducted at the average recommended temperature and at 11 i_l°C (20 i 20F) above and be- low this temperature. The remaining procedure for quantitative weight loss is that after having obtained the initial specimen weight to the nearest 0.001 g the specimen is transferred to the test tubes containing the preheated clean- er solutions. After exposure for a 2 hour period or other appropriate time the specimens are removed from.the solution and treated as follows: (1) Hold the specimen in forceps and rinse thoroughly in a l-liter beaker into which tap water is flowing rapidly. (2) Rinse thoroughly in dis- tilled water at room temperature. (3) Rinse with a stream of acetone from a wash bottle. Shake free from acetone and dry. (4) Examine spec- imen appearance before and after removal of corrosion products with re- gard to the following: Discoloration, Dulling, Etching, Presence of accretions and relative amounts and areas, Type of pitting--wide, medium , or narrow, and Presence of selective or localized attack. (5) Corrosion products may be removed from iron and steel by chemical treatment in a 53 boiling solution of ammonium citrate (10 percent by weight) with a 2 or 3 minute immersion time. Shaw and McCallion (1959) of Chas. Pfizer and Company, Inc., manu- facturer of a sodium gluconate compound which is used in combination with caustic systems as a sequestering agent for water hardness encount-- ered in bottle washing, conducted a study of the corrosiveness of gluconate- caustic systems. Their paper and additional correspondence with McCallion (1959) indicate the laboratory procedure used. The laboratory test method involved the use of a standard roller, bushing and pin obtained from a well-known chain manufacturer. The bushing, pingand roller were thoroughly cleaned in detergent solution and rinsed. Any residual scale or rust was removed by a short dip in 5 percentammonium citrate solution at about 65°C (150°F), followed by rinsing, drying and weighing. Speci- mens were assembled and placed in 500 ml polyethylene bottles, (the metal surface area was 195 cm2 in a test solution volume of 200 ml, the roller contributing 125 cmal The pin was externally connected to a motor and rotated at 25 RPM, which is the equivalent of a roller linear speed of approximately 16 ft/min. The roller remained stationary. Test solutions of 3 percent sodium hydroxide plus additives were prepared using water of 12 grains per gallon hardness. The addition of gluconate to the solutions varied from 0.045 to 0.15 percent based on the total solution weight. Various other additives were also used. The test sol- utions were added to individual bottles which were then placed in a con- stant temperature bath at 60°C (140°F). 'The length of exposure was for 7 days. IMost of the testsawere for a single 7-day exposure. A few tests were for two 7-day exposures. The weight loss of only the bushing and pin was determined, because as the‘authors pointed out the roller was so heavy as to make its weight loss inaccurate for data conversion. The specimens were then treated with a dilute acid dip (2 percent hydrochl- oric acid plus Rodine inhibitor) for 2 to 5 minutes to insure removal of any residual scale. The specimens were then rinsed, dried, and weigh- ed again. It was stated that, "since the differences in weight before and after the dilute acid treatment were negligible, the weight losses prior to the dilute acid treatment were used in calculating corrosion rates. The acid treatment was however, continued throughout the test." 54 In addition to measured weight losses, iron analysis of each test solution was performed. They stated that this provided a check of corrosion fig- ures. They said that the corrosion figures based on the two different data agreed in 13 out of 15 runs, which was said to be well within the range of experimental error. It is interesting to note that this close agreement was obtained in the iron analysis where the total amount of. corrosion due to the bushing, pin, and roller was determined, while in. the weight loss measurement the amount of corrosion was determined using only the bushing and pin. The iron analysis was performed colorimetrically using hydroxylamine to reduce all iron to the ferrous state and adding phenathroline to form the colored ferrous-phenanthroline complex. (Before analysis all insoluble iron in any of the test solutions was dissolved with hydrochloric acid so that the analysis gave total iron removed from the metal.) This would not give the total corrosion since the corrosion products on (or adhering to) the metal were not included in this iron analysis as des- cribed above. An appropriate statement at this point seems to be one made by Fontana (1960) at the end of his paper on corrosion testing. He stated that, “in conclusion, I should like to state that hundreds of different types of corrosion tests are made. Apparently the type and detail are limited only by the ingenuity of the personnel involved. Types vary from the 'quick and dirty' to exotic arrangements. The important point is that the test should produce data suitable for the intended use or application." ' ;§§§DLTS AND DISCQSSLON OF IESTS PEBTAININQ TQ THE DEVELOPMENT OF THE EXPERIMENTAL PROCEQQRE Preliminary tests were conducted to become familiar with the tech— niques and essential details commonly employed in total immersion corro- sion studies. Tests were conducted to obtain acceptable processes for pickling and removal of corrosion products from corrosion test specimens. Certain processes either recommended for use or used in previous studies reported in the literature were evaluated for use in this study. Since each test is an entity in itself, it will be treated as such as far as possible. . A preliminary test was conducted as follows in order to become . 9H 55 acquainted with and to facilitate an appreciation for the various tech- niques and details required in corrosion.evaluations: Pint Mason jars with glass covers were used, each jar containing about 450 ml of cleaning solution under study and one hot-rolled mild steel specimen (0.25cm x 4.0 cm x 8.0 cm). Prior to use, the specimens were washed in a household deter- gent solution (Tide), rinsed in water, wiped dry, dried in a vacuum oven and weighed to the nearest 0.0001 g. The specimens were placed in the jars and laid at about a 45 degree angle to the base of the jar. At the end of the exposure period, the specimens were removed from the jars and again washed, rinsed, dried, and weighed as before. The data obtained for duplicate specimens and repeated immersions were obviously neither reliable or reproducible. The apparent reasons for this are probably one or more of the following: (1) The washing pro- cedure may have been inadequate. (2) Presence of and variations in the amount of scale already on the surface could have interfered with corro- sion and the obtaining of reproducible results. (3) Since the strip laid at an angle, corrosion products might have accumulated more on one sur- face than on the other, thus influencing corrosion.i (4) The washing pro- cedure employed after exposure did not facilitate the removal of corro- sion products, leaving them to interfere with subsequent corrosion. (5) Glass containers were used with alkaline cleaning solutions, and changes in solution corrosiveness could occur, due to exhaustion of cor- rosive constituents, or the accumulation of constituents that might affect further corrosion. Several jars broke during the test. A second test was conducted as follows: Polyethylene containers and caps with nylon string attached to the cap for support of one hot-rolled mild steel specimen (0.25cm x 4.0 cm x 8.0 cm) in 500 ml of cleaning sol- ution were used. Prior to use the surface of the specimens were milled down to base metal, degreased according to a standard procedure given in .ASTM (1958), rinsed, dried in a vacuum desiccator and weighed to the nearest 0.0001 g. The specimens were suspended from the caps in con- tainers of solution for exposure. During exposure of the specimens a test using control specimens was conducted to evaluate the acceptability of a corrosion products removal method suggested in the Appendix of an .ASTM (1958) standard. The method was considered unsatisfactory for use, 56 since considerable base metal was removed from the specimens.. (The re- sults of this test and others to develop a satisfactory corrosion prod- ucts removal procedure without appreciable removal of base metal are presented later in this section.) As a result, this test had to be term- inated with an unsatisfactory method for removal of corrosion products and the results are therefore not presented. However, the results ob- tained did appear to be more reliable and reproducible than those of the previous test. Two important problems remain to be dealt with. First, the removal of surface scale and oxides which vary in amount, so as to obtain a uni- form, reproducible surface for the determination of corrosion. A chemical treatment would be advantageous compared with the rather time consuming and expensive milling process to remove surface scale and oxides used in the above test. Second, an acceptable corrosion products removal process (one that effectively and efficiently removes corrosion products with a minimum removal or attack of base metal) is needed. Tests to obtain acceptable processes for the removal of surface scale before exposure and the removal of corrosion products after expos- ure follow: It was pointed out in the literature review that Shaw and MeCallion (1959) used a short dip in an ammonium citrate solution (5 percent by weight) at 65°C (150°F) to remove scale and rust before exposure of their specimens. This pickling process was evaluated for use in this study as follows: Prior to pickling, hot and cold rolled mild steel specimens were first measured to facilitate calculation of the surface area, then degreased according to the procedure given in ASTM (1958) standard, dried in a vacuum oven and weighed. After pickling the speci- mens for 2 minutes they were rinsed, dried and weighed. The mean weight loss of duplicate hot and cold rolled mild steel specimens was 21.7 mg/dm and 4.2 mg/dmz, respectively. A visual examination of the specimens in- dicated that the cold rolled specimens were adequately treated, however, there appeared to be some dulling of the metal finish. This treatment *waa inadequate for hot rolled specimens, since there was little removal 2 of the heavy scale deposits present initially. Acid solutions are commonly used for pickling specimens as pointed 57 out in the‘literature review. The use of inhibitors has also been recom- mended to retard the action of the acid on the base metal. Rodine inhib- itors are recommended for this purpose. Rodine III inhibitor was obtained from a local chemical supplier and used with hydrochloric acid as recommended for pickling mild steel specimens as follows: The specimens used in the previous test were im- mersed for 2 minutes in a 65°C (150°F) solution of 2 percent hydrochloric acid (by weight) plus 0.02 percent Rodine III (by weight), rinsed, dried, weighed and visually examined and then this same process was repeated again. The mean weight loss of duplicate cold rolled mild steel speci- mens was 5.6 and 6.9 mg/dm2 after the first and second treatments, res- pectively. The mean weight loss of duplicate hot rolled mild steel speci- mens was 178.2 and 166.4 mg/dm2 after the first and second treatments, respectively. The scale deposits on the hot rolled specimens had loosen- ed up and some rubbed off on the forceps used in handling them after the first treatment. Most of the scale, but not all, was removed after the second treatment. These results indicated that an inhibited acid pick- ling solution of higher concentration, and/or higher temperature could be satisfactory. Another test was conducted using the same type inhibited acid pickling solution, at boiling temperature with a 2 mmnute immersion on new degreased hot rolled specimens. The mean weight loss of the dup- licate specimens was 428.3 mg/dmz. The specimens were microscopically examined after 1 minute and 2 minutes in the solution. After 1 minute the surfaces were free of scale and no further change in the surface was noted after the second'minute of immersion. The results of this and the previous test showed that this solution (2 percent hydrochloric acid (by weight) plus 0.02 percent Rodine III (by weight)) would be satisfact- ory for pickling hot and cold rolled mild steel specimens. However, correspondence with Linden (1960) of the Amchem Products, Inc., producers of the Rodine inhibitors, resulted in the recommendation that Rodine 213 be used in preference to Rodine III because Rodine 213 is a more effi- cient, stronger, heat stable product. The following test was conducted using Rodine 213 as the inhibitor. Duplicate hot and cold rolled mild steel specimens were measured, de- greased, dried and weighed. The specimens were then immersed and brushed 58 for 2 minutes in a 93°C.(200°F) solution of 15 percent hydrochloric acid (by weight) plus 0.4 percent Rodine 213 (by volume), rinsed, dried and weighed. The mean weight loss of the duplicate hot and cold-rolled. specimens was 354.7 mg/dm2 and 3.3 mg/dmz, respectively. Examination of the specimens after pickling showed that the scale had been removed-. Successive immersions of duplicate hot rolled mild steel specimens after this first pickling treatment showed the following mean weight losses: 1.7, l.5,1.6, and 1.4 mg/dm2 when pickled exactly as described above a second, third, fourth and fifth time. These results further indicated the complete removal of scale in the first treatment, and also that once the scale has been removed that very little attack on the base metal occurs. The weight loss due to this pickling process was also evaluated when hot and cold rolled mild steel specimens were prepared for a study to determine if the amount of corrosion in several cleaning solutions was significantly different for these two types of specimens. The re- sults of this test are reported later. The mean weight loss and stand- ard deviation for 16 hot and 16 cold rolled AISI No. C 1008 steel speci-1 mens due to pickling was 385.8 mg/dmz and 21.6 mgth and 3.6 mg/dmz and 1.4 mg/de, respectively. Since this pickling process was both effect- ive and efficient, and did not result in appreciable attack of the base metal it was accepted for use in these corrosion studies. With an acceptable pickling process specimens can be prepared and immersed in the respective solutions for corrosion rate evaluations. After immersion the corrosion products must be completely removed, with a minimum removal of base metal before assessing the amount of corrosion by weight loss. It is important to point out that the weight losses are not expected to be large, and therefore, drastic methods (those which remove considerable base metal) can not be used. The effect of several processes used for corrosion products removal, on base metal of iron and steel specimens is reported in the literature. 'They are pre- sented below to provide an indication of what might be expected. Champ- ion (1952) reports that immersion of steel in 20 percent sulphuric acid solution containing 0.05 percent di-orthotolylthiourea as inhibitor for 1 hour at 60°C (140°F) resulted in 13 g/m? (130*mg/dm2) metal loss. He also reports that immersion of blank specimens in 20 percent ammonium 59 citrate solution for about 20 minutes at 75 to 80°C (167 to 176°F) re- sulted in a weight loss of less than 3 g/m2 (30 mg/dmz). Knapp (1948) shows weight losses ranging from 0.0 to 6.0 mg/dm2 for a number of clean metals subjected to electrolytic cleaning in inhibited 5 percent sul- furic acid at 75°C (167°F) for 3' minutes. It is essential as pointed out by Champion (1952), ASTM (1952) and Knapp (1948) that regardless of the method of removing corrosion products, the rate of base metal removal must be determined for the particular materials under test, in order to avoid major errors. The ASTM (1958) and Speller (1935) have suggested that the corro- sion products on iron and steel be removed by imersing the specimens. for 2 minutes in a boiling solution of ammonium citrate (10 percent by weight). This process was evaluated as follows: Duplicate hot rolled mild steel specimens which had been milled to remove the scale and ex- pose the base metal, were measured, degreased, dried, and weighed. They. were then given the above corrosion product removal treatment, rinsed, dried, and weighed. The mean weight loss of the duplicate specimens after each of 4 consecutive treatments was 9.5, 6.1, 6.1 and 5.0 mg/dmz. Another test was conducted exactly as described above using 5 percent amonium citrate at 65°C (150°F) and boiling to determine if a decrease in concentration at the same temperature or if a decrease in both con- centration and temperature would appreciably decrease the amount of ‘ base metal removed. The mean weight .loss of duplicate specimens after each of 4 consecutive treatments was 5.6, 4.7, 4.9 and 5.1 mg/dm:Z with the boiling solution, and 4.2, 4.3, 4.6 and 3.5 mg/dm2 with the 65°C (150°r) solution. Since considerable base metal was still being removed, a change was made to the use of an inhibited acid solution for corrosion products removal. Evans (1960) has pointed out that inhibited acid is generally recomended for corrosion products removal and that preliminary experi- ments are necessary to find the solution and lowest concentration which is effective without seriousl'y’attacking the base metal. Knapp (1948) suggests some inhibitors for use, one of these was Rodine. To evaluate this process, duplicate milled (to remove scale and expose base metal) specimens were measured, degreased, dried and weighed. They were then 60 _immersed for 2 minutes in 2 percent hydrochloric acid (by weight) plus 0.02 percent Rodine III (by weight) solution. The mean weight loss of the duplicate specimens after each of 4 consecutive treatments was 10.5, 10.9, 11.8, and 17.1 mg/dm2 in boiling solution, and 3.1, 2.7, 4.2, and 2.3 mg/dm2 in 65°C (150°F) solution. The over-all mean weight loss of the 4 consecutive treatments in the 65°C (150°F) solution was about 25 percent less than the 5 percent ammonium citrate solution at 650 (1509F). An additional test was conducted exactly as described above except that 004 percent RodineIII was used in a solution at 65°C (150°F) to deter- mine if the amount of metal removed could be reduced still further. The mean weight loss of duplicate specimens after each of three consecutive 2-minute immersions in 65°C'(1509F) solution of 2 percent hydrochloric acid plus 0.04 percent Rodine III was: 2.2, 2.7, and 1.4 mg/dmz. The over-all mean of these 3 immersions'weight losses.is about 32 percent below the over-all mean of the 4 immersioni‘weight losses in the previous inhibited acid solution at 650 (150°F) and about 50 percent below the mean of the 4 immersions’weight losses in the 5 percent ammonium citrate at 65°c (150°p). . As mentioned earlier in the section on obtaining an acceptable pickling process, Linden (1960) suggested the use of Rodine 213 in pre- ference to Rodine III because it is a stronger, more efficient, and heat stable inhibitor. Their literature showed that the maximum amount of base metal removed from mild steel immersed in a 5 percent (by weight) hydrochloric acid plus 0.1 percent (by volume) Rodine 213 solution at 82°C (180°F) would be about 0.17 mg/dmzémin. A solution of 5 percent hydrochloric acid (by weight) plus 0.1 percent Rodine 213 (by volume) *was evaluated for its effect on base metal as follows: Hot and cold rolled AISI No. C 1008 steel specimens were measured, degreased, pickled (using the pickling process adopted above), rinsed, dried, and weighed. They were then immersed and brushed, in the above solution at 82°C (180°F) for 2 minutes. The mean weight loss of duplicate specimens after each of 6 consecutive treatments was as follows: 1.7, 1.7, 1.9, 1.8, 1.8, and 1.6 mg/dm2 with cold rolled steel and 2.0, 2.1, 1.7, 1.7, 2.0 and 1.6 mg/dm2 with hot rolled steel. . A comparison of the mean and standard deviation of each of the bet- ter processes of each type evaluated above follows: The 5 percent . .- -. 61 ammonium citrate solution at a temperature of 65°C (lSOoF) and a 2 minute immersion of the specimens resulted in an over-all mean base metal re- moval of 4.1 mg/dmzand a standard deviation of 0.50 mg/dmZ. The 2 per- cent hydrochloric acid plus 004 percent Rodine III solution at 65°C (150°F) and with a 2 minute specimen immersion had an over-all mean base metal removal of 2.1 mg/dm2 and a standard deviation of 0.58 mg/dmz. The 5 per- cent hydrochloric acid plus 0.1 percent (by volume) Rodine 213 soTution at 82°C (180°F), with a 2-minute immersion of both cold rolled and hot rolled AISI No. C 1008 steel specimens resulted in an over-all mean base metal removal of 1.7 mg/dmz and standard deviation Of 2.4 mg/dm2 for the cold rolled specimens and a mean base metal removal of 1.8 mg/dm and a standard deviation of 0.26 mg/dm2 for the hot rolled specimens. Since the latter process resulted in a minimum removal of base metal 2 (small mean) and a small deviation in weight loss for consecutive treat- ments (small standard deviation) it was adopted for use in this corrosion study. The results of a direct comparison of the corrosion products re- moval method suggested by the ASTM (1958) and the one to be used in this study follows: Eight cold rolled AISI No. C 1008 steel specimens were Ineasured, degreased, pickled, rinsed, dried and weighed. Four specimens were immersed for 2 minutes in a boiling solution of 10 percent ammonium citrate, and the other 4 specimens were immersed for 2 minutes in 82°C (180°F) solution of 5 percent hydrochloric acid plus 0.1 percent Rodine 213, then rinsed, dried and weighed. This procedure was repeated three times with the inhibited acid solution and six times with the ammonium citrate solution. The mean weight loss after each treatment was 2.0, 1.9, and 1.9 mg/dm2 with the inhibited acid solution, and 8.4, 5.8, 3.5, 442, 4.0 and 3.5 mg/dm? with the ammonium citrate solution. The com- ‘bined over-all mean weight loss and standard deviation (based on the individual weight loss values for each specimen) of the inhibited acid and ammonium citrate were 1.9 mg/dm2 and 0.14 mg/dm2 and 4.9 mg/dm2 and 2.1 mg/dm2 respectively. The data show that the inhibited acid removes about 61 percent less base metal with about a 93 percent less deviation from specimen to specimen than the ammonium citrate solution. CORROSION STUDIES THE EXPERIMENTAL PROCEDURE Laboratory total immersion corrosion tests were conducted to assess the corrosiveness of alkaline sequestering agents to mild steel speci- mens. The amount and intensity of corrosion was determined by weight loss measurement and visual examination. The experimental procedures used are grouped according to: Metal, Corrosion media, Laboratory expos- ure conditions, or Assessment of effects on the metal and media. Metal factors Commercial quality AISI No. C 1008 cold rolled and hot rolled steel plate corrosivity was determined in several alkaline solutions represent- ative of those to be studied in this investigation. No significant difference in corrosivity was found for cold rolled and hot rolled steel specimens when immersed in distilled water solutions of 3% sodium hydrox- ide, 3% sodium hydroxide plus 0.168% sodium gluconate and 3% sodium hy- droxide plus 0.24% ethylenediaminetetraacetic acid; therefore, cold rolled steel specimens were used throughout the remainder of the invest- igation. Test specimens of about 1 mm in thickness, 3 cm in width and 10 cm in length were furnished by the U. 8. Steel Corporation with the detailed chemical analysis given in Table 9. The microstruétures of both materials were normal for C 1008 steel. Table 9. Composition of hot rolled and cold rolled AISI No. C 1008 steel specimens. Materials ARL No} Composition, weight percent C Mn P S Si Cu Ni Cr Hot rolled L 1746 0.083 0.37 0.013 0.028 0.012 .0.037 0.003 0.018 steel Cold rolled L 1747 0.057 0.34 0.007 0.024 0.006 0.001 0.003 0.01 steel EARL = Applied Research Laboratories Of U. S. Steel Corporation Preparation 0f the specimens involved drilling a hole 1 mm in diam- eter at the end of each specimen (to be used in supporting the specimen), measurement of each specimen's dimensions with vernier calipers to 0.1 mm (the specimens generally measured about 1.3 mm in thickness, 4.15 cm 62 63 cm in width, and 10.17 cm in length giving a surface area of 88.13 cm2 and having an initial weight of about 44.4523 g), and degreasing and pick- ling of the specimens. The specimens were degreased according to the following procedure recommended in the ASTM (1958) standard method of total immersion corrosion testing for soak tank metal cleaners: (1) Immer- se the test specimens in a beaker of carbon tetrachloride or trichloro- ethylene and swab the surfaces of the individual specimens thoroughly using clean forceps to hold both the cotton swab and test specimen. (2) Shake off excess solvent and transfer to the carbon tetrachloride or trichloroethylene vapor degreasing bath long enough to observe the vapor completely covering and condensing on the specimen. (3) Swab the specimen separately in a beaker of alcohol at 50°C (122°F), and shake free from excess alcohol. (4) Transfer the specimens to a beaker of distilled water at 50°C (122°F), swab and shake free of excess water. (5) Immerse the specimens separately several times in the beaker of acetone, and shake free of acetone. After degreasing,the specimens were pickled to remove surface scale and provide a uniform, reproducible surface. The specimens were immersed and brushed with a nylon bristle brush for 2 minutes in a 93°C (200°F) solution of 15 (weight) percent hydrochloric acid plus 0.4 (volume) percent Rodine 213, rinsed and weigh- ed. Two specimens were pickled in each beaker containing 500 m1 of pick- ling solution. The specimens were rinsed for one-half minute in each of two 1-1. beakers of distilled water, followed by an acetone rinse from a ‘wash bottle. They were then wrapped in a dry, smooth, hard-surface paper and dried for an hour in a vacuum desiccator. After immersion in corrosiOn media, the specimens were removed and rinsed as described above. Next, it was necessary to remove the corrosion products before *weighing and subsequent re-immersion of the specimens. (Four replicates 'were used for 5 consecutive immersions with the weight loss determined after each of the five 7-day immersions in order to evaluate each treat- tnent effect.) The corrosion products were removed by a 2-minute immer- sion of the specimens in 5 (weight) percent hydrochloric acid plus 0.1 (volume) percent Rodine 213 solutions at 82°C (180°F). Two specimens ivere processed in each 500 m1 beaker of corrosion product removal sol- ution. The specimens were then rinsed, dried and weighed as described above. 64 Corrosion media Boiled distilled water and water with 12 grains per gallon total hardness as calcium carbonate were used to prepare corrosion media. The hard water was prepared by diluting tap water of 19 grains per gallon total hardness with distilled water. The hard water of 12 grains per gallon total hardness consisted of 8 grains per gallon as calcium hard- ness and 4 grains per gallon as magnesium hardness. The iron and chlorine content were each less than 1 ppm. Water, sodium hydroxide, sodium gluconate, sodium versenate (tetra- sodium ethylenediaminetetraacetate), trisodium phosphate, and tetrasodium pyrophosphate were used as agents in this corrosion investigation. The manner in which these agents were used is described in Figure 7; the agent combinations are: water; 3% sodium hydroxide; 3% sodium hydroxide plus sodium gluconate; 3% sodium hydroxide plus sodium versenate (tetrasodium ethylenediaminetetraacetate); 3% sodium hydroxide plus trisodium phos- phate (TSP); and 3% sodium hydroxide plus tetrasodium pyrophosphate (TSPP). Concentrations of the agents can also be noted on Figure 7. Three con- centrations of sodium gluconate and sodium versenate (tetrasodium ethylene- diaminetetraacetate) were evaluated; these are the calculated theoretical amounts of the chemical necessary to sequester the hardness in 12 grains per gallon, water, two times this amount and four times this amount, which correspond to the 1x, 2x, and 4x values in Figure 7. One concen- tration of trisodium phosphate and tetrasodium pyrophosphate was eval- uated. This concentration was the commercial recommended amount which is approximately two times the theoretical concentration for 12 grain per gallon water. The Specific amounts of each agent added on a percent by weight basis is given with the results (Table 26). The amount of sequestering agent added is based on the amount of water and not on the total solution weight. The entire experiment diagrammed in Figure 7 was carried out for distilled water and for hard water of 12 grains per gallon total hardness. WhenLthe distilled water was used, the agents and concentrations were the same as with hard water. Since the sequestering agents are added to control hardness, the unreacted sequestering agent in the distilled xaater test is at a higher concentration than in the hard water tests. 65 usuaquomxo oouoouuoo .u .wum unoaouaamou Hoowuouoonu x e uooaouwovou -1. , HmoauouooAu x N sumamuoamowhm aouoonouuoa oumnmmonm asaoomwus uooaouusoou Houwuouoonu x a summonuo> aoaoom ouooooodw moo usuucooo aowoom a no A ooqxouohs soaooa Rm 5 euoomc noun: one: ooauom use macaw NH 66 Laboratory exposure conditions The specimens were tested in 500 ml polyethylene bottles- The specimens were supported by a nylon string attached to the plastic bot- tle closure with epoxy resin; one specimen to a bottle. This method of suspending the specimens prevented them from touching the container. The ratio of solution yolume to specimen surface area was 5.68 ml/cm2 (or 500 ml/88cm2). 3 The majority of the tests were carried out at 150 i 10F in an automatic temperature controlled agitated water bath. (The water in the bath was agitated and not the bath or the test solutions.) Tests limit- ed to one agent concentration in distilled water. were carried out at 130°F and 170°F.. All test solutions were preheated to the desired'tem- perature before adding the test specimens to the bottles. All tests were replicated four times. The lines in Figure 7 illust- rate the replicated samples. The tests were repeated for five consecut- ive 7-day imersions to determine corrosion variation with successive innersion. Fresh test solution was added to cleaned bottles after each imersion. ' Assessment of corrosion effects on the metal and media The effects of corrosion on the steel specimens tested were deter- mined after each imersion by weight loss and by subjectively observing the change in the appearance of the specimens. The specimens were weighed to 0.1 mg on a laboratory analytical balance having a sen- sitivity of 0.05 mg at full load. The corrosion results are reported as milligrams weight loss per square decimeter-week (mg/dmz-week). After removal of the test specimens from the test solutions and rinsing, the appearance and general nature of corrosion was assessed by visual exam- ination of the test specimens, both before and after removal of corro- sion products. The specimens and test solutions were examined and scored according to a system developed by Darrin (1946), with the ex- ception of one modification. All categories used for assessing the visual specimen corrosion involved a qualitative measure or type of score with the exception of the’ pitting category. Darrin's procedure requires a quantitative measure of pit depth and the score given was bases on this pit depth. In this investigation pit depth was not 67 measured, but rather pitting was given a qualitative score based on the amount and intensity of pitting. Darrin's system for visual examination and scoring of the specimens and solutions follows: System for visual examination and scoring of the corrosion effects on the test specimens before and after removal of corrosion products: Discoloration--none(3), slight (2), moderate (1), bad (0). Roughening--none (4), slight (3), moderate (2),-bad (0).- Local corrosion--none (9), slight (6), moderate (3) bad (0). Pitting--none (12), slight (9), moderate (6), bad (3), very bad (0). General corrosion--none (12), slight (9), moderate (6), bad (3), very bad (0). “Maximum total score--40. System for visual examination and scoring of the corrosion ef- fects on the test solutions: Cloudiness--none (4), slight (3), moderate (2), bad (0). Precipitate--none (8), slight (5), moderate (2), bad (0). General appearance--good (8), fair (6), poor (4), bad (2), very bad (0). Maximum total score--20. In addition to the assessment of the particular features of corro- sion, the over-all condition of the specimens or solutions given a part- icular treatment are expressed as a percentage of the possible maximum; while the over-all condition of the system is expressed as a percentage of the possible maximum for the sum of the liquid and the specimens in the system. The percentages are interpreted in accordance with the following table: Table 10. Interpretation of scores according to Darrin (1946). Score, 1 Designation Degree of corrosion 100 Perfect . No indication 95-99 Excellent ' Minor, but very satisfactory 85-94 Good Definite, but probably satisfactory 75-84 Fair Questionable 65-74 Poor Probably unsatisfactory 64 Bad Severe corrosion In presenting the above experimental procedures a number of the tests conducted in this investigation have been mentioned. In addition to these, 68. the reproducibility of the above experimental procedure with different individuals was determined. The variation in corrosion with length of immersion for several agents was assessed. The effect of a wetting agent on the corrosiveness of an alkaline sequestering agent solution was determined. The corrosiveness of aqueous sequestering agent sol- utions was also determined. 69 W This investigation is concerned with detergent solution corrosion of equipment used in cleaning reusable beverage containers, and in partic- ular, sequestering agent corrosion of bottle washing machines. The re- sults of this investigation are reported according to the following phases of study: (1) Studies on materials and methods, (2) Studies on distilled water-sequestering agent systems, (3) Studies on distilled water-sodium hydroxide-sequestering agent systems, (4) Studies on hard water-sodium hydroxide-sequestering agent systems, and (5) Distilled water and hard water studies combined and compared. It should be noted that these studies are not necessarily presented in the order in which they were conducted, but rather according to an order which tends to facilitate their discussion. Studies on materials and methods Materials The results of a test to determine if cold rolled and hot rolled AISI No. C 1008 steel used in beverage bottle washing equipment varies in susceptibility to corrosion, as measured by weight loss after each of three immersions of separate replicate sets of specimens in four deter- gent solutions, similar to those to be evaluated in this study, are pre- sented in Tables 11 and 12. The mean and standard deviation for each Table 11. Corrosion weight loss data of steel specimens immersed in distilled water for one week (7 days). Steel, Weight loss, mg/dmz-week AISI Steel, No. C 1008 Immersions over-all l 2 3 means Cold rolled Mean 128.87 144.63 151.96 141.82 Standard deviation 3.70 4.86 5.02 Hot rolled Mean 117.54 143.03 131.87 130.81 Standard deviation 6.28 6.90 6.16 Immersion over-all means ~ 123.20 148.83 141.91 136.31 detergent by each immersion and steel are included in these tables. (The means and standard deviations are for 4 replications in these tables and all subsequent tables.) The variances calculated from the standard 70 mo.n nm.¢ mo.m co.n mdflufi ammwmooo sauna waste He.n, hm.n .Hn.m mn.m s~.o oc.a an.o m~.¢ “use em.m mo.n c~.o Ho.o mo.o co.m oaqn mmtu mate nN.~ mm.o n~.o mm.N swan Nm.~ om.N OQUOE M N H dH¢1M0>O unowd ocowmuoaau . oosaou no: on.m om.n n~.¢ um.n oe.o eo.a ee.o mn.¢ mo.¢ mm.n «a.n c~.o mm.o em.c mn.n ~¢.n an.n «mar w~.c mm.o o~.o s~.n u¢.m ma.n om.~ undue m N. H gaswuo>o ucow< amownuoaau meadow odoo wood 0 .oz HmH< .Hooum ages; cocoa Hasuuo>o dooum cocoa ~Hm1u0>o soauuoaan oouuma>oo pudendum one: odes owuoomwuuou -oeaaaaeoaoeaauu ne~.o + «venous»: asseom um nowumw>oo pudendum - use: ouoooooam Bounce smog.o + seasons»: asaeom an nowuog>oo oueoomum , use: «venoueaexasaeom oucowc you moowuo~0n unannouoo onwaoxao nu oonuoaaw mooauoomm Hooum mo sumo mmoa usages .Auhmo my 3003 ono noweouuoo .md manna 71 deviationfin Table 11 and Table 12 were tested for homogeneity of variance using Cochran's Test (Dixon and Massey,(1957». Homogeneity of variance was found at the 5% level of significance in both cases; thus satisfying the requirement for homogeneity of variances, essential to the use of analysis of variance methods. The results of an analysis of variance of the cold and hot rolled steel specimedb weight loss data obtained after immersions in distilled *water are presented in Table 13. The interaction was significant (5% Table 13. Analysis of variance of corrosion weight loss data of cold and hot rolled AISI No. C 1008 steel specimens immersed in distilled water. Source Degrees of sum of Mean F freedom squares square Total 23 3709.71 ** Immersions 2 2077.55 1038.77 33.22** Steels 1 726.99 726.99 23.25* Interaction 2 342.28 171.14 5.47 Error 18 562.89 31.27 *:Significant, 5% level. Significant, 1% level. probability level of significance) and main effects were highly signif- icant (1% probability level of significance). The interaction is shown graphically in Figure 8. The error mean square was used to calculate estimmtes of the standard errors. The estimate of the standard error for the steels (S!) was l.62.mg/dm2-week andfor the immersions (SI) was 1.98 Ins/dmz-week. The difference between steel over-all means was highly significant as shown by the F value (see Table 13). Cold rolled steel suffered more corrosion than hot rolled steel. The immersion over-all means were tested for significance by Harter's (1960) critical values for Duncan's new multiple range test. (This method of testing for significant differences is used in the remaining analyses.) Significant- 1y more corrosion was found in immersions two and three than in immersion «one. The amount of corrosion was not significantly different for immer- sions two and three. The analysis of variance of the weight loss data presented in 'Table 12 for cold and hot rolled steel specimens immersed in the alka- line detergent solutions is presented in Table 14. The analysis shows I55 I50.- 2 ‘ ht Ioss, mg ldrn -week “5'9 A; G O 1 Mean 2? “5‘- F Cold Rolled SIeeI, AiSi Hot Rolled No. C I008 O Immersion I A Immersion 2 D Immersion 3 Fig. 8. Immersion-steel interaction with distilled water as corrosive liquid. .3 8 6' a I O N g A \ g 5% 0 v? m 2 E 4" O '5 3 C D g 3+- Fig. 9. Immersion-agent interaction with cold steel specimens. Immersion I Immersion 2 Immersion 3 30/0 SOdium hydroxide 3%, sodium hydroxide ‘0.I68°/o sodium un- conate ‘ Agents 3°/osodium hydroxide *0.24°/o ethylenediam- inetetraacetic acid and hot rolled 73 Table 14. Analysis of variance of corrosion weight loss data of cold and hot rolled AISI No. C 1008 steel specimens immersed in alkaline detergent solutions. Source Degrees of Sum of Mean F freedom sguares sguare Total 71 85.16 - Immersions 2 2.26 1.13 2.92 Steels 1 0.485 0.485 1.25** Agents 2 19.46 9.73 25.16; Interactions ' 1x8 2 0.818 0.409 1.06** IxA 4 40.59 10.15 26.24‘ SxA 2 0.042 0.021 0.054 IxSxA 4 0.620 0.155 0.401 Error 54 20.88 0.387 ** Significant, 1% level. that the steel and immersion main effects are not significant, and that the agent main effects and immersion-agent interaction effects are highly significant. The interaction is shown graphically in Figure 9. The grand agent over-all means for 3% sodium hydroxide, 3% sodium hydroxide plus 0.168% sodium gluconate, and 3% sodium.hydroxide plus 0.24% ethylene- diaminetetraacetic acid are significantly different from each other (given in order of increasing corrosivity). Visual examination results scored according to Darrin's (1946) system (modified by giving a qualitative score for pitting rather than a quantita- tive score based on measured pit depths) for the cold and hot rolled steel specimens before and after corrosion products removal are shown in Tables 15 and 16. The visual examination results for the solutions. scored according to Darrin's system are shown in Table 17. Included in these tables are percent figures representing the total scores obtained with the specimens or solutions expressed as a percentage of their total possible score. The over-all condition of the system, sum of the solution and the specimens in the system before removal of corrosion products, as a percentage of the possible maximum is presented in Table 18. The re- sults shown in these tables indicate no appreciable difference in vis- ual scores for the cold and hot rolled steel specimens. Methods A limited test was conducted to detemmine if different people would 74 ooH mm ooH ooH OCH 0% no no ooH ouoon Hsuou aaaaxoa . mo uooouom Houoa Huuuaoo moHuuHm cm mN wH cc UHOOO NH m NH NH NH 0H m 0 NH down souuoo NH NH NH NH NH NH oH m NH max ”0‘ O\Ch mm 0‘ down nouuoo Hmuog ~¢ rfitfl ~¢~¢ ~¢~¢ ~¢~¢ onoo Anson .AHu>oaou uuoovoun donouuoo wouumv «no wound .AHn>oaou nauseoum ooHoouuoo ouomonv «mo ouomum N H UflOG UfiUOOGQHUOU n «no nouu< iuofiauHoodonnuo NQN.O + H mac ououom uvwxouvhs afidvom Rm m «mu uauu< ouddooon.a=Hoom NmoH.c + n «no ouomom oonouvhs auuuom Rm n «no ~33 N “ho vacuum 26.33.3853 85.25% Rm o MAO Hound 0 ”mac ououom noun: ooHHHumHa n snout 82.:qu aoHun ruoH oomHm dong—H.538 uaoauuoua .numm mooHauoaaH nous» mom moaHn> one: .oooaHuonm Haouu mooH 0 .02 Hde voHHou vHoo do dowmouuoo mo mucoumo no» mo :OHudaHadxo onoomouomz .mH oHndH 75 mm as ooH coH OOH mm me no ooH swoon Hduou -asaaxua mo uoouuum Heuoa Huuooou moHuuHm mm 0N mH oe UHOUQ NH 5 NH NH NH 0H m o NH aoHo nouuoo NH NH NH NH NH NH HH NH GO [‘0‘ O\O\ NC” a oOHo nouuou aha mound «no ouomom «no wouu< “no ououom duo wound duo ououom emu nouu< “NU ”HOMQQ swoon aaamex ~¢ vdtn ~d~d ~¢~¢ ~3-d v3 c>c> once «so: warm vHum OHuOUGQuuuu iodHauHooaonnuo N§N.o + ovonuohA asHuom Rm oueoooon asHvoo NmoH.o + season‘s; season an masseuse: season «n can»: nuaaauuaa muwno ooHun Hoooa unmaom -uoHouon oOHundHauxu anonymous .euoo udoHouoaaH dunno now eoaHa> mum! .eagaHoona Houuo mooH 0 .oz HmH< ooHHou no: no ooHaouuoo mo auoomwo any mo noHueoHanxo onoooouoex .oH quoH 76 OOH OOH OOH m OOH onoom Houom . -fihfiwfidfl mozudmvmwm ON ON ON H ON onooo Heuoa m d’x‘l’ O onooo 32:4qu viva 0Hu00¢nnuou unawannvuaoHaAuu NQN.O + ovonnv a aaHvom Rm ounnouaHm annooa MQOH.O + «unease»; asneom en monsoon»; annuom an nanosreonnnuana ooaonnonno Hononoo ouuanHoonm emooHvaoHO aoHuooHaaxm unoauoona once on» .nueo nooHonoaaH dunno now ooaHs> new: .aouanoono Hooum ooHHom no: no ooHHon mo connnuaan nuuun unonuoHoo on» no ooHuHooou on» no ooHuonHaixo OHAoouonumz .NH oHnwa 77 anon oHuoocnnuuu anouonuenuoo nonaaunoonowhnuo N§~.O + Sansone as- mooning Boo B 8 . «3.6.3? 82.8 .2 I ouuoouon lsHvoo NmoH.O + 838%.: oz 383.— 2: 2: ”.3333 588 an Aneuoeuonunu 3:38.. u:5 .3338 coco 2 2 «33.6.3 .538 um ; -cmnuonndo und>om oem um «a noun: ooHHHuan donownmo we ounuun nonunounnoa quHon no: voHHon oHoo .-Ao¢mHv flwuuun HO vonuufi may umgma kuum QOOH U .02 HmH< eonooo oonnaaoo any «0 nonuouononoudn nooonomlqonoom oocnnaoo nooaunona . .uuno oooHunoaaH ounnu now cooHn> one: .uuonoonn nonnonnou mo He>oaun onouoo Bunch» can on nooanoomn on» one nonusHon on» no no. son now onouo anHunon aoanxna ago no omeuooonom e on ououunnxe annexe onu no nonunvnou HHnnnu>O .OH «Hana 78 get the same results using the prescribed experimental procedure. Corro- sion test results for three different individuals following the same experimental procedure are reported in Table 19. The mean weight loss, Table 19. Corrosion weight loss data obtained by three individuals using cold rolled AISI No. C 1008 steel specimens immersed in 3% sodium hydroxide plus 0.084% sodium gluconate for one week (7 days). Immersions » weight loss,mg/dmg-week Immersion _____}ndividuals overéall A .B .0. means I Mean 4.08 4.03 4.48 4.19 Standard deviation 0.61 0.29 0.29 Coefficient of ' variation, % 15.1 7.2 6.5 11 Mean 4.68 3.90 3.95 4.18 Standard deviation 0.59 0.41 0.17 Coefficient of ' variation,% 12.5 10.5 4.4 Individual over-all means 4.38 3.96 4.21 4.18 Individual over-all ' ' ' standard deviation 0.640 0.333 0.356 Individual over-all ' " " coefficient of variation, % 14.62 8.41 8.46 Individ a1 over-all mean ' ' ' error , % £12.65 17.28 $7.32 1with confidence coefficient 0.95 standard deviation, coefficient of variation, and mean error are included in the table. An analysis of variance of the data is presented in Table 20. The P values indicated no significant difference between individuals Table 20. Analysis of variance of corrosion weight loss data obtain- ed by three individuals. Source Degrees of Sum of Mean P __i freedom sguares sguare Total 23 5.23 Immersions 1 0.002 0.002 0.01 Individuals 2 0.691 0.345 1.92* Interaction 2 1.30 0.649 3.60 Error 18 3.24 0.180 * Significant 5% level. 79 or immersions, but a significant interaction. The significant immersion- individual interaction is shown graphically in Figure 10. The data were analyzed further, taking the data for each immersion» separately. Table 21 gives the analyses of variance for each immersion. Again no significant difference between individuals was found. However, it is interesting to note that the sampling error mean square (error or within mean square) was the same for all three of the analyses._ Table 21. Analysis of variance of corrosion weight loss data ob- tained by three individuals. Source Degrees of sum of Mean F freedom sguares sguare Immersion 1 Total 11 2.11 Among 2 0.49 0.245 1.36 Within 9 1.62 0.180 Immersion 2 Total 11 3.12 Among 2 1.50 0.750 4.17 within 9 1.62 0.180 Visual examination data for the specimens and liquids are not includ- ed, since the maximum possible score was reported by each individual throughout the test. Studies on distilled water-seguestering_ggent systems Results from a study conducted to determine the effects of certain sequestering agent, distilled water solutions on the corrosion of mild steel are reported in Table 22. Large differences in mean weight losses were found with the various sequestering agents. The mean weight loss decreased with each successive immersion of specimens in tetrasodium ethylenediaminetetraacetate solutions. Visual examination results for the steel specimens scored before and after corrosion products removal are reported in Table 23 and the visual examination scores of the sol- utions are reported in Table 24. Included in these tables are percent figures representing the total scores obtained with the specimens or solutions expressed as a percentage of their total possible score. The over-all condition of the systems is reported in Table 25. With the sodium.gluconate solutions considerable corrosion was observed. The 80 4.80P q 470*- O Immersion i q 43 lmnnnkm 2 460- 4 2‘ fig 45q- P J \. g 4.40" -l a? (h .9 4.30~ . .E c! 6 420- 4 3 c 8 4'0?" d 2 4.00i- - 3.90” _I A B C Individuals Fig. 10. Immersion-individual interaction obtained with steel specimens immersed in 3% sodium hydroxide plus 0.084% sodium gluconate. 81 N0.0¢ OO.N¢ mN.N¢ em.m¢ mm.m¢ nO.Nn mamas HHmuno>o dowmnaaaH NONmodzv ouonnmonaonmm annoomonuou NmmN.O + mH.n OO.m On.m mO.m OO.m wH.m nouns oaHHHuoHO Hanson. Honmnn.eommuzv manganese asnvonnnu sumo.o + ”No” “*9” ON.” 0”.” MN.” ONoN . Hug“: ”QHHfiUQHn oumuoomonuauaaHaaHvoaoHanus aunvomdnuou NOnH.O + mm.HmH mu.NmH Oo.o¢H mm.OmH mO.HOH mm.noH nouns ooHHHuoHn oumaoonHw annoom HQO0.0 + mw.mm nH.mN mm.¢m On.¢m mO.Nn mH.mm nouns oaHHHumHO seams .m .e m N n HHmwno>o naonnoaaH .oaome . I ll xuozuNao\ua «wmoH gamma: anon. undead .thno NV 3003 ooo nom nnoHuoHoa uaomo maHnouaoovoo qnouns voHHHumHo aH vounoaaH maoaHooam Hoaum OOOH 0 .oz HmH< voHHon vHoo mo muHamon mooH uanoa noumonnoo .NN oHan 82 OOH Oe NH NH m mm mm NH NH m OOH on NH NH m OOH Oe NH NH m mm Hm N NH m ON wN o NH N NO mm O OH a NO mN e HH N OOH Oe NH NH a anode Houou none acne Boanxoa onouo ionnoo -onnou mo uaoonom Hmuoa Honoaou waHuuHm Hmooq mmu noum< mac onowom mm mmo noum< «no unamom \fd' FIN) mmo noum< Mmo onowom who noum¢ mmu unamom onoom aoamez €01?) NN ("Iv-‘0') F‘N AON=OH Nonemzv ouozamoamonmn anHoommnuou flmmN.O + nouns ooHHHumHO oommmzv ouonnmonm anqvoeHnu NNm0.0 + none: voHHHumHO anouoomonuouoanaonvoaoHmane aoHoooonuou NOmH.O + none; voHHHueHO ouaaouaHm asHvoo NQOO.O + nouns ooHHHuoHO Honmnn wdHco nonuo lawman inoHoomHO aoHuocHaoxu uaoaumana .ouov oaoHnnoaBH o>Hw nom mooHo> ado: .maoaHooao Hooum mOOH O .02 HmH< voHHon vHoo no oaoHuaHon uaowm onnoumoovoo inane: uoHHHumHv mo muuowuo o>Hnonnoo on» no nonuoanamxo uumoouonunz .NN oHan 83 OOH OOH Om we OOH onoom Hanan aoanxoa mo ucoonom ON ON ON onoom Hence ouamnmommm Hmnoaou w e onoom abamez oumanHoonm mmoaHoooHO oOHuocHamxm Noneozv ouosnmosnonmm annoomonuau NmmN.O + nouns uoHHHuoHO Henson. AONmNH.¢Ommsz manganese anHvOmHnu Num0.0 + nouns ooHHHumHO ououmononuouoanamnoonlonuo aaHvomonuou NomH.O + nouns ooHHHumHO ouooooon anHoom Nam0.0 + nouns voHHHumHO uaoaudona .mumo maonnoaaH o>Hw now moaHo>.omuz .mcoaHoonm Hooum OOOH 0 .oz HmH< ooHHon mHou mo aonnanH nouwn mooHuaHom man no GOHqucoo man we aoHuaoHadxo owmoomonomz .eN oHan . 84 mnouummeumm hno> non anoaHz ucoHHooxm ooHumoHocH oz uuomnom aononnou ono>om com cOHoonnou ono>om mom nonmonnoo mo oonmmo coHumcmeoO r if AonOH.NOchazO muonmmonmonmm asnuoaansou unm~.o + mm nouns wannnsmna Ao~:~n.eommazv manganese .5333 £35 + can nous: cannnsunn ononcommnuouonHaaHooaonnuo anHoomnnuou NOmH.O + Hoemnv annnoa no venous can waHm: monooo moaHnEou any «0 coHumuonanoucH nonoom omcHnaoo an _ nouns ounnnnmnn ., ouocooaHm annoom NemO.O + Nm nouns vaHHumHO uaoonom uaoauoona .muoo moonmnoaan o>Hw now moaHm> and: .uuuooonn cononnoo onu mo Ho>oaon onomon aoummm can oH mooaHooAm ago one aoHuoHou any «0 Ban man now onooo oHnHomoa soaHxaa onu mo owouaoonon n on ooumonaxo saunas onu mo aoHuHocoo HHoino>O .mN oHnoH 85 corrosion products were generally black or dark brown-to-yellowish in color, existing as a tightly adhering, slightly rough layer on the sur- face of the specimens. Some pitting and considerable general corrosion were visible before removal of the corrosion products. After their remov- al some of the general corrosion effects were still apparent and more pit- ting was visible. The over-all nature of the corrosion products was such that they were effectively removed by the corrosion products removal process after each immersion. Precipitation of corrosion products, solv ution cloudiness, and a generally poor solution appearance was observed. The over-all condition of the system (specimens and solutions) was. assessed and interpreted according to Darrin's (1946) methods as bad, with severe corrosion. Considerable corrosion was also apparent with the tetrasodium ethylenediaminetetraacetate solutions. The corrosion prod- ucts were a dark red-to-brown color, and existed as a thin, slightly . rough layer on the specimen surface before corrosion products removal. The nature of these corrosion products were such that they were particu- larly difficult to remove. They seemed to become increasingly more dif- ficult to remove after each successive immersion. Some of the corrosion products could not be removed and the amount increased with successive immersions. Considerable precipitation of corrosion products was obser- ved, along with considerable solution cloudiness and a poor solution general appearance. The over-all condition of the system was assessed and interpreted according to Darrin's (1946) methods as bad, with severe corrosion. The phosphate solutions showed no visible corrosion effects. Studies on distilled water-sodium hydroxide-sequestering_agent systems Sequestering_ggent effects The results of a test conducted to determine the effects of four water softening agents on the corrosion of mild steel in 3% sodium hy- droxide solutions are presented in Table 26. In comparing the effects of these four agent? one to. the other and to the 37. sodium hydroxide con- trol, the "use concentrations" (that concentration of agent required to soften water of 12 grains per gallon total hardness as CaC03) are used. Three concentrations of sodium gluconate (AI) and tetrasodium ethylene- diaminetetraacetate (A11) and one concentration of trisodium phosphate (A and tetrasodium pyrophosphate (AIV) were tested. The use concen- III) 86 NN.m Om.m Ho.N HO.N mH.m N¢.¢ momma HHnuno>o ooHnnoaaH AONmOH.NONm¢on ouosnmonnonha aoHoomonuau NmmN.O + ¢H.m mN.n mm.N Om.N mm.N mH.m aononomn anHoom Nm Ho~=~3.eommuzv manganese aaHvoeHnu RNn0.0 + OO.m ow.n mN.N ON.N mm.N mm.N ovonnohn aaHoom Rm aumuooonnuauoawadHooaoHheno nonvononuou NNHn.O + nm.m mm.m mn.~ mn.~ mm.m mo.m cannons»; asnuom um ouauoooonuouoonaoHooooHanus aanooeonuau NOmH.O+ me.m mm.m mm.N mm.H nm.m mm.m oonxonomn nonvom Rm ouauoomonuouoaHaoHoonoHanus asnoonnnuou NmNo.O+ mH.m OO.N ww.N mm.H mO.N mO.o oononv»s Bannom Nm auoaooaHm aanoon NmOH.O + Hm.m me.~ mn.m me.m mm.m mm.e ounxonene anneom an ouoaooaHm annoon NQm0.0 + mN.m no.m ON.N mo.N Om.m mo.¢ oonxonohn aonvom Rn . ouooounHw aoHvom NN¢O.O + mH.m O¢.N mN.N mO.m mO.m mo.¢ oononoh: aSHvom Rm en.~ m~.m na.n nn.n on.~ om.~ unnxoncna asneom an a... in II. . IN. JI HHoino>o noonenoaaH uaow< xooaiNavxum .omoH asunus and: undead .eaoHuoHoo nouns ooHHHunHo nuaomnauoo onHmeHo an Anzac NO Hoes ado now vannoaan naoaHooau Human wOOH O .oz HmH< ooHHon vHoo mo.muHonon nuoH uanaz aOHmonnoo .ON oHnaH 87 trations are 0.084% with AI, 0.156% with A11, and the concentration tested with AIII and AIV' The results are presented in a manner that allows not only comparisons of agent effects, but also comparisons of specific agent effects to other agents at more than one concentration, and still further comparisons of the effects of A and A concentration on corrosion. The I 11 specific effects of concentration of A and A11 on corrosion are analyzed, and discussed in detail in the last seition of this dissertation for. direct comparison of these effects in distilled water and hard water. An analysis of variance of the weight loss data (Table 26) for the steel specimens immersed in the alkaline detergent solutions is given in Table 27. The analysis shows that the agent, immersion, and interaction Table 27. Analysis of variance of corrosion weight loss data for steel specimens immersed in alkaline detergent, distilled water solutions. Source Degrees of Sum of Mean F freedom squares square Total 179 216.93 ' ** Immersions 4 78.55 19.64 100.72** Agents 8 10.79 1.35 6.92** Interaction 32 101.23 3.16 16.20 Error 135 26.36 0.195 ** Significant, 1% level. effects are each highly significant. The agent and immersion over-all means are ranked from smallest to largest values and tested for signifi- cance by Harter's (1960) critical values for Duncan's new multiple range test in Figures 11 and 12. A horizontal line connects means which are not significantly different. Codes were used to facilitate the present- ation of these results. The immersion-agent interaction is shown graph- ically in Figure 13. The results of a number of statistical tests on the data to assess some of its general characteristics follow. Individual replicate weight loss values ranked by immersions for 3% sodium hydroxide and 3% sodium hydroxide plus 0.156% tetrasodium ethylenediaminetetraacetate did not show a significant concordance (Walker and Lev (1953)). This indicates that the rank of a given replicate from immersion to immersion is independent of its rank in a previous immersion. 88 Annual OH.ouasaaosaonha Hanna-anon”. anNO + 03:93.3 annoom fin an “nouns NH.ouonaeonm gnooanu RNmOO + cannon—:3 Ian—vow Rn no mannuaoanuauananoaaogfie annoonanuou NNHmO + 03:30.3 Saddam Rm neon mauuuaonanuouoanadnooaaHanua 35.30333 .3an + oononPE San—vow Rn “Non muuouounanuononnaanooconfio Bonvoeanuou .336 + 03.8.3.3 Bacon Rm aHon mannaooaHm Bantam .836 + 03383.? Eon—vow Nn neoN “3330an Han—coo NemOO + canon—v.3 gnoom Rm .NON “nuance—Hm aonooo «~36 + «3.8.6.3 .838 an :2 835%.? 53am an .5 332. 3:35. fins 83 23338 snow. wannuuooovoe ocHanHa aH mounds-5 oaaaHoaao Hooua wo aoHeonnoo mo nan-can an» no noenneafioo .HH .wnm w?&6< «on com mom «on. Gm m .on v . o m m. .D U. I I. i w m .I S E IN ”w I m w I oo.n Em in NW. . 86 m3 En on w w «on an men. .2 I 1 M ”I m 89 CD 3 4.42 I CNE 4(- °1a 9’\ Eta 330 E 3.I5 = 3— ? a? 2.6I 2.6I a s: m 632 an. E .9 m I- 3 G 3 4 2 5 I Immersions Fig. 12. Comparison of the amount of corrosion obtained with different imersions of steel specimens in alkaline sequestering agent solutions made with distilled water. Mean weight losses, mg/dmz-week b 01 N L. / r—Ji A ’ -, / '8'» " - v‘ 47 ‘ 4‘ 8» . “ \ I O Immersion I . Immersion 2 A Immersion 3 ‘ Immersion 4 D Immersion 5 2Cl 2C2 2C4 3Cl 3C2 3C4 4 5 AqenIs Fig. 13. Imersion-agent interaction. 91 The corrosion weight loss results obtained with 3% sodium hydroxide and 3% sodium hydroxide plus 0.156% tetrasodium ethylenediaminetetraacetate , solutions from immersion-to-immersion were tested for dependence between immersions. Linear correlation coefficients were calculated and tested for significance at the 5% probability level, for each possible combina- tion of immersions. There were ten combinations for each agent. The correlation coefficients were not significant, indicating that the re- sults from any particular immersion are not dependent upon the results of some other immersion. The same results were found for correlations of. weight loss results of immersion one against immersion two for 3% sodium hydroxide solutions of: 0.084% sodium gluconate, 0.057% trisodium phos- phate, and 0.293% tetrasodium pyrophosphate. The linear correlation coefficients were not significantly different from zero, indicating that the outcome of a subsequent immersion is independent (or not significant- ly affected by) of a previous immersions effects. The data for five of the previously mentioned agents each showed. a highly significant concordance when the immersions were ranked by the replicates. This shows that when a given immersion is either more or less corrosive than another immersion its effect is obtained over all replicates. The experimental design as shown in Figure 6 was set up so that the corrosion weight loss data for the 10 agents (distilled water plus the nine alkaline solutions) could be analyzed together by analysis of variance. However, Cochran's Test for homogeneity of variances (Dixon and Massey (1957)) showed that the variances were not equal (homogeneous) at the 1% significance level. The variance for the distilled water data was much larger than the variances for the alkaline solutions data. Cochran's Test showed homogeneity of variance for the alkaline sol- utions’data. Since homogeneity of variance which is essential to anal- ysis of variance was found for the alkaline solutions, they were anal- yzed together. . To answer the question as to whether the agents as analyzed (five different agents and three concentrations of two of these, to give the total of nine agents) would give the same results if broken down into a five agent analysis and analyses comparing the three concentrations (of 92 each of the two agents) directly, it was found that the answer lies in the fact that a common variance exists for the nine agents (homogeneity was found). The results of such an analysis of variance and tests of significance of the ranked agent over-all means (not presented) for these five agents (the upper and lower concentrations of sodium gluconate and tetrasodium ethylenediaminetetraacetate were not included in this - analysis) showed the same significant differences for these five agents as was found in the nine agent analysis shown in Figure 11. Individual analyses of the two agents at three concentrations showed the same sig- nificant differences for the three concentrations of each agent as was found in the nine agent analysis shown in Figure 11. Visual examination results for the steel specimens scored according to Darrin's system before and after corrosion products removal are . shown in Table 28. The visual examination scores of the solutions are shown in Table 29. Included in these tables are the total scores of the four replicate specimens or solutions as percentages of the total score possible. In addition, Table 30 presents the over-all condition of the system as a percentage of the maximum possible score.) The 3% sodium hydroxide control showed some discoloration, local and general corrosion before corrosion products removal. After corrosion products removal the effects of a small amount of local corrosion was apparent. The A AIII’ and A solutions showed no visible corrosion ) damage to the spicimens or conEZmination of the solutions. The appear- ance of the specimens after immersion in these solutions seemed improved rather than harmed. With AII the specimens showed discoloration (dark red-to-brown color), local corrosion and considerable general corrosion before removal of the corrosion products. Some of the discoloration, local and general corrosion effects were still apparent after removal of the corrosion products. A comparison was made between distilled water solutions of 3% sodium hydroxide, 3% sodium hydroxide plus 0.057% trisodium phosphate (N33PO4‘12320), 3% sodium hydroxide plus 0.293% tetrasodium pyrophosphate (Na4F207'10820) (data in Table 26) and distilled water solutions of 0.057% trisodium phosphate (Na3P04‘12820) and 2.93% tetrasodium pyro- phosphate (Na4P207~10H20) (data in Table 22). The analysis of variance 93 OOH OOH OOH OOH Nm mu Nm mu Nm mm OOH OOH OOH OOH OOH mm mm mm OOH on OO Oe hm On an Om NM Hm Od O¢ Oe Od mm mm mm Oe onoon Hmuou ananxda mo uaoonom Hnuoa Honoaou waHuuHm OHOOO NH NH NH NH HH N NH NH NH NH NH HH NH OH NH down Ionnoo NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH h” 00‘ mm NO NO com mas O‘O‘ N” m aOHu Ionnoo Hoooa d’x‘l’ Qd’d‘ Hw now uoaHa> one: .maoaHoomo Hooum NOOH O .o% HmH< ooHHon vHoo ao maOHuoHom noun: ooHHHunHv nuaawnouoo oaHdeHa mo museums aononnou onu mo aoHuaouaoxo unmooaonouz .mN oHAoH 94 OOH Om OOH OOH OOH OOH OOH OOH OOH OOH onoon Houou asamea mo uaoonom ON OH ON ON ON ON ON ON ON ON onoun Hones m m x? onouu essaxaz Honmon.no~neuzv unannounaonna asneounnuou sna~.c + wenxonene.a5neom u n Honmnn.eomnuzv «uaaauona anneownnu unmc.o + «enacts»; aaneom um aunuoonwndauaaHamHvoauHAnus auneoaunuas uwnn.o.+ .ounxonuna annuom.Nn onenooannuauaauaoHvarathuo asnuouannuu sen3.o + cannons»: asnvom um ounusuaonuuuaauaonuoaaHmenu .asneoaunuuu nmno.o + ounxonuna annuom um ounaooan BSHvom NmOH.O + mongoose: sanuom an ouaaooaHm.a:Hoou NQQ0.0 + cannons»: asnuom an ouaaouaHm aonooo NNdO.O + cannons»: asneom um «anxonvan annuom um oonunaanna Honoaou ounuHaHoonm unannoooHO aoHuooHaaxm neoauoona maonnoaBH o>Hm n0m noaHo> can: .onso .nooawoonm Hanna OOOH O .02 Hde ooHHon vHoo mo oonnoaaH nouma moowuo IHon noun: oaHHHunHu nuaownouoo ooHHaxHa ago no conuHocoo man no aoHuaaHamxa unmounonomz .mN aHan 95 Table 30. Over-all condition of the system expressed as a percent- age of the maximum possible score for the sum of the solution and the“' specimens in the system before removal of the corrosion products. Mean values for five immersions data. Treatment Combined Interpretation of the combined score, scores using the method of percent " Darrin (1946) Designation Degree of corrosion 3% Sodium hydroxide 92 Good Definite, but prOb- ably satisfactory 3% Sodium hydroxide 97 Excellent ‘Minor, but very + 0.042% sodium gluconate satisfactory 3% Sodium hydroxide 100 Perfect No indication + 0.084% sodium gluconate . .,. 3% Sodium hydroxide , 100 Perfect No Indication + 0.168% sodium gluconate 3% Sodium hydroxide ‘ 85 Good Definite, but prob- + 0.078% tetrasodium ably satisfactory ethylenediaminetetraacetate 3% Sodium hydroxide 83 Fair Questionable + 0.156% tetrasodium ethylenediaminetatraacetate 3% Sodium hydroxide 83 Fair Questionable + 0.312% tetrasodium ethylenediaminetetraacetate 3% Sodium hydroxide 97 Excellent Minor, but very + 0.057% trisodium. satisfactory phosphate (Na3PO4.12H20) 3% Sodium hydroxide 100 Perfect No indication + 0.293% tetrasodium pyrophosphate (Na4P207.10H20) results of the corrosion weight loss data for these solutions are given in Table 31. The agent and immersion over-all means are ranked and tested Table 31. Analysis of variance of corrosion weight loss data for steel specimens immersed in distilled water and 3% sodium hydroxide,phosphate sol- utions. Source Degrees of Sum of Mean F freedom squares square Total 99 53.79 ** Immersions 4 15.65 3.91 31.28** Agents 4 3.37 0.832 6.74** Interaction 16 25.42 1.59 12.72 Error 75 9.35 0.125 Significant, 1% level. 96 for significance in Table 32. Table 32. Comparison of agent and immersion over-all mean. weight losses. Agents Agent Immer- Immersion over—all sions over-all mea s, mea s, mg/dm ~week mg/dm -week Distilled water 3.28 5 3.85 + 0.057% trisodium phosphate (Na3P04-12H20) 1 2.95 Distilled water 3.18 3 2.92 + 0.293% tetrasodium pyrophosphate (N4P207'10H20) 4 2.83 3% Sodium hydroxide 3.14 _ 2 2.80 + 0.293% tetrasodium pyrophosphate (N4P207-10H20) 3% Sodium hydroxide 3.00 + 0.057% trisodium phosphate (Na3P04‘12H20) 3% Sodium hydroxide 2.74 The results show that the distilled water plus phosphate solutions used tend to be more corrosive than the 3% sodium hydroxide plus phos- phate solutions, which are in turn more corrosive than the 3% sodium hydroxide control. The distilled water plus 'phosphate solutions were effective inhibitors in light of the concentrations used. These .results.indicate that the corrosion-rate in the rinse sections of bottle washers could be reduced to a value near that found in the alkaline wash sections through incorporation of such agents. Temperature effects The corrosion weight loss results for the study conducted to deter- mine the influence of temperature an alkaline detergent, distilled water solution corrosivity are presented in Table 33. The lSOoF data is taken from Tables 26, 28, 29, and 30. Analysis of variance results for each of the agents tested are reported in Table 34. Temperature, immersion and interaction effects were highly significant in each case. The over- all means are ranked and tested for significant differences in Table 35. 97 Table 33. Corrosion weight loss results of cold rolled AISI No. C 1008 steel specimens immersed in alkaline detergent, distilled water solutions at different temperatures for one week (7 days). Mean weight loss, mgldmz-week Treatments Temperature Immersions over-all _1__2___3_._£I__§_ ““33 3% Sodium hydroxide " ’ ' ' 1300F 2.20 2.40 2.73 3.88 2.78 2.80 lSOoF 2.90 2.50 3.15 1.95 3.23 2.75 170 F 3.33 3.18 3.80 3.13 4.03 3.49 Immersion over-all means 2.81 2.69 3.23 2.98 3.34 3.01 3% Sodium hydroxide + 0.084%osodium gluconate 1300F 4.10 3.90 2.78 4.20 3.83 3.76 1500F 4.03 3.90 2.65 2.78 3.03 3.28 170 F 4.63 4.33 4.78 4.78 6.58 5.02 Immersion over-all means 4.25 4.04 3.40 3.92 4.48 4.02 3% Sodium hydroxide + 0.156% tetrasodium ethylenegiaminetetraacetate 1300F 2.30 1.63 3.90 3.98 3.25 3.01 150°F 5.98 3.53 1.95 2.33 3.38 3.43 170 F 4.25 2.85 3.68 3.03 4.55 3.67 Immersion over-all means 4.18 2.67 3.18 3.11 3.73 3.37 Visual examination results for the in Tables 36, 37, and 38. specimens and solutions are presented 98 Table 34. Analysis of variance of corrosion weight loss data for steel specimens immersed in alkaline detergent, distilled water solutions at different temperatures. Source Degrees Sum of Mean F of- squares square freedom 3% Sodium hydroxide Total 59 31.01 ** Immersions 4 3.58 0.895 3.89** Temperature 2 6.93 3.47 15.10** Interaction 8 10.16 1.27 5.53 Error 45 10.34 0.230 3% Sodium hydroxide + 0.084% sodium gluconate Total 59 68.34 ** Immersions 4 7.86 1.965 7.68** Temperature 2 32.25 16.125 62.99** Interaction 8 16.69 2.086 8.15 Error 45 11.54 0.256 3% Sodium hydroxide + 0.156% tetrasodium ethylenediaminetetraacetate Total 59 79.59 ** Immersions 4 16.51 4.128 18.76** Temperatures 2 4.47 2.235 10.16%; Interaction 8 48.70 6.088 27.67 ‘ Error 45 9.91 0.220 W Significant, 1% level. 99 NO.N ._Hqu mHum com v-th -.o QM - . . O eases ##QMM QMQHO‘ o e 0- 6'3wa“) .. ... MMNNN mHNx'fm Hand'N “MQF‘N sous- . \ma nuance Hsino>o ooHonoaaH naowonuaaH HO . n OMH muduoufldhuuuoq—uawvudflfihAUQ _ne.n can .asnoooonooo nomn.o + no.n onn . .oonxonons sonoom an m~.n own on.n onn. ooooooonm aonoon.uooo.o + no.“ can oonxonono_aonoom an nn.~ own . oo.~ onn oo.n onn oonxooono eonoom an xooaiNio\ma «undue HHsInooo . onsuonanauu. .. nonsuwwunaun. . ouaowc .aonouonooaou uaonomwuo no naofiuoHoa nouns quHHunHo nuaawnouuo oaHHoaHa oH vounuaan naoEHooOu.Hooum mo ooHoonnoo mo assess on» no oonwndnaou .mm oHnoH 100 loss, mg /dm -week I I I I 500i- - N 4.oo~ - 3.00 L- - 0— 03%, sodium hydroxide Weight CI 3% sodium hydroxide +0.084% sodium gluconate [332/0 sodium hydroxide +0.I56°/o tetiasodium ethylenediaminstetraacetate 3I5 325 335 345 355 Absolute temperature, °K Fig. 14. Corrosion weight loss of steel specimens imersed in alkaline detergent, distilled water solutions at different temper- atures for one week (7 days). Temperature over-all mean weight losses plotted. 2.0 I I I I I i 0.700i- ~ NI 3 O E ._ 0600i- - In In 2 E i" osooi- . ‘8 E O 30/0 sodium hydroxide 'E 0 3‘70 sodium hydroxide-0 "O a 0.084% sodium gluconate 3 E 3% sodium hydroxide 00.I56% tetrasodium ethylenediaminetetraacetate am i A l l 1 0.00275 0.00205 0 0&95 0.0030“ 0.003l5 Reciprocal of absolute temperature,’l< Fig. 15. Arrhenius plots of corrosion weight loss of steel specimens imersed in alkaline detergent, distilled water sol- utions at different temperatures for one week (7 days). Temper- ature over-all mean weight losses plotted. 101 Om Om Nm ON mm .Nm Om Om OOH OOH OOH OOH OOH OO Om OO Om Om OOH anoua Houou .eaanxda no unaunoN Heuoa HonoooO OoHuunN mm on nn nn Om NO mm on co co OO Oe Oe mm mm mm mm mm O¢ OHOUQ NH OH HH HH NH HH HH NH NH NH NH NH OH NH OH HH HH NH noun Ionnoo NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH [‘0‘ ”Q Q” O‘C‘ mm [‘0‘ 0‘0\ NQ I‘m m aoae Ionnou Hnuoa QQ N‘rN‘f N'I'Q‘ N‘I'Q Qd’ 0"? NTN‘I‘ d’fl {G o i OaHoo $9.8 -nonooono O ONO naumd N ONO unouom NoONH N ONO nouNO H ONO unouam NoOmH N ONO noumd N ONO anomam N oOnH uunusuannuauoaHanNmoauHNnuu asnoooanuau OOOH. O + oonxouono.asnoom an O ONO nuumd N ONO anowum NoONH n ONO nouu< m ONO onowam NoOnH n ONO nauud m ONO unouam NoOOH ouaoouoHO Ban—non NOOOO + oonxouono sonoom an O ONO nouwd N ONO unouaO NoONH n ONO nuumd N ONO anomoO NoOmH m ONO nuuu< m ONO onouam N OOH , oonxouonn manoom um O 9.30» 62.—«aux ooHuo . nonueananxm «unannouna o>Hm now mosHo> one: 9 ofluflv udOwaHflEw .oaaaHuoNn Hooum OOOH O .02 HOH< OoHHon OHou no monounnuNaou uaonoumHO um moouuoHou nouns ooHHHuoHO nuauOnoqu uaHHeOHo mo uuuowuu nonnonnou onu mo aoHuaaaaaxa oHNouoonunx .OO OHAdH 102 OO OOH OOH nO OOH OOH OOH OOH OOH OOH onouo Hmuou ,asanxaa No uaounoN NH ON ON NH ON ON ON ON ON ON Hmuoa ”0" ”an O QQI" QQQ onouo uuannaoNNo HonoauO unouNNHuonN Qx‘fm d'QN‘I’ mooawvaoHO onoom 6:8.“de noonn NOOnH noonn aunusuaanuouuadauNooaaHNnuo esHoooanuou OOOH.O + «3.8.62 5:8 an NOONH NoOmH NoOmH ouoaooaHO aswoou NQO0.0 + oonxouoso.asnoom so noonn noonn noomn oonxouono sonoom an aoHuoaHame ouaoauoona .uumv maonnaaaH o>Hu now moaHd> one: nannumnumaou uoonoumwv no can: mooauaHon anu mo aoHuHOoou any «a nowudduamxa oHNouuonumz .NO oHaoH .naoaHooNu Hoouo OOOH O .02 HOH< OaHHon OHoo mo oonnoaaH nauma 103 NnOuuowonuem meaaonN non nauNoHuaO ..; - -NnOuuouonuao NHnoaonN nap nouHaHuoO Nnouuomeuon.mnae app .noaHx NnOuonuoHuam mHnnoonN use nauHaHuaO aoHunquau oz oonooonoon oz Nnouuauenunu mHnanonN no: NauHanoO . , ,Nnouuameuueu NHnoaonN nan mouwaawan - NnouunNOHnum Nna> nap nnooqz ooqmonnou mo oonOaO voou OooO uaoHHaoxO vooO uuownoN uuaunaN voou oooO uaoHHooxO nonudaunoon NOOmHv aHnnnO no monuua can wanna ounouo oaanoeou «On No nonunuonNnouoH OO mO mm OO OOH OOH Nm Nm Om noonn NoOmH NoOOH «unusuawnuauoawaanvuaaHNnua .eswooaonuou OOOH.O + oonxouono_a=noom um noonn noonn NoOOH ouaaouaHO anHooa NQO0.0 + oonxonono sonoom an noonn noonn noonn ovonnan_a=NOom NO noounaN «anode OaaHnaoo uaoauaana .ouov maOHnnoaBH a>Hu nom naaHa> one: .nuunoonN aononnou No Ho>oaon anamuo auuomu can an oaoaHouNa can use :oNuaHom can no sun can now onoun aHnHoaON Hosanna can No.0OsuaounoN a on monaoano nonaunnomauu uaonomunv um uaounmu on» no naoHuHanu HHoIna>O .OO anoH 104 Surface-active agent effects Wetting agents are sometimes added to alkaline sequestering agent. solutions to further improve their detergency. One such surface-active agent is Triton QS-15. _ It is an amphoteric surface—active agent, of the oxyethylated sodium salt type, containing both anionic and cationic cen- ters. A test was conducted to determine the corrosive effect of this wet- tirg agent in combination with sodiu- hydroxide and sodium gluconate. The corrosion weight loss results of this test are presented in Tabled39. Table 39. Corrosion weight loss results for cold rolled tusx so. 0 1008 steel specimens iusersed in an alkaline sequestering agent aolotion with and without a wetting agent for one week (7 days). Solutions pre- pared with distilled water. Agents Mean weight loss, qlimz-Iweek ' Agent Immersions over-all 1 2 means 3% Sodium hydroxide 4.08 4.68 4.38 + 0.084% nadir-I gluconate 3% Sodium hydroxide 4.65 4.58 4.61 +3 0.084% sodium gluconate + 0.015% Triton QS-lS Innersion over-all means 4.36 4.63 4.49 Table 40 gives the analysis of variance results for this test which show no significant difference between the two solutions (one with and one Table 40. Analysis of variance of corrosion weight loss data for steel specimens inersed in an alkaline sequestering agent solution with and without a wetting agent. Source Degrees of Sun of Moan F freedoml 232555 a re Total 15 4.29 Immersions 1 0.276 0.276 0.99 Agents 1 0.226 0.226 0.81 Interaction 1 0.456 0.456 1.64 Error 12 3.33 0.278 without the wettim agent) or between the two i-Iersions in corrosivity. Visual examination data for. this study (not presented) showed no evidence log of corrosion with either solution. The maximum possible score for the specimens and solutions was found throughout the study. Length of immersion effect Results of a study conducted to determine the variation in amount of corrosion as affected by the length of immersion are presented in Tables 41, 42, 43, 44, 45, and 46. Analyses of variance results, presented in Table 42, show that only with 3% sodium hydroxide plus 0.156% tetrasodium ethylenediaminetetraacetate in the first immersion, was there a significant difference between the amounts of corrosion found with l, 3, 5, or 7 days of immersion. (It should be noted that separate sets of specimens were employed in each immersion period.) Presented in Table 43 is a comparison of the mean weight losses for each agent and immersion. With 3% sodium hydroxide plus 0.156% tetrasodium ethylenediaminetetraacetate, signif- icantly more corrosion was found with the 7 day immersion than with the 1, 3, and 5 day immersions which were themselves not significantly dif- ferent. The visual examination results indicated a poorer over-all con- dition with the 7 day immersion period for both 3% sodium hydroxide and 3% sodium hydroxide plus 0.156% tetrasodium ethylenediaminetetraacetate and a poorer over-all condition with the 5 day immersion period for 3% sodium hydroxide plus 0 084% sodium gluconate. Studies on hard water-sodium hydroxide-sequestering agent systems Corrosion weight loss results from a study conducted to assess the effects of four water softening agents on the corrosion of mild steel in 3% sodium hydroxide solutions are presented in Table 47.. Hard water of 12 grains per gallon total hardness as calcium carbonate was used in pre- paring the solutions. As in the distilled water-sodium hydroxide- sequestering agent system studies three concentrations of sodium gluconate (AI) and tetrasodium ethylenediaminetetraacetate (A11) and one concen- tration of trisodium phosphate (A ) and tetrasodium pyrophOSphate (AIV) were used. In comparing agent effiits the use concentrations are used. For AI this is the 0.084% level (22C2) and for AII this is the 0.156% level (33C2) and these are compared to the control (3% sodium hydroxide (11)) and to the use concentrations of AIII (44) and AIv (55). (The numbers given in parentheses are codes for the various agents and concen- trations of agents tested, which are used in Figure 16 to facilitate the 106 oN.O oo.m wo.~ actua— HHoIno>o naoO< Om.O Om.O ompu n O0.0 OO.o oo.m n N0.0 N0.0 Om.N O NO.N ON.O OO.N H mmnoinnOHnnoaaH mo summon N nonnnanH NEONOE qnmoH umuaoa one: ON.O HO.¢ O0.0 manna HHoIno>o unoO< Om.m OO.o Om.N N Om.O O0.0 OO.N O NN.N O0.0 OH.O O NO.N OO.o O0.0 H nNoo qnonnnoaan No nummuH H donnnoaaH. ouououoonuouonnaonooaonnuo annooonnuuu OOOH.O + oonxooono eonoom um ouooouoHO eunuch NQO0.0 + oonxouono asnoom an oonxouone sonoom um nuaomd .noOHunHon nouo3.0oHHHunno xunoOnouoo oaHHoOHo on sang «a onanoH Oannoonoan now noonoaan nooanuoNn Hooun OOOH O .02 HOH< OoHHon vHou mo nownonnoo .Ho oHnnH . a>o a and oO H H NH O Hun «was nnm.o mn.n .Nn on.n ommqo monn n ooqm on .1 no.3 on.~n on 4.... m nono 22o «in . n onqon an non.o ~n.o nu oo.~ no.3 on.“ n No.nn an OHS—um O a” .HH...“ Q EOVUQHH N com: «o BOO mo noonOuO N nno.o NA.“ «3 no.o~ on.o o~.o~ n «I no.mm nn nnn.o nm.« «3 mo.n no.3 ~o.n m oo.o. an eon.o mn.~ Nn oo.~ So... :4 m om.n an onoomn condemn savanna N anus. no now, No noonOoO H naoHnnoaaH . annuws aouauom Honda ouooouoonuouaadaonooaonnuo .eSNOononuou OOOH.O + oonxonono.a5nooa um 3:33 nooauom . . Houoa ouooounHO annooo NOOOAO + oonxouono aunoom an onoons nooaoom Hooch oonxouono asnoom Rn ounoom .ndoHuoHon nouns OoHHnonHo .uooOnuuoo onNHoOHo on mean no nOanoH OoHooonuoH nou connoaan naoanuoNo Huoun now noon onOH uOOHos nownonnou mo ounoHno> mo unnNHno< .No oHooH 108 “Inca . o In OONO QONN Q OOHDN s- ~o- s -s N'rnnao cannot Ono GM“ 0 MNNN .finil .nnooH unmask one: N InanI-I mNn—I InfirNu-I «Non «nononuaan no 59.3 NO.N H NN.N O ouauouoonuauoanaonooaoHNOuo O0.0 m asHoooonuou OOOH.O + «on n «3.8.62 .38.... «m OO.¢ N OO.o H 86 m 3333» .538 .586 + mod n 333%? 338 um OO.N m Om.N N OH.O O 86 n 828%.. 538 an NEONOE nNoo 4nononuaan unoH .uAOHoa.aouz. mo :uOouA H naoHnnoaaH nuooO< .naoauaHon nouns ooHHHunHO «udoOnauoo oaNHoOHo on oflwu No I.onuOsoH Oannnunuan now nonnoaan nauanuoNn Hoouo No nonnonnou mo uaooao can No aoanoNaou uOo oHnnH 109 OOH OOH OOH ON OOH OOH OOH OOH OOH Om. OOH OOH OOH OOH OOH OOH OOH NH NH NH NH NH NH NH 33 88 333 33 NH m NH NH NH NH NH NH 3 33 33 33 88 NH onooo Houou .anawxaa onooo Ionnou wo uauunoN Hauoa HonoauO OnwuowN InoaaH osu now nonHo> one: down NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH mm O‘O‘ GO‘ O‘O‘ O‘O‘ O‘C‘ mm 0‘0 m down Ionnoo Houoa N‘I‘Q’ G’N‘f #N‘l’ d’d‘ N'TNT d’d‘ Qd’ #x'f o Onwao Anson O ONO nouwo n «no ouowon once A O ONO nouwd O ONO onowom ammo O O ONO nouwd n «no ououon onoo n O ONO nouwd n «no ouowoo use A ouoaouoHO annooo NOO0.0 + oonxonono asnooa an O ONO nouwd a «no ouowon oaoo N O ONO nouwd O ONO onowoO nNoO O O ONO nouw< O ONO onowoO «Nov O O ONO nuuwd n «no ouonon moo n oonxonono asnoom an O onouo .5303: nowun InoHoumHO aowuoowaoxu nuaaauoona .ouoo oaoHn .noowuaHom noun: OoHHHunHO .uooOnouoO oowHoOHo aw dawn wo oOuOnaH unanowwwo now oonnoaaw ooaawuoNn Hooun OOOH O .oz HOHO OoHHon OHoo wo noHuoawaoxo uwNouoononz .oo oHanH 110 OOH OO OO. ON OOH OO Om. OO OOH OO OO OO Om OO Om OO onooo Houou asawxna onoon NH NH NH NH NH OH NH sown Ionnou NH NH NH NH NH NH NH NH wo uooonoN Houoa HonoaoO OownuwN \00 cm VDO‘ \OO‘ none Ionnou Hound ~¢~¢ ¢fl~¢ ~¢~¢ ~¢~¢ Oowoo Ignace n «no noowe o «no ononon onoo N m Onouoone . N Ono ouowon nnoo m n emu noow< N emu onowon once m n Ono uoow< I N Ono ouowom moo n ounuoooonuouoawaownoaoHNnuo Eflwfiouduuflu meHoc + oonxouone aonoom an oowuo InoHouowO nownonwaoxu nunoauoona .oooonoooo .44 «noon 111 OOH ON OOH ON OOH ON OOH ON OOH ON OOH ON OOH ON OOH ON OOH ON OOH ON OOH ON OOH ON OOH ON onoon Hnuou ,anawxna onoun wo uoounoN Houoa ”””” ”””” ”””” O ouomnnoNNn HonoaoO c o once N o o once m m 4 once n o o Noe n oyououonnuouonwaowoononnuo aonoooouooo Nemn.o 4 oonxouono aonoom an o o oNoo N a o onoo m o o onoo n w o moo n muonouaHO aowooo.NOO0.0 + oonxouoNo aunoom Nn o o oNoo N o o oNoo m o o oNoo m o o Noe n oonxouoNo sonoom Na O o onoom Boawxm: ououHquonN neonwonoHO nonuonwaoxm nuaoaunona .ouno nnownnuaaw osu now nooHn> one: .nooEHoon Hooun OOOH O .02 HOHO ooHHon OHou wo oownnuaEH wo onuOnoH noonowwwo on» now won: nnowuoHon any wo nowuwonoo unu wo ooHuonwaoxo owooomononz .mo oHaoH 112 oHoooowunooO NnOuuowmwuon NHnooonN non NonwanoO NnOuunmeunm NHnnnonN uoo NouHonoO NnOuunwnwunm NHnononN nan NoanHwoO canuouwoow on oHnonowunooo nowumuwoow oz nonunowoon oz NnOuunwnwuon NHnononN non Nounowwon oowunononw 0: ooHuouHoaw oz nowuoowoow oz ooHnonnou wo-oonmoO. nNON @000 @006 @000 uoownoN anN uoownoN uuownoN ooou noownoN uoownoN uuownoN nowuocmwnoa IIHoomHO onunoa no oocooa on» Onwn: nonoun oonwnaoo onu wo conuOuonNnouoH NN Om Nm Om OOH OO OOH OOH Om OOH OOH OOH UGUUHGN Nonoun oocwnaOO ammo N ammo O ammo O Noo n ououmononuouoowaowoooonnuo annoonmnuou NOOH.O + oonxonono aonoom NM ammo N ammo O nmoo O Nno n ouooouoHO aswoom NOOO.O + oonxonoNo asnoom No nNoo N nmov O nNoO O Nov H oonxonoNo aonoom Nm unoauoona odudv onownnoaan 03u now moan> coo: .nuoooonn nownonnoo wo Hw>oaon onowoa naounNn oou ow nooawooNn one ooo.aoHu:Hoo can wo Ban ago now onoun oHannON aoawxoa any wo uOnooou InoN n no OoononNXo oawu wo nouwaoH.unwnowwHo now oounuu nauummn osu wo nowwwonou HHoIno>O .Oo oHnoa 113 ON.¢ mo.O ON.O H0.0 O0.0 OH.O O0.0 HN.N N0.0 OH.O momma HHnIno>o unoO< no.“ NN.o no.n Hm.o no.4 oN.o oN.o mn.~ mo.~ m~.n mN.m mn.m NN.N oo.~ mo.N mm.m no.n mn.o mo.N on.m mo.o mo.N oo.m mo.m mm.m Om.N mo.N mm.m on.m no.m mo.N mm.o mm.m oo.m om.m wo.N mm.o mo.w mm.o NN.N on.o mm.N mo.o mN.o mo.m mo.m mo.m oo.o mm.m Om.N IImI. .Imn. .Imw. .IIMI .Imm. OGOH OHOSH OoosuNaOMmm qmmoH nouns: anon manna HHoIno>o nownnanH Nonmon.NoNNoon ooonooonoonNo aowoononuou NOmNAO + oonxooonn aonoom Na AONONH.¢ONOnzv aunnmnonm aowoonwnn NNO0.0 + oonxouoNo aonoom Nn ouououoonuouoowaowoonoHNana anwoooonuou NNH0.0 + oonxouoNo aonoom Nn onounconnuouocwanwoonoHNOuo nonvononuou NOOH.O + oonxonoNo annoom Na ounuouoonuouoonaowooooHNnuo anwoononuou NON0.0 + oowxonomo anwoom NO ouooouoHO annoon NOOH.O + oonxouoNe sonoom Nm ouooooaHO anwoon NOO0.0 + oonxonomn annoom Nm muonouoHO aowoon NNOO.O + oonxouoNn annoom Nn oonxouone annoom Na nucoO< .ouooonnoo Bowano on nnoovnnn HooOu noHHom noN noHnnO NH wo nooHuoHon nouns Ono: nunownouoo oanoOHn ow NnNoO Nv Ono: ono now ounnosaw nnoawuomn Hooun OOOH O .02 HOH< OoHHon OHou wo muHonon nnoH uOOHos downonnoo .No oHOnH 114 Aneuaa OH.ouoONaoanonNN lwoooonuou flOmNO + canon—um: Eamon NO .OO "nous: 3.33.— ..ooAN anon-«nu NNO0.0 + canon—v.3 Bow—vow NO :3 NonnuoudnuouoswflowooaoHNOuo sumo-snug NNHO.O + sowxonomo annoom .NO .OOOO mouououoonuouonwaowvoaoHNous gwooaonuon OOOH.O + 03.8.33 apnoea NO .NOOO NounuouoonuonoowaowooasHNn—uo swoooonnou HONO.O + wagon—{E Bow—vow NO .HOOO «cannon—HO flow—5o NOOHO + «3.853 Snoom N... .32 m3.3.83» 93o. N486 + 32893 338 an .82 «32883 332 .385 + «3.8.62 9:8... a... .82 83:83.. 5:8 .8 .ns 83:. on... not. so! 332:3 one». Nennouoooooo oawHou—Hs on moons-IN aaslwusNa Hooua wo nowaonnou wo annals can we aoawnsnloO .OH .OwN £504 _UNN NUNN v UNN __ mm Nonm v? .Umm Vunn l I; m g D I IN N m a D .0 [n u FIJI _O.m an a 8.... m. m m. 0?.0. m Alu. u m2.» 0: s s r Inn w b w t .2 men. A _NN II I l 0 de 115 presentation of the results.) An analysis of variance of the weight loss data (Table 47) for the steel specimens immersed in the alkaline detergent solutions is given in Table 48. The analysis shows that the agent, immer- sion, and interaction effects are each highly significant. The agent and Table 48. Analysis of variance of corrosion weight loss data for steel specimens immersed in alkaline detergent, hard water solutions. Source Degrees of sum of Mean F freedom squares square Total 179 887.87 ' ** Immersions 4 47.96 11.99 33.87** Agents 8 720.78 90.10 254.52** Interaction 32 71.29 2.23 6.30 ‘ Error 135 47.84 0.354 *‘A‘ Significant, 1% level immersion over-all means are ranked and tested for significant differ- ences in Figures 16 and 17 respectively. The results are presented in a manner that allows not only comparisons of agent effects, but also com- parisons of specific agent effects to other agents at more than one con- centration, and still further comparisons of the effects of AI and AII concentration on corrosion. The Specific effects of concentration of AI and AII on corrosion are analyzed, and discussed in detail in the last section of this dissertation in order to allow direct comparison of these effects in distilled water and hard water. Visual examination results for the steel specimens scored according to Darrin's (1946) system before and after corrosion products removal are shown in Table 49. The visual examination scores of the solutions are given in Table 50. Included in these tables are percent figures repre- senting the total scores of the four replicate specimens or solutions expressed as a percentage of the total possible. In addition, Table 51 presents the over-all condition of the system, sum of the liquid and the specimens in the system before removal of corrosion products, as a per- centage of the maximum possible score. The 3% sodium hydroxide control showed considerable discoloration and general corrosion before corrosion products removal. After their removal less discoloration and general corrosion was observed and the effects of some local corrosion became 116 6 , 5.64 x 5. so _ 4.52 § 427 ‘43I '--n \ E 8’ = 3*- 1 5’ «r . s 2 In 02 2 I- _ E .9 S 'F - 0 4 2 I 3 5 Immersions Fig. 17. Comparison of the amount of corrosion obtained with different imersions of steel specimens in alkaline sequestering agent solutions made with hard water. IH 0% H9053 Vienne— - UnnpwuaINUUUNU CC! pens—ran or“ Iiiihwuii 117 ooH o¢ cm on ooH oe ca on mm mm N¢ mm coH o¢ No um ooH oe Nm hm ooH 0* mm mm ooH ow mm mm ooH co cm on mm mm on Nn ooH o¢ «noun Hana» asaqua «noun -Hou uuu¢3.uuwv acounumaEH u>Hu you aoaHu> can: .maoawumam Human mooH o .02 HmH< voHHou vHou do muouu: uaumuouuu ucaHuxHu no magnum» aofiaouuou on» no aofiuuaaaaxo agnouuouuux .ae «Hana NH oH NH oH NH NH NH NH NH NH NH HH NH HH NH m m 0 NH aoHu nouuoo mo uamuuum Huuoa Huumaoo wdHuuHm [Illllllllly .Uuun \ NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH NH INCD QO‘ ”0‘ [\m 0‘” mm O‘C‘ O‘O‘ NO\ m 60H¢ nouuoo Huooq ¢ . m «ho Houw< c N MAO vacuum ¢ n mun wound ¢ N duo vacuum Q n duo uuumd N mac ouowom duo Houm< amoouomum ‘3'G 01¢) amo Houm< who vuoumm mac umum< «mo ouowom ume umum< mmo muouum «mu umumd duo ououon mac hound Mme ouommm uuoua BSBHuqz ~¢ ~¢~3 ~¢~¢ ‘3‘: ~¢~¢ ~¢~3 MF'N NM 010') NM NM Houmofi.so~m¢azv «unnaoonmouan Ebvauduuuu flnmN.O + ovgxounan auguom Hm Hous~a.eomnuzv ouunauona aufluouwuu sumo.o + ovfixouuas aaueom um ouuuuuuuuuuuuaHaaHvuauHmnuo auauoaauuou «Nan.o + ouwxouumgvaawuom um oumuoouuuumuoawaaHvuuuHmnuu asgvouuuuou nom~.o + ouaxouvmn asavom an ouuuouuuuuouocwauwvuauHanuu 35NUOnQHUUU Nwhc.c + oufixouvzs aaflvom so ouucouaHm abwvon Ron.o + ougxouums asfiuom an «uncouaHm aawvon semo.c + uufixouumn asfivom an uuaaouaHm aaHvoa NNeo.o + ouwxouvmn aswnom an uuflxouuan asauom Rm waHaw GOHua -gwsom -uoHoumHa coHuuuHauxm “coaquss i 118 AoNucH.Non¢¢2v «unanuosmouha > 8:300:33 RnoN .o + an ad o ouuxouuag azauom an HonNH bangs 3233a.— auwvoouuu fluno.o + no «g o «waxouaan asauom an 3333.30usages—«Hana» sauce-auuou RNHn.o + an «a s oufixouu»: asflvom an 3332.30uocHaqumauHmnuo 523-850» flwan + an s“ s oufixcuuma asfivom um cuuuuuquuuuuocwaugunoHanna aswvouquuuu sumo.o + no “a s ouaxouua: sauuom Rm 3333» 9:3. $36 + as a“ o oufluouuas aswvom an 3282:» 5.2.2. 830.9 + on o“ .5 ouflxouvan.as«uom um . 3.333» .323- N§.o + as «H o unauouuha asfluom an an “a o ouuxouvas asuuom an 2: ON a «you. gun: oucuo Huuou ape—«Ian onco- manna—woman mo uauoHom H38. Havana 33:30qu :ogwaofio 53315:“ unanimous / ILII acwuuav ucowuuoaaw 0>Hm now auaHu> can: .aamaHuunu Huwua mooH 9 .oz HmH< voHHou vHou no =0HnuoaaH uuuma HuaHou noun: than quauuuuuuv ocHHoxHo unu mo coHqucoo mnu mo :oHuucwauxu anounouunz .om oHnoH 119 Table 51. Over-all condition of the alkaline detergent, hard water systems expressed as a percentage of the maximum possible score for the sum of the solutions and the specimens in the systems before removal of the corrosion products. Mean values for fiVe immersions data. Treatment Combined Interpretation of the combined score, scores using the method of percent Darrin $1946) Designation 'Qggree of corrosion 3% Sodium hydroxide 78 Fair Questionable 31 Sodium hydroxide 85 Good Definite,but prob- + 0.042% sodium gluconate ably satisfactory 3% Sodium hydroxide 92 Good Definite, butrprob- + 0.0842 sodium gluconate ably satisfactory 31 Sodium hydroxide 90 Good Definite, but prob; + 0.168% sodium gluconate ably satisfactory 32 Sodium hydroxide 90 Good Definite, but prob- + 0.078% tetrasodium ably satisfactory ethylenediaminetetraacetate 32 Sodium hydroxide 90 Good Definite, but prob- + 0.1561 tetrasodium ably satisfactory ethylenediaminetetraacetate 31 Sodium hydroxide 90 Good Definite, but prob- + 0.312% tetrasodium ably satisfactory ethylenediaminetetraacetate 3% Sodium hydroxide 82 Fair Questionable + 0.0571 trisodium phosphate (Na3P04-12H20) 3% Sodium hydroxide 85 Good Definite, but prob- + 0.293% tetrasodium ably satisfactory perphosphate (N34P207-10H20) ‘1' apparent. The Alsolutions caused slight discoloration and general cor- rosion cf the specimens at the lower concentration and some slight general corrosion was apparent at the upper concentrations before corrosion prod- ucts removal. No visual corrosion effects were apparent after removal of corrosion products. With All discoloration and local corrosion of the specimens were apparent before, but not after corrosion products removal, except some local corrosion was apparent at the upper A con- II centration. Discoloration, local and general corrosion of the specimens immersed in A111 and ATV solutions were apparent before corrosion prod- ucts removal but not after. 120 The condition of all the solutions employed in this study, which were made with hard water, were considerably poorer than the solutions employed in the previous study, which were made with distilled water. The 31 sodium hydroxide, AI, and ATV solutions tended to show similar appearances, with some solution cloudiness and precipitation. The A11 solutions exhibited less precipitation, less cloudiness and a better general appearance, while the A111 solutions exhibited the poorest over- all condition with considerable cloudiness and precipitation. The over- all condition of the systems evaluated is summarized in Table 51. Distilled water and hard water studies combined and compared Distilled water versus hard water Corrosion weight loss results for steel specimens immersed in dis- tilled water and hard water of 12 grains per gallon total hardness.as calcium carbonate are presented in Table 52. An analysis of variance of the data is given in Table 53. The difference between the water over-all means was highly significant, with distilled water being the most corro- sive. The immersion over-all means were ranked and tested for significant differences. Immersions l, 2, and 3 which were not significantly differ- ent from each other were each significantly more corrosive than immer- sions 4 and 5, which were themselves not significantly different in over- all effect. Visual examination results for the specimens and solutions are presented in Tables 54, 55, and 56. Distilled water, distilled water plus tetrasodium ethylenediamine- tetraacetate,'and hard water effects compared The distilled water and hard water weight loss data of Table 52 and the distilled water + 0.1562 tetrasodium ethylenediametetra- acetate weight loss data of Table 22 were analyzed together by analysis of variance. The analysis_of variance results are presented in Table 57. The agent and immersion over-all means are ranked in decreasing order and tested for significant differences in Table 58. Alkaline detergent, distilled water and hard water studies combined and compared The nine alkaline detergent solutions made with distilled water can be compared with each other and with the nine alkaline detergent solutions made with 12 grain per gallon hard water by an analysis of variance pro- cedure which treats these solutions as 18 different solutions (or agents) if homogeneity of variance exists between the two sets of data. The 121 mN.¢NH Hm.mHH mopmmo cocoa HHowyo>o uaom<\ mH.NHH Hm.¢HH cm.mNH ow.¢mH mm.mNH «some HHouuo>o ooHuuoaaH coHHom you ma.mm om.moo ow.o- mmonwo nm.o~H magnum NH .uouua sun: n¢.mNH mm.¢mH o~.mmH mo.m¢H mm.mMH nouns uoHHHuoHn _m .q. Lon N .H II, ooOHouanH auoowd xooanNavVuE «mooH wmmHoz one: .Aomov NV 3003 0:0 you soon: 6H voouoaaH acoaHoomo Hoouo wooH 0 .oz HmH< voHHou uHoo wo wuHooou ooOH uanok ooHnouuoo .Nm oHooH 122 Table 53. Analysis of variance of corrosion weight loss data for' ' steel specimens immersed in distilled water and hard water of 12 grains per gallon total hardness as calcium carbonate. Source Degrees of sum of' Mean F freedom sguares sguare Total 39 12544.56 ' ** Immersions 4 3221.11 . 805.28 5.70** Agents 1 4787.35 4787.35 33.90 Interaction 4 "299.96 74.99 0.53 Error 30 4236.14 141.20 **Significant, 11 level. 1Denotes distilled water and hard water. sampling errOr variance of the distilled water solution data (from Table 27) was compared to the sampling error variance of the hard water solution data (from Table 27) ‘by Cochran's Test for homogeneity of variance (Dixon and Massey (1957)). Cochran's Test showed that the hypothesis of equal variances should be rejected at the 11 level of sig—M nificance (the calculated value was 0.6448 and the.critical value for re- jection at the 12 level was 0.6146). Since homogeneity of variance-was not found at the 11 level of significance for these two sets of data one can not combine or analyze the two sets of data together without invalid- ating to some extent the analysis of variance test for means. However, in order to allow approximate comparisons of the various treatments the data was combined and analyzed together. The analysis of variance re- sults are presented in Table S9. The agent over-all means are ranked and tested for significant differences in Table 60. These results are to be interpreted as approximate only and the conclusions weighted accordingly. Concentration effects of sequestering agents in alkaline distilled water and hard water solutions. The sequestering agent concentration effects in alkaline distilled water and hard water solutions are presented here to facilitate direct comparisons between these two systems. Analysis of variance results for each agent and water type-solution are presented in Table 61. The concen- tration and immersion over-all means are ranked in'decreasing order and tested for significant differences in Table 62. The effect of sequestering agent concentration on the corrosiveness 123 ooHHom won 8 e H ~ H «39.5 NH 632. 3am w H o H o ~32. eoHHHuza ooH ON a a .H swoon 6253 «noon Houou -agaHmoa ouooo sooouoonmo mo uncomom Houos Houooou ououHoHooum. noooHvooHo ooHuooHauxu uooauooya .wuov nooHouoaaH o>Hw you ooaHo> one: .oooauoooo Hoouo wooH 0 .oz HmH< uoHHou vHoo mo ooHouoaaH woumo nooHuaHom.uouo3 who; was nouns voHHHuoHv on» no ooHuHuooo ago no noHuooHauxo uHmoooouuoz .mm oHnoH as Hm m «H a e H «.5 ~33 , :2qu .8.— 8 an H HH m m o «no 3&3 2:3» «H :32. Ba. 2 on .c NH m e H «.8 $34 8 3 H, HH m n o «no 383.. ~32. 32335 2: 3 NH «H a .H n 988 5&3: fluouu HQUOU _ dowu flown .ananga ououo nouuou nouuoo onoo ooHuo uo unoonom Houoa Houooou wdwuuHm HuooH uawooa uuoHoooHn ooHuooHaixu uooauooua .ouov nooHouuaaH o>Hm now nooHo> one: .nooamwomo Hoouu mooH 0 .oz Hand uoHHou so soon: who: was noun: voHHHuoHu no euuowmo ooqoouuoo oau mo oOHuooHiixo oHnoooouonz .em «Hons 124 Table 56. Over-all condition of the system expressed as a percent- age of the maximum possible score for the sum of the solution and the specimens in the system before removal of the corrosion products. mean values for five immersions data. Treatment Combined Interpretation of the combined score, scores using the method of percent Darrin (1946) Designation Degree of corrosion Distilled water 40 Bad Severe corrosion Hard water, 12 grains 47 Bad Severe corrosion per gallon 1. Table 57. Analysis of variance of corrosion weight loss data for steel specimens immersed in distilled water, 12 grain per gallon hard water and distilled water plus 0.156% tetrasodium ethylenediaminetetraacetate. Source Degrees of sum of Mean P freedom? sguares' square Total' ’ 59 25620.98 “' " ** Immersions 4 5550.26 1387.57 14.31** Agents 2 14859.68 7429.84 76.64 Interaction 8 848.09 106.01 1.09' Error 45 4362.95 96.95 " ** Significant, 1% level. Table 58. Comparison of the amount of corrosion of steel specimens immersed in distilled water, 12 grain per gallon hard water, and distilled water + 0.1561 tetrasodium ethylenediaminetetraacetate. Agents Agent Immer- Immersion over-all sions over-all mea s, mea s, dm -week dm ~week Distilled water 151.73 2 143.54 + 0.1561 tetrasodium ethylenediaminetetraacetate 1 142.34 Distilled water 135.19 3 136.64 Hard water, 12 grain per 113.31 4 125.48 gallon 5 119.03 A 125 Table 59. Analysis of variance of corrosion weight loss data for steel specimens immersed in alkaline detergent solutions made with dis- tilled water and 12 grain per gallon hard water. A T ' Source Degrees of Sum of N ‘Hean F freedom sguares sguare Total 359 1316.28 . - ** Immersion 4 62.06 15.52 56.42**, ‘Agents 17 943.05 55.48 201.73**' Interaction 68 236.89 3.48 12. 61 Error 270 74.28 0.275 ** Significant, 12 level. of alkaline sequestering agent solutions can be assessed analytically by determining the regression of weight loss on concentration. The concene tration data for each agent and water type were analyzed for the follow» ing: (1) regression, (2) linearity of regression, and (3) when linearity of regression was found, the linear regression predicting equation was determined. The results are presented in Table 63. The.variation in corrosion weight loss as affected by concentration of sequestering agent (sodium gluconate or tetrasodium ethylenediaminetetraacetate) in 3% sodium hydroxide solutions made with either distilled water or 12 grain per gal- lon hard water is shown in Figures 18, 19, 20 and 21. 126 2.“ A _ . , 83333.3 .588 so + 33: 33.33: H co.n wousa_NH.¢uonooo£m aawuoowuu Nsmo.o + ouuxouv a nonvoo uh + nouw3.uoHHHuan c . , ououuuosuuou Ho.n nonwlsaooooHanuo soHuooouuou NNHn.o + ouonuu : Esauoo Rn + nous: vusm «Una . ouououoouuouooqsswu . mH.n nooonnuo sawmooouuou um~o.o + uvHNouv n asHvoo Na + nouns voHHHuofim Hon . nous: oH.ouwnm «H.m noonmoummgasHvooouuou flmmN.o.+ ovHuouuan saHuoo Rn + nous: voHHHuoHn n . . ousuooosuuou mH.m aooHEUHoooonnuo asHvoosuuou sumo o,+ ouwxouvhnkaswuoo 8n + nouns boom Honn 3." 3333a SSSRNSS .+ 8333.3 933 «n .+ 332. 333.3 Hon mN.n nous: NH.ousnmoonm asHuooHuu.Nsmo.o + ovonuvhn asHvoo Na + noun: can: «a 36 3383» 5:38 .985 + 3on.62 5:3. «n + ~32. 333.3 «3 . uuouooosuuou , . nn.m noowaoHvooloAuo asHvooouuou NonH.o + sowxouvmn.aoavoo an +.uouos whom Nona A. ousuoooouuouooaauwv 9." , .3393 338333 .5an + 3833.? 52.2. an + 833 3:33.. «on ., nouns oH - - . . $6 . ageing—8%.. 533833 .536 + 8338.6? 538 «m + ~33 Bum mm Hna 32.83» 9H3. .536 + «3333-532 an +_ ~32. 32343 «3 wan auououoouuouonHaoHv . . . {4 -38152838333 «Sn... + 338%.. 588 .3 +833 83333 3n 2.4 3onqu .538 «n + ~33 33 .HH move uuoooosHmjaswuoo non.c + ouonuuhs isauoo an + nouns own: «ONN HN.N monsoon w luHooo Nonc.o + ouwxouumogaSHvoo Na + noun: can: NUNN umqm uaoooasHm asHvoo RN¢O.O + oonouu A Bowvoo 8n + nouns can: HONN . » some: sowum . . novoo «cocoa HHowuobo nomad A ouoow< uoow< ;; g . , - . .uouns who: ooHHou nwmtooHowu NH so nous: voHHHuoHu nuH3.omoB oaoHuoHoo uooufllumw uuouoosuoo osHHoxHu oH woouoaaa oouaHoooo.Hoouo mo coHoouuoo mo unooau ago no sooHuomaoo .co «Hana. 127 .Ho>oH RH .uaaoHuHeuHm .Ho>oH um .uaaOHMHamHmue Nae.o MN.HN me oon.o NN.oH no woman mmo.c mHoqo nn.n w Hc.N «no.9 . Om.N w , . dowuumuouuH --ocHH condo oqu N «ma.~ No.H 8c.N N «aoHuauuamuaoo «exaHNH HHHo .meq8H .3 4HmH.ea ‘ H~.o~ «n.8oH 8 aeoHauoaaH . _ ._.. modem an .. . mn.~mH an i. Hagan ououoooouuouoowaowooolosuo, asHuooosuou + oeHuoueaa aeHeom um -- omN.c MN.MH no MHH.o no.n 3 no uouum «st.m .oqu quw w stoo.o mmn.o No.0 m ooHuuououoH «snoqmo omqu ;Noimm N som.¢ .oon.c _ NH.H N nooHuuuuoouooo fi¥NHumH «can. _ oanN q *¥n@.w¢ cm.m mm.HN e uaowmuuaaH msqmm mm ON.¢m mm Houoa . ouoooosHm asHvoa +. ouonuvae asHeom an unmade nouoomo aovoouu ouoomo oouoomm Housman .1: m comm mo ism uo nooumon m can: no Bum. no «common noHHou non uonmu NH .uouoa who: nouns ooHHHuoHn ensues oouoom .AnHo>oH ooHuouuoouooo omega no use: unomd wdHuouooovom sooov ououoooouuouoowaquoolonuo anHuOoouuou onHm oonouvhs aoHvoa Rm vow ouooooon aoHuOo ooHo ovonuu»: asHooo Rm mo nooHuoHoo soon: was: ooHHom woo oHoum NH van nouns uoHHHu aon ow voouoaaH nooaHoooa Hoouo you «and nooH uanos ooHoouuoo mo ooooHno> mo «HozHood «Ho oHouH 128 ououoooouuouoowsaquuoonauo flowvounuuou + oeHaouuse asHeom an ouseouon sawvou + oeonuv e asHeom an Ho.n ~HM.° _ nH.m oao.o thm oxo.o _ n8.m - omH.o main .onH.o am.m NHm.o no.8 moH.o _ mH.n Neo.o H~.a 3a¢.o _ m~.n eao.° an.» Heo.o Hn.m moH.o soon- Bowman. x803: .HH—{fl a we advance «unowo «moose unmouom «nouns . Hwouuo>o wdwuouooovoo uo HHosuo>o onnounoovoo mo gflUflHUflQUgU GOHUHHUNHWOQOU GOHUQHUQQOQOU fiOflUCHUflUUOOU ooHHou non ocHumu NH «mugs: who: nouns euHHHuan muons: ouoow< jl .uouoa who: ooHHow use oHouw NH one noun: voHHHuoHu :uHs owns unoHuouuoooooo udouomuwv mo nooausHoo nomad onhuunooooo ooHHoxHu oH uouuoaaw nooaHoooo Human mo ooHoouuoo mo unooao any «0 ooowuomaoo .No oHnoH 129 .N :0 M mo unoHonuooo ooHooouwuu HoooHH ago no onHH one no omoHo any «was .umoououoH 1% oar no .Nozoox o Bonn » 5553: no «0 muse—Hana 05» "k. N x.» .25”. I + o «xv on no N a + e um? mo oooHuosvo Houooou . E use .823 HawHo mo Hm>oH 85. “V .ooHumHomom on”. now onus ow. unoHonmooo ooHooouwou onu uosu 933325: 05 now moon. no u Nwa "m Bus unouoc no.9 u Bum umooud O In C 0 ll N 0 0 $386 8 2. 3 H a + 8:5 n w m3 35 u .83 3384 Hod 3.0 n Bum 3.83.3 Bun umooo< no.0 u .8: unooo< NmmNH + 306 n w no.0 \( 310 u 8...» uuwfiom Bun unooo< no.0 u uouo H3034 NHcoN + moo.n n a mo.o 2 oooHuosuo ooHoooumou ooHooouuou mo huHuoooHH "m o n “SM u: soon: mouoaHuom «wow anon. «you noon. . a. 33«unannouooHeoHuoolonuo asHvooouuou + «3.83.2 32.8 am 32303» Baton + 03:93.3 gHuom Rm ooHHow you .2323» NH Conn: chum ouououoouuouonHaoHvoolonuo SHuooouuou + 26.33533 gHvom Rm oumooonHw HHS—coo + 83.8.82 5:88 am noon: voHHHuoHn ouaomd . CC gnome moHnouooouou mo aoHuouuoooooo on» no va nooH ustos :oHoounoo mo ooHooouwou mo «HohHoo< .mo oHooH A 130 6‘r——r—" I f T l I l r 5 - H .8 a» c: .34» - NE ‘1, U ___(‘ \ ,— 83- 4' . «r? (0 9 2r. «.4 E .9 O ExperimentoI 0090 g ' - Over-all Means H 0 1 1 4 4 1 1 1 1 0.02 0. 04 0.06 0.08 0J0 0. I2 0. l4 0. I 6 0. l8 Concentration. oercent Fig. 18. Corrosion weight loss of steel specimens immersed in alkaline solutions made with distilled water and containing differ- ent concentrations of sodium gluconate. The estimated linear re- gression line (weight loss versus concentration) is drawn. 1.006 T I 1 T —' .K 0 o a "S 0900 -+ \ ' r— O E u? m 2 3.0.800- - 3 '5 OExpsrimentol Dots E Over-all Moons '5 0.700 - -1 8‘ _1 1 1 1 1 0 5 IO IS 20 25 Reciprocal of concentration, percent Fig. 19. Corrosion weight loss of steel specimens immersed in alkaline solutions made with 12 grain per gallon hard water and containing different concentrations of sodium gluconate. The esti- mated linesr regression line (log of weight 1088 versus reciprocal of concentration) is drawn. 7- .. 6- .H x 0 Q3 3 N. 5’- —1 E 'U \ U’ Ea~ H 6~ r‘fiD 1n / 2 31— «- E .9 i’ 2’ " 0 Experimental Data Over-all Means 1 1 J 1 1 l 1 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Concentration , percent Fig. 20. Corrosion weight loss of steel specimens imereed in alkaline solutions made with distilled water and containing different concentrations of tetrasodium ethylenediaminetetra- acetate. 132 I I I f i I 1 7i— .0 6- .H x 3.3 I N 5— c— E '0 \ 0" E 4- — loss , 58 1 \ ‘0 _4 E .9’ a: 2- ~ 2 l r- 0 Experimental Data -‘ Over-all Means 1 1 1 1 1 1 1 0.00 0.05 0.10 ms 0.20 0.25 0.30 0.35 0.40 Concentration, percent Fig. 21. Corrosion weight loss of steel specimens imersed in alkaline solutions made with 12 grain per gallon hard water and containing different concentrations of tetrasodium ethylene- diaminetetraacetate. 133 DISCUSSION OF Rssgirs The problem under investigation was the determination of the corre- sion inhibiting or accelerating effects of organic sequestering agents, one of the hydroxycarboxylic acid type and one of the aminopOlycarboxylic acid type, relative to the effects of the more traditionally used inorganic phosphates and the other components of the system (water, alkali, and hardness-forming salts in water), under conditions similar to and varied in accordance with what is commonly encountered in their use in cleaning reusable glass containers in mechanical washing machines. Studies on materials and methods Materials In the experiment to determine the susceptibility to corrosion of cold rolled and hot rolled steel (AISI No. C 1008) in distilled water it was determined that the difference in amount of corrosion obtained with the two steels was highly significant (11 probability level of signifi- cance), with cold rolled steel found the most susceptible to corrosion. It was also determined that no significant difference (51 probability level of significance) existed between the two steels in the alkaline solutions. Since the agent, immersion, and interaction effects are of secondary interest at this paint their discussion will be delayed until the distilled water-sodium hydroxide-sequestering agent section. The difference in results for water and for the alkaline solutions could be explained as follows: In both cases a maximum of no greater than 6.78 ml of oxygen could become available for reduction at the cathodes. A maximum oxygen availability of 6.78 ml could theoretically account for about 33.8 mg of iron being corroded to the ferrous state. In cases where 6.78 ml of oxygen is available for reduction, and the corrosion obtained does not exceed to any great extent the 33.8 mg corrosion of iron, the corrosion would be classified as predominately oxygen-type corrosion. In cases where the corrosion greatly exceeds this value the corrosion would be classified as predominately hydrogen- type corrosion. The grand over-all means for the steels in distilled water and in the alkaline solutions were 136.31 and 3.67 mg7dm2-week, respectively. It appears then that hydrogen-type corrosion predominates in the former 134 . and oxygen-type corrosion predominates in the latter.. Uhlig (1948) in his review and discussion of the effect of cold working on iron and steel corrosion pointed out the following: Cold working has no appreciable effect when oxygen diffusion controls the corrosion rate, but when corrosion is, by hydrogen evolution a definite effect of heat treatment employed to re- lieve internal stresses, results in decreased corrosion. Johnson (1946), Speller (1935), and Evans (1960) have also reviewed and discussed the ef- fects of cold working on corrosion of metals. They note that the corro- sion rate of iron and steel is accelerated by cold working, particularly if the cold deformation is localized and that the action of chemicals upon strained and unstrained metals is different. Evans.states that straining can greatly increase corrosion velocity and that the strained metal acts as a more reactive material than the annealed metal. Speller. states that the effect of strain may be due to variations in surface-film protection, caused by the breaking of the film on certain areas, or to differences in surface finish of the strained areas, or to differences in grain distortion. Johnson, Speller and Evans did not make any dis- tinction between effects of cold working in hydrogen-type and oxygen- type corrosion as did Uhlig. The results reported here are consistant with Uhlig's interpretation of cold working effects on iron and steel, undergoing either hydrogen- type or oxygen-type corrosion. The distilled water results in which hydrogenbtype corrosion appears to predominate showed a highly signif- icant difference, with cold rolled steel experiencing the greatest amount of corrosion. In the case of the alkaline solutions1 results, in which oxygen-type corrosion appears to predominate, no significant difference between the steels was found. Based on the findings, that the steel means were not significantly different in alkaline solutions similar to those to be studied in this investigation, it was decided to use one type of steel. Since the cold rolled steel specimens had more uniform dimensions and were easier to prepare for testing (degreasing and pickling) they were selected. Methods It was determined that different individuals could reproduce experi- ment results (when the 52 probability level of significance was used as 135 the criteria of reproducibility) using the experimental procedure to be employed in the subsequent studies. It should be noted that individuals 8 and C both had a college degree and considerable training in chemistry and chemical techniques. Individual A on the other hand did not have a college degree, but did have some chemistry training and some laboratory technician experience. One might,in the opinion of the author,state that individuals B and C demonstrated a higher degree of interest and conscien- tiousness relative to the studies, as well as better laboratory techniques than individual A. This opinion appears to be justified by the fact that individuals B and C had considerably lower and quite similar values for their individual over-all standard deviations and mean errors than indi- vidual A. Despite these differences, the results obtained with these individuals were not significantly different and it would appear then that the experimental procedure as developed has merit with regard to the above. §tgdies on distilled water-sequestering agent systems It has been determined that marked differences exist in the effects of various types of sequestering agents on the corrosion of mild steel. While sequestering agents are not usually used in systems as simple as these in practice, knowledge of their behavior here is basic to explain- ing the behavior of sequestering agents in the more complex systems of practical importance. Distilled water solutions of phsophates (trisodium phosphate, AIII’ and tetrasodium pyrophosphate, AIV), a hydroxycarboxylate (sodium gluconate, AI) and an aminopolycarboxylate (tetrasodium ethylene- diaminetetraacetate, AII) resulted in 2.39, 25.1, and 112 percent as much corrosion as distilled water (Table 52), respectively. It is obvious that the differences in corrosion between these solutions and distilled water are highly significant, except in the case of the aminopolycarboxylate, A A statistical comparison (presented in Tables 57 and 58) has shown A:: to be significantly more corrosive than distilled water alone. The concentrations and ionic strengths, , ( presented in parenthesis) of the solutions used in units of molality, m, (moles of solute per 1000 g of solvent) were for AI 0.0039m (0.0039m), for AII 0.0041m (0.04lm), for AIII 0.0015m (0.009m) and for AIv 0.0066m (0.066m). These concen- trations represent the commercially recommended quantities of agents for 136 softening water of 12 grains per gallon hardness. These "use concentra- tions" were employed in all comparisons of agent effects in each phase of this investigation, except where concentration effects themselves were under study. In the case where concentration effects were under study the levels used were one-half, one, and two times the use concentrations. The use concentration represents, in the case of AI-and AII a sequester- ation ratio of 2 moles of sequestering agent per mole of metal ions to be sequestered. For AIII and AIv the ratios were 0.73:1 and 3.2:1 respectively. The discussion and explanation of results will be of a general rather than a specific nature because of the lbmited information avail- able on the behavior of these agents in these systems and under the conditions of interest. For example, chelate formation or stability constant values needed are generally unknown except for simplified sys- tems and limited conditions. In addition, the lack of specific qual- itative and quantitative information on the nature and distribution of corrosion products, imposes some restrictions on the degree of certain- ty that can be attached to the reactions and explanations proposed in the following discussions. An attempt is made, however, to present what seem to be reasonable explanations for the findings based on certain. V811 known relationships. The effects of precipating agents on metal che- lates (or sequestering agents on metal precipitates may be expressed in general by the following equilibrium (Chaberek and Martell (1959)): MA+B‘___ ml+A K," where A is a sequestering agent and B is a precipitating agent, which combine with the metal to give the soluble chelate MA and precipitate MB respectively. The exchange constant Xx is related to KMA’ the metal chelate formation constant, and KMB’ the solubility product of MB, by 1 Km . the insoluble metal salt, measured by Ex, is a function of both the the equation Xx = Thus the tendency toward precipitation of solubility product and the metal chelate formation constant. The pre- cipitation of MB may be prevented if Ex is sufficiently small. The equilibrium constant for the reverse reaction, KX" measures the tendency of the chelating agent to solubilize the slightly soluble salt MB and is equal to the reciprocal of Xx. 137 Further discussion of results will be by the individual types of sequestering agents studied. Sodium gluconate, AI(hydroxycarboxylate) Sodium gluconate is not an effective sequestering agent for ferrous or ferric ions in distilled water of pH 7 (Mehltretter, Alexander, and Hist (1953)). The approthate value of K , is equal to the product of 1.1x10-36, the ferric hydroxide solubilit: product constant at 1890, and 3.16x10-6, the gluconate-ferric metal chelate formation constant at 25°C (the gluconate-ferrous metal chelate formation constant was not given in Chaberek and Martell's (1959) tabulation of reported chelate formation constants), giving 3.48x10.42 which would predict that ferric hydroxide would not be solubilized. It is thus evident that the re- action or equilibrium for K MA +Bf—f‘1mlvi-A X I I would lie predominately to the right. Since sodium gluconate does not tend to solubilize (sequester) these ions, it might be expected that the amount of corrosion would be near the value for distilled water alone. However,such was not the case, and about 75 percent less corrosion was found with sodium gluconate in distilled water than with distilled water alone. It appears then that the sodium gluconate is functioning as a hydrogen-type corrosion inhibitor (by controlling hydrogen evolution). The supporting evidence is that the amount of corrosion (33.88 mg/dm2 -week) found did not appreciably exceed the amount accountable for on the basis of the calculated maximum amount of oxygen present. Since the amount of weight loss found is approximately equal to the maximum for oxygen-type corrosion it is expected that oxygen-type corrosion predominated. The addition of AI to distilled water would increase its pH. By increasing the pH (increasing OH' ion concentration) the solubility of the ferrous hydroxide would be decreased. By decreasing the ferrous hydroxide sol- ubility some corrosion inhibition would be expected in oxygen-type corro- sion, since the ferrous hydroxide surface film of corrosion products tends to decrease oxygen diffusion rate to the cathodes. undoubtedly hydrogen- type corrosion predominated in the case of distilled water, since the amount of corrosion found so greatly exceeded that accountable for on the 138 basis of the maximum oxygen (6.78 ml) available for reduction.. It would seem from the above that sodium gluconate may be functioning as a corro-. sion inhibitor, by limiting hydrogen evolution, which is the major corro- sion rate controlling factor in hydrogen-type corrosion. Evans (1960) with the aid of the literature has put forward some tentative explanations of the corrosion inhibition mechanisms involved here. He has pointed out that in the case of hydroxycarboxylic acids, the hydroxyl groups may help anchor the anions, which once they reach the metal surface are likely to be held fast. He notes two possibilities pertaining to removal from the surface: first the removal of the organic anion with metal attached, forming a complex anion or second the removal of the radical, leaving oxygen behind, as in the case of chromate. According to Evans, which of these occurs, probably depends on the rel- ative strengths of the chemical bonds which must be broken if complex- anion formation (leading to corrosion) or oxide-formation (leading to. passivity) is to be obtained, and without knowledge of bond strengths, prophecy is difficult. Evans also points out that certain organic compounds result in decreased corrosion rate and thishdecrease'has been attributed to adsorption of the compound on'the metal surface. He notes that the molecules tend to attach themselves at points where the greatest free force-field exists to attract them, and that these are generally the places where corrosion would tend to start. As a result inhibition of corrosion may occur. From the above it is apparent that at least two mechanisms of corrosion inhibition exist which would tend to explain the decreased corrosion found with distilled water plus sodium gluconate as compared to that of distilled water. The results reported here tend to agree with those of a previous study by Hieland, Maguire, George, and Kahler (1950b) where their inter- est centered on flow and corrosion in iron pipes. They reported 84 per- cent protection with 0.084 percent sodium gluconate in Philadelphia tap 3 at 120°r and flowing at 3.5 ft/sec, while we found 75 percent protection with distilled water at water of 42 ppm total hardness as CaCO 150°? under essentially static conditions. The smaller amount of hard- ness, lower temperature, and flow could have contributed to the greater 139 degree of protection obtained in the Nieland, Haguire, George and Kahler study. The visual examination data for the.two studies tend to agree, since in both instances a dark, hard, corrosion products film was observed. and pitting was reported at the gluconate concentration used in this study. It appears from the above that sodium gluconate could serve as-a useful inhibitor of mild steel corrosion in distilled water. Nieland, Maguire, George, and Kahler (1950b) found corrosion protection with hydroxycarboxylic acids in waters of greater hardness, but pointed out that care must be exercised in selection of the concentration to be used, since high concentrations may product pitting. However, they noted that the percent protection tended to decrease as the water hardness in- creased. ‘ Tetrasodium ethylenediaminetetraacetate, AII (aminopolycarboxylate) Addition of tetrasodium ethylenediaminetetraacetate (0.156 percent) to distilled water resulted in a significant increase in corrosion. The over-all mean weight loss was 12 percent greater with A than without it. 11’ An increase in the amount of corrosion waa*expected since it is known that A11 is an effective sequesterant for ferrous ions at about pH 7. Solubilization of the corrosion products is predicted by the product of the ferrous hydroxide solubility product constant of 1.64x10.14 at 18°C and All-ferrous metal chelate formation constant of 2.15x1014 at 20°C. The product of these two values gives Xx, value of 3.53, which would indicate that sequestration (or solubilization) of the ferrous hydroxide should occur to some extent under these conditions. The equilibrium for the following reaction would then tend to lie to the left, favoring some formation of the mefial chelate (MAII) X d MII+Bv—K' m¢+AII from the metal hydroxide (MB). A theoretical maximum of 0.1145 g of iron could be sequestered with 500 ml of 0.0041mAII solution. The Eesults of the study showed an added metal weight loss of about 16.5 mg/dm -week. The data above tends to indicate some sequestration of the corrosion products if the action of A is confined to that portion of the corro- II sion which was predominately oxygen-type. Sequestration can produce increased corrosion by decreasing the amount or thickness of the corro- 140 sion products film on the surface, which will increase the rate of oxygen diffusion to the cathodes, resulting in an increased amount of corrosion. The ferrous hydroxide corrosion products film is the primary corrosion rate controlling factor in oxygen-type corrosion; however, hydrogen evolution is the primary corrosion rate controlling factor in hydrogen- type corrosion. Since hydrogen-type corrosion was the predominate type. of corrosion (occurring after the more spontaneous oxygen-type corrosion) the possibility ofAII causing the increased corrosion by functioning in some manner that would affect the rate of hydrogen evolution from the metal surface also exist. It seems somewhat preferable to accept the explanation that the A11 participates and functions to increase corrosion by ferrous ion sequestration during the oxygen phase of corro-g sion. If the effect of AI was limited to the oxygen phase of the corro- I sion (maximum weight loss accountable for on the basis ofthe maximum oxygen available in these systems is 33.8 mg) then the 16.5 mg weight loss represents an increase of about 50 percent during the oxygen phase of the corrosion. ,It is entirely possible that A. may have affected both phases of the corrosion in a positive mannerliincreasing corrosion) or by affecting the oxygen corrosion phase by a greater positive manner‘ than suspected and then countered with a negative effect during the hydrogen corrosion phase such as to have given a net increase in corro- sion of the 12 percent found. It obviously can not be stated how the agent produced the given effect, but it seems somewhat more reasonable to suspect that the increase occurred during the oxygen phase of corrosion. With regard to the visual examination results obtained in this system two particular observations were noted: The first was the very noticable dark red-to-brown colored corrosion products (also pink-to- purplish tint in some cases) on the specimens and in the solutions. This noticeably different color for the All solutions, from that obtained with the distilled water solutions, probably represents the color of the metal complex formed. It is known (Chaberek and Martell (1959)) that such complexes exhibit various colors dependent upon experimental con- ditions. The second observation of interest was the adherence of these colored corrosion products to the metal surface. It was very difficult to remove these films with the prescribed corrosion products removal 141 process used throughout these studies. A visibly small amount of the corrosion products could not be removed after each successive immersion and the amount that remained seemed to increase with successive immersions. This difficulty was apparently responsible for the decrease in weight loss data found with each successive immersion. It would appear from the above that All should not be used with soft water in equipment fabricated with mild steel without considering the increased corrosion that could result. No studies were conducted with hard water solutions of A . However, it seems reasonable to expect a corrosion rate betweeilthat for the distilled water plus AH and that for distilled water alone; perhaps this value would be closest to the lower distilled water value. The basis for this expectation is that at least some of the AII’ if not all of it, would be involved or tied up in calcium.and magnesium chelates and the free A available to function as a corrosion accelerator would be decreased. II Trisodium phosphate and tetrasodium pyrophosphateb AIII and ATV (phosphates) ‘ The over-all mean weight loss found with the phosphates was about 97.6 percent less than that for distilled water. This reduction in corrosion is attributed to anodic corrosion inhibition. The concen- tration of phosphate anions (0.0015m, or 0.0066m, AIV) was consider- A ably greater than the hydroxyl anion coiiintration (about 10.7 moles per liter for water). As a result the product of the oxidation reaction (ferrous ion) can readily combine with phosphate ion, and since these iron-phosphates are insoluble they are immediately precipitated directly on or in close proximity to the anodes. This results in anodic corro- sion inhibition (Evans (1960)). The product of the cathodic reaction, sodium hydroxide, is soluble. The over-all reactions might theoretically be demonstrated as follows: 6Fe + 4Na3P04 + 30éT + 6H 2Fe + Na4P207 + 021 + 2nzo = pezrzofi .+ 4Na0H The appearance of the specimens and solutions was the same both be- 20 = zre3(ro4)2l + 12Na0H fore and after immersion in the phosphate solutions. However, it is known that ferrous-phosphates are white-to-blue in color, and this along with the small quantity of corrosion products formed was probably responsible 142 for making their detection difficult, and the general observation of no visible corrosion. The fact that phosphates decrease the corrosivity of water under'-- various conditions is well known by corrosion experts (Speller (1935) and ,Evans (1960)), but is not so well known by beverage industry personnel.~. This study serves to compare the relative.effects of these four agents on a quantitative basis which has not previously been done. 0n the basis of these results one might reasonably expect that the use of these agents in water rinse sections of a washer could result in reduced corrosion, particularly if more phosphate than that-required to- soften the water was used. It is known that calcium-phosphate complexes can act as effective corrosion inhibitors (according to Evans (1960)). The mechanism by which they inhibit corrosion has been described as'a cathodic one (Putilova, Balezin, and Barannik (1960)). They reported that glassy phosphates, for instance, migrate towards the cathodes in water containing calcium ions and that this differs sharply from the be- havior of phosphates in distilled water, where they move toward the anodes. They attributed the migration to the cathodes as due to positive- ly charged colloidal particles consisting of numerous diphosphate chains joined by calcium atoms and pointed out that the positive charge on the colloidal particles may be due to adsorption of calcium ions by colloidal calcium phosphate. Studies on distilled water-sodium hydroxide-seguestering_ggent systems Agent effects The corrosion weight loss value of 2.74 mg/dmz-week (or 0.07 mils/yr) found with the 3% sodium hydroxide control compares well with the reported literature value of about 0.1 mil/yr for studies on mild steel conducted under similar test conditions (Uhlig (1948) and Speller (1951)). Trisodium phosphate, (4), was not significantly more corrosive A than the control (1). Tetriigdium pyrophosphate, AIv (5), sodium gluc- onate, AI (202), and tetrasodium ethylenediaminetetraacetate, AH (302) listed in order of increasing corrosivity were each significantly more corrosive than the control. In 3% sodium hydroxide solutions an insol- uble layer of corrosion products, Fe(0H)2, will be readily formed. The solubility of ferrous hydroxide decreases from about 3.8 ppm to about 143 0 ppm (or nil) as the pH increases from 7 to 12 (Speller (1951)). The effect of the above mentioned agents can be explained as follows: In a 32 sodium hydroxide solution the equilibrium for the reaction Fe (0H)2 :Fe“ + 20H“ lies predominately to the left; however the ferrous ions in solution can be complexed by the AI, A11, A111, and AIv anions and form either soluble or insoluble complexes, thus tying up the free ferrous ions which will be replaced by ionization of Pe(0H)2 to re-establish the equilibrium. The above process should continue as long as sufficient complexing agent re- mains. As a result of these reactions the nature and distribution of the corrosion products can be modified to such an extent that the over-all corrosion rate may be significantly increased or decreased. Apparently _ the AIII did not appreciably modify the nature and distribution of corro- sion products by perhaps ferrous-phosphate formation and combination. with the corrosion products, and/or by precipitation of ferrous-phosphate at a distance from the metal surface and the settling of the complex to the bottom of the container. The AIV on the other hand did apparently have a significant effect through modification of the nature and dis- tribution of corrosion products. Sodium.gluconate's significant increase in corrosion over that obtained with the control would appear to be due to the fact that AI forms stable, water soluble iron chelates in alkaline solutions (Mehltretter and Watson (1953)). On the other hand, it is re- ported that A is not effective as a sequestering agent for ferrous ions II above pH 12 ("Keys to Chelation" (1959)); however, it was the only agent found significantly more corrosive than both the control and the A 1- 111 ‘° ution. Apparently even at a pH of 13.8 the A I possesses sufficient sequestering power for ferrous ions to cause :his increase, or the increase is due to some other effect. It can also be observed from the results that the three sequestering agents (A1, AH and ATV) produced significantly more corrosion than the control, while the non-sequestering agent AIII did not. Further,there seems to be'a tendency exhibited by the agent over-all means to indicate more corrosidn'with'organic sequestering agents than with the inorganic agents in this system and for the agents tested. The analysis of the alkaline detergent, distilled water solution co- rrosion weight loss data.(Table 27) showed a highly significant immersion- agent interaction effect. A significant interaction is one which is too 144 large to be explained on the basis of chance and the null hypothesis of no interaction. In the case of a significant interaction the factors are not independent of one another. The simple effects of a factor (such as agent) differ and the magnitude of any simple effect depends upon the level of the other factor (such as immersion) of the interaction term. This can also be stated by saying that the differences in re- , sponses to the various treatments (agents) are not of the same order of magnitude from block to block (immersion to immersion). An interaction may mean different things to different people with particular interests- and problems. Failure of effects to be additive is the measure of inter- action to the statistician. To the chemist an interaction is much the same as reaction. Steel and Torrie (1960) point out that where factors interact, a single factor experiment will lead to disconnected and possibly misleading information. The fact that the interaction and the differences between immersion over-all means were found to be highly significant was not entirely ex- pected. Both are rather difficult to explain. There are at least four factors which may be involved: (1) Corrosion media, (2) Laboratory ex- posure conditions, (3) Corrosion products removal, and (4) Metal factors. The first, corrosion media, was the same for each immersion, in that enough solution was prepared for at least five immersions and fresh solution was provided in clean plastic bottles for each immersion. Since solutions of uniform composition were used and the potential effects of the particular agent or agents are the same from immersion-to-immersion this would not seem to be the.source of the interaction and immersion variances. The second factor, that of laboratory exposure conditions, was adequately controlled as described in the experimental procedure. The third factor, that of the corrosion products removal process, would not seem to be the source of difficulty either since this was the same after each immersion. The composition of the solution and the process were controlled as well as possible. The data (Table 26) and the anal- "ysis of variance results showed that the variation in corrosion from immersion-to-immersion is highly significant. It would seem that since the corrosion media, the laboratory exposure conditions, and the corro- sion product removal process were well controlled and uniform from 145 immersion-to-immersion, that the source of variation is primarily related to the metal or metal surface. It is difficult to elaborate on this factor; however it is important and the results show that the amount of corrosion not only varies from immersion-to-immersion, but that the order a of immersions as far as corrosiveness varies from agent-to-agent. Analyses of the data did not indicate that the results of a given immersion were dependent upon another immersiodh results. It seems that the metal is in; some way exhibiting some control over the agents effect from immersion-to- immersion and from agent-to-agent. A study should be conducted in the_ future to determine if this, some other factor or a combination of-fac- tors is the cause of these observed variations. speller (1951) has pointed out that studies of the process of corrosion in the light of the electrochemical theory have demonstrated that the influence of sur- face finish and similar external factors may have much more to do with irregular corrosion than any variations in chemical composition or other internal factors. Such effects appear to be primarily responsible for the interaction found. While the immersion over-all means were ranked and tested for significant differences, the results found are not of appreciable practical value in this investigation. For this reason little attention will be given to them in the remainder of these discussions. Likewise, while the agent means were compared for each immersion, it would appear that in view of the role the metal itself may play in the over-all effect of a given agent from.immersion-to-immersion, that the agent over-all means averaged over all immersions provides the best est- imate of the corrosiveness of a given agent in relation to that of others. Therefore when the corrosiveness of different agents are com- pared the agent over-all means are used. In certain instances and for certain agents only, the weight loss data from immersion-to-immersion indicates additional information about the general effect of the agents as far as corrosion and these affects are discussed. There appears to be a decrease in corrosion from immersion l to immersion 2 with AI and AH for each concentration used (except, possibly for the 0.084% concentration of AI). It would seem that since the main difference here is a previous immersion in the second case and no pre- vious immersion in the first case that perhaps surface effects arising 146 as a result ofea previous immersion may have caused the decrease._ However, in Table 12 just the reverse effect, namely that of an increase in corro- sion was apparent from immersion 1 to 2 with ethylenediaminetetraacetic acid added to 31 sodium hydroxide as compared to the above described decrease with tetrasodium ethylenediaminetetraacetate added to 3Znsod- ium hydroxide. In addition, analyses of the control and the A data“ (Table 26) for dependence between immersions by linear correlaiion.co- efficient determinations and tests of significance for each possible combination of immersion pairs (10 possible) showed no significant corree lations and thus no dependence. The data for AI, A111, and AIv (Table 26) were tested for dependence between immersion 1 and 2. This data and the ' above indicated that the outcome of a subsequent immersion is independent of a previous immersiodb effects at the 51 probability level. The ex- planation for the variation in results from immersion-to-immersion with a given agent and for the order of immersion over-all means (Fig. 12) seems to lie in surface, agent, or oxygen availability effects or some combination or interaction of these effects. While the immersion and interaction effects (illustrated in Figures 8, 9, and 10) found through- out these studies are of academic interest, and of secondary interest from a practical standpoint, and since their explanation must be rather general and non-specific as the above, no further time is devoted to them in these discussions. Statistical analyses of the data showed the following character- istics: (1) The response of a given replicate from immersion-to-immersion is independent of its response in a previous immersion. (2) The response for a subsequent immersion is independent of a previous immersion; effects. (3) When a given response was obtained with a given immersion it was generally uniform over the replicates. (4) The requirement of homogeneity of variance for analyses of variance is satisfied when the distilled water data is analyzed separately from the nine alkaline detergent sol- utions’data. (5) ,The same significant differences were found for agent effects and concentration effects whether analyzed together or separately. The statistical checks described above provide a basis for the fact . that the statistical methods employed are sound for the type of data and ' responses obtained in these studies. . 147 Visual examination results indicated a better over-all condition for AI, IV than for the 3% sodium hydroxide control. The A on the other hand showed a poorer over-all condition than that of the AIII’ and A II control. Temperature effects Significantly more corrosion was observed at 170°F than at 130°F~ with each of the three solutions tested. However, in the case of 32 sod: ium hydroxide plus 0.0841 sodium gluconate (AI) significantly more corro- sion was found at 130°F than at 150°F. In addition, the corrosion weight loss difference between 130°F and 150°F, and between 150°F and 170°F. was not significant for 3% sodium hydroxide and 31 sodium hydroxide plus 0.1561 tetrasodium.ethylenediaminetetraacetate (A respectively.. Arith- b metic and Arrhenius plots of the data for each oflihe three solutions (Figures 14 and 15) did not show a significant linear relationship be- tween corrosion rate and temperature for the respective plots (corrosion rate versus absolute temperature or logarithm‘of corrosion rate versus the reciprocal of the absolute temperature). The over-all effect of temperature on corrosion is likely the combined effect of temperature on reaction rate, gas solubility, and corrosion products for each of the systems tested. It should be recalled that the results presented in Table 33 consist of the combination of two separate tests’data. The 150°F corrosion weight loss results used here are from a previous test (Table 26), while the 130°F and 170°F corrosion weight loss results are from another test conducted sometime later, but by the same individual. If our attention is centered on the 130°F and 170°F results alone we find a rather definite increase in corrosion rate with increase in temper- ature for each of the systems tested. The visual examination results indicated that in general as the temperature increased the over-all appearance and condition of the sys- tems evaluated tended to decrease, but not to a large extent. Thus, general agreement was found between visual examination and weight loss results. Surface-active agent effect Addition of a surface-active agent (Triton 08-15) that is recom- mended for use with alkaline detergent solutions, to improve upon their 148 detergency, did not produce a significant increase or decrease in the corrosivity of a sodium hydroxide-sodium gluconate composition (Table 39). Schwartz, Perry, and Berch (1958) have reviewed studies on the corrosive‘ effects of surface-active agents. They have noted, with regard to re- presentative anionic, cationic,and nonionic agents that their action seems to be specific for individual compounds rather than types. Evans (1960) also concurs with this. It is thus apparent that one should not general- ize on the basis of the above findings with regard to the potential effects of other surface-active agents on corrosion. Length of immersion effect A common practice is to assess the corrosion which has occurred during an arbitrary period of time, and to adopt the ratio of these two as a representative corrosion rate noting the total time of exposure (Champion (1952)). The purpose in conducting this study was to deter- mine the variation in corrosion weight loss with length of immersion time, since the decision to use a 7 day immersion in all the tests was somewhat arbitrary. Use of one length of exposure, however, allowed standardization and control of the unknown effects of this variable. In addition, the following served as a basis for selecting 7 days as the length of immersion: Preliminary tests seemed to indicate that a 7 day immersion would be satisfactory in producing a reasonable amount of corro- sion. It has been recommended that the alkaline wash sections of a washer be emptied, cleaned, and new solution added each operating week. Shaw and McCallion (1959) used a 7 day exposure period in their gluconate- caustic studies, and use of a 7 day exposure in these studies would allow comparison of results. This study on the effect of length of exposure on corrosion was one of the last to be conducted. The reason for this was several delays due to experimental control difficulties. The results of these tests suggest that future research be conducted using 12 hours, 1 day and 2 day exposure periods, except possibly in the case Of the sodium hydroxide-tetrasodium ethylenediaminetetraacetate solutions. The reason for little apparent difference in amount of corrosion with 1, 3, S, and 7 day immersions is probably due to a build up or accumulation of corrosion products which could effectively limit further corrosion. Advantages and disadvantages still remain, however, as to the use of a .149 1 day or 7 day immersion period. Use of a 1 day immersion period does not itself assure that even its use is best because perhaps an even shorter immersion period might still be better in this respect in-some instances. The use of a 7 day immersion period and comparisons based“ on this, allow any inhibitive or non-inhibitive effects which take.p1ace with time, and would occur in practice and have important practical im-. plications to be taken into account in assessing the comparative effects under study. The sodium hydroxide-sodium gluconate solution is signif- icantly more corrosive than the sodium hydroxide solution when the-mean for the two immersions'results are compared for either the l, 3, 5, or 7 day immersions. This is not the case with the sodium hydroxide- tetrasodium ethylenediaminetetraacetate solution, since the amount of corrosion weight loss is significantly greater for a 7 day immersion than for l, 3, and 5 day immersion periods. So it is evident that in the first case the same results are found no matter what immersion period is used for comparison of effects, while in the second case this is not so, since here a significant increase in corrosion occurs with increased length of immersion. The fact that in one of the three cases a significant increase in corrosion weight loss was found, is itself support for the use of the longer immersion_period. Studies on hard water-sodium hydroxide-sequestering_ggent systems The effects of fOur water softening agents in 31 sodium hydroxide, 12 grain per gallon hard water (consisting of 2/3 as calcium hardness and 1/3 as magnesium hardness) solutions on the corrosion of mild steel were determined. Significantly less corrosion was found with tetrasodium A111 (44)’ and tetrasodium pyrophosphate, AIv (55), thanowith the 31 sodium hydroxide control (11). The AII (3302),AIII (44), and AIv not significantly different in effect. Significantly more corrosion was found with sodium gluconate, AI (2202), than with the control. (Numbers ethylenediaminetetraacetate, AH (3302), trisodium phosphate, (55) were,themselves, given in parentheses are the codes as used in Figure 16 and they denote the use concentrations of the respective agents.) ‘ A comparison of the weight loss data (t test) of the 3% sodium hy- droxide, distilled water and hard water controls showed that the 31 sodium hydroxide, hard water control was the significantly more corrosive of the 150 two. The addition of sodium hydroxide to hard water results in the following reactions: 0a(H003)2 + zuaou-qoaco3$ + Na2003 + 2:120 If all the calcium and magnesium ions are complexed, the concentration of AIII’ and A solutions. 1’ A11" IV In the hard water-sodium hydroxide-sequestering agent systems it is like- sodium carbonate would be the same with A ly the mutual interactions of the several ions present will play a major role in affecting the nature and distribution of corrosion products, and thus, ultimately affect the corrosion rates. It is known for example that hard waters tend to precipitate an insoluble layer (e.g. 0a003) particularly on the cathodic surfaces of corroding iron (Uhlig (1948)). Such an insoluble layer may contribute to impeding the diffusion of oxygen to the metal in cases where a relatively unprotective layer of corrosion products exists (such as with soft water solutions). 0n the other hand, such a precipitate may alter an already protective film (such as the ferrous hydroxide film on iron in alkaline solutions) decreasing its over-all protectiveness. .It appears that this is the case with the 32 sodium hydroxide, hard water control. «The water hard- ness probably results in alteration of the corrosion product film in. such a manner that less resistance is afforded the diffusion of oxygen . to the metal. Addition of a water softening agent that will prevent the- deposition of an insoluble layer of water hardness compounds on the metal surface then will decrease the corrosion rate found with the 31 sodium hydroxide,hard water control. This appears to occur withAII andAIv which softens by sequestration of the alkaline earth metal cations and with A which softens by precipitation. The increase in amount of III corrosion with A may be due to a calcium- or magnesium-gluconate com- plex existing on1 the surface of the metal. The metal-gluconate com- plexes formed, normally considered as-solubilized, may exhibit some charge that will allow'migration to the cathodes. There they may modify the nature and'distribution of corrosion products and perhaps allow'more rapid oxygen diffusion to the metal surface, and is so doing.promote oorrosion. If increasing the concentration of the agent (Ai) tends to more completely.sequester the metal-glucbnate, thus forming a.more stable water soluble complex, more of this complex is then in solution and less on the metal and the corrosion rate is then expected to de- 151 crease with increasing concentration of the agent. Concentration effects of AI and AII are analyzed and discussed in more detail later in this dissertation. Shaw and HcCallion (1959) found more corrosion with 31 sodium hy- droxide and 31 sodium hydroxide plus sodium gluconate in their studies. than was found in the tests reported in this study. They also.reported. an increase in corrosion with increase in sodium gluconate concentration, which also differs from the results found in this investigation. Several factors differ between these two studies, the most important of which ~may be the use of rotating specimens in the Shaw and McCallion study.-v They also employed a test solution volume-to-test specimen surface area~ ratio of considerably less than that which is recommended. In addition, 7- only two replicates were used, the results were not analyzed statistically, and the error involved was not measured. The type of metal andcorrosion products removal procedures employed by Shaw and McCallion differed from those used in this study. The corrosion products removal procedure employed in their studies was found unsatisfactory for use with milde, steel in these studies. The claim (000per (1960)) that the use of sod- ium gluconate results in increasedcorrosion has been substantiatedwand the claim (DVorkovitz and Hawley (19528,b)) of decreased corrosion with its use has been repudiated in these laboratory tests. However, it should‘ be noted that as far as the practical application of these findings, the corrosion aspect due to the use of this agent may be out-weighed by other advantages that may be obtained as a result of its incorporation in cleaning compositions. The visual examination results indicated a better over-all condition AII andAIv systems than for the AIII and the 31 sodium hy- droxide control systems (Table 51). The over-all condition of the con- trol, A A I’ and.A I’ II IV . that for the same systems made with distilled water. However, with the for the AI, systems made with hard water were poorer than AII system its over-all condition tended to be poorer with the distilled water solutions than with the hard water solutions. On an Over-all basis, more corrosion was found with the 3% sodium hydroxide hard water systems than with the 31 sodium hydroxide distilled water systems, as indicated by both the grand agent over-all mean weight loss and the 152 grand agent over-all visual examination results. Distilled water and hard water‘studiesfcombingd and compared- Distilled water versus hard water 7 Results of a test conducted to determine the corrosivity of distilled water and hard water (12 grains per gallon total hardness as calcium car- bonate) using the experimental procedures developed for these studies have shown that distilled water is considerably more corrosive (significant, 11 level) than hard water. This finding in itself is not new. However, having corrosion weight loss data for these solutions, determined under; the conditions and procedures used throughout the rest of the studiegyis advantageous in that more confidence may be attached to comparisons made.~ using these values than to literature values, where experimental conditions may have differed to some extent from those employed here.. The corrosion rate found with distilled water of 3.5 HPY (or 135.19 mg/dmz-week) com- pares with literature values of 3.1 HPY to 3.5 MPY (Speller (1951) and Uhlig (1948)). This finding itself is important since it reflects favorably upon the experimental methods employed throughout these studies. The difference in corrosivity of the two solutions can be ex- plained as follows: As corrosion proceeds, ferrous hydroxide is formed on the surface of the metal, and due to this and the accompanying in-. crease in pH, calcium carbonate is formed from the calcium bicarbonate in hard water and precipitated on the surface of the metal intermingled with corrosion‘products. This results in the formation of a less perme- able film over the surface of the metal which retards oxygen diffusion. The over-all corrosion rate is decreased in hard water as a result of this retardation of oxygen diffusion to the metal surface. Visual ex- amination results indicated that the appearance of the specimens were slightly poorer in distilled water than in the hard water, while the condition of the distilled water itself was considerably poorer than that of the hard water. The over-all appearance of the two systems was bad with severe corrosion. The visual examination and weight loss data both indicated more corrosion with distilled water than hard water. Distilled water, distilled water plus tetrasodium ethylenediamine- tetraacetate,gand hard water corrosion weight loss effects compared Comparison of the corrosivity of distilled water plus 0.1562 tetra- sodium ethylenediaminetetraacetate with distilled water and hard water 153 has shown that the former is significantly more corrosive than distilled water and hard water. These results have been explained in a previous section (see the distilled water-sequestering agent section). There are important practical implications that can be drawn from the above find- ings: (1) The use of tetrasodium ethylenediaminetetraacetate with soft water in the rinse sections of mechanical bottle washing machines could significantly increase corrosion over that normally obtained with soft water alone. (2) An excess of this sequestering agent in washer rinse. sections with hard water could also be expected to increase corrosion. Based on the results in Table 22 phosphates could be used to ad-. vantage in corrosion inhibition in the rinse sections of a washer. An excess of phosphate type agent would probably be required in order to both soften the water and to inhibit corrosion. The objections to the use of phosphates which occur when they are used with alkaline solutions at high temperatures are not a great problem in the rinse sections. In the rinse sections of a washer the pH is near neutral and the phosphates are very good sequestrants at this pH. In addition, one of the chief objections to the use of polyphosphates, namely that of reversion to the orthophosphate, which occurs at high temperatures, would not be a serious problem in the rinse sections, since the rinse water is at about room temperature. Sodium gluconate could be used to retard corrosion in rinse sections, but it is not an effective sequestering agent for calcium and magnesium at the pH 7.0 of the rinse sections. Alkaline detergent, distilled water and hard water studies combined and compared Caution: The results reported and discussed in this section are pre- sented to indicate trends, and they mmst be considered as approximate, since results for the two sets of data were pooled even though homo- geneity of error variances did not exist for the two sets of data. Anal- ysis of the results (Table 59) in this manner despite the disadvantage noted above allows many comparisons of results that would otherwise not be possible, as shown in Table 60. The marked differences which exist between sodium gluconate in alkaline hard water and distilled water sol- utions as well as the marked difference between alkaline hard water and distilled water controls are readily apparent. The differences between 154 the alkaline, hard water and alkaline, distilled water trisodium phos- phate and tetrasodium pyrophosphate solutions and the tetrasodium-ethylene- diamdnetetraacetate solutions (except at its highest concentration) were found not significant. The distilled water used in preparing the test solutions made with distilled water was boiled before preparation of the solutions; the hard water was not boiled,since boiling would have induced precipitation of hardness salts. This point in the experimental procedures is important because the oxygen content of the hard water solutions could be expect- ed to.be greater than that of the distilled water solutions. The oxygen content in the distilled water solutions at the start of the test was primarily dependent upon the amount of oxygen dissolved from head- space gases during pre-heating of the solutions before adding the test specimens. Concentration effects of sequestering agents in alkaline distilled water and hard water solutions The corrosion weight loss of steel specimens in alkaline distilled water solutions containing 0.168% sodium gluconate was significantly greater than that in 0.0422 and 0.0841, and between 0.084% and 0.1682 sodium gluconate. A plot of corrosion weight loss versus concentration of sodium gluconate gives a straight line of the form ’3‘? = 3.065 + 2.6011: over the concentration range evaluated. The 9’18 an estimate of an .unknown Y (corrosion weight.1°88) from a known X (Concentration in per- cent). The slope of the line (be,called the linear regression coefficient of X on X) and the intercept (a) values being 2.601 and 3.065 respectively, over the concentration range evaluated. Apparently the corrosion weight loss does not follow a linear relationship somewhere between 0.000% and 0. 0421 sodium gluconate, since the corrosion weight loss found experimen- tally at 0. 000% sodium gluconate was 2. 74.mg/dm.2-week. The tendency for corrosion weight loss to increase linearly over the concentration range evaluated (be significant at 201 level) can be explained as follows: As the concentration of sequestering agent increases, the potential for more sequestration of corrosion products exists, and as more of the corrosion products which tend to inhibit corrosion in oxygen 155 type corrosion (corrosion products tend to control oxygen diffusion to the cathodes which tends to control corrosionbrate) are sequestered, the corrosion rate tends to increase. From a practical standpoint these re- sults would tend to indicate that very heavy excesses of sodium gluconate should be avoided with 3% sodium hydroxide soft water cleaning composit- ions used for purposes other than softening water (such as detergency). In alkaline hard water solutions the amount of corrosion decreased as the concentration of sodium gluconate increased. The three possible differences in weight loss as a function of concentration were each significant. A plot of the logarithm of corrosion weight loss versus the reciprocal of sodium gluconate concentration gives a straight line of the form » ' 1 log 1" = 0.7782 + 9%{9-43-2- over the concentration range evaluated. The slope (be, significant-at the 12 level) and intercept (a) values being 0.006442 and 0.7782 respect- ively. It was anticipated that either one or two mole concentrations of sequestering agent (0.042% or 0.084% sodium,gluconate) per mole of alkaline earth metal cations (12 grains per gallon total hardness as- calcium carbonate) would result in complete sequestration of the metal cations. Theoretically at leasg.sequestrant-to-metal mole ratios-of .. less than one, one, and two have been described for sodium gluconate and calcium.(0haberek and Hartell (1959)). In practice.it is assumed that a mole-to-mole ratio exists and this quantity is then doubled so that commercially 2 moles of sodium gluconate per mole of alkaline earth metal ion are recommended (Shaw and McCallion (1959)). If all the calcium and magnesium ions of the calcium and magnesium carbonates which interfere with the corrosion producté (ferrous hydroxide) normal metal protection were completely sequestered by just sufficient sodium gluconate (such as 0.042%), and thus prevented from becoming intermingled with the corrosion products, a corrosion rate less than that obtained with the 3% sodium hydroxide hard water control (4.13 mg/dmzdweek) and perhaps nearer the 3% sodium hydroxide distilled water control (2.74 mg/dmz-week) would be expected, but apparently this is not the case, since the test using the mole-to-mole ratio of one (0.0421 sodium gluconate) did not produce such an effect. Instead, a very marked increase in corrosion 156 was found. There are at least two approaches to explaining this marked effect, the latter of which seems most reasonable since it can be used more readily to explain the decrease in corrosion with increase in sod- ium gluconate.concentration as well as the marked increase at the lower concentration. The first approach is to look upon the calcium-gluconate complex formed as an accelerator of corrosion (by a catalytic means for example) where increasing the concentration of the free gluconate from 1 mole (0.0421) to 2 moles (0.0841) to 4 moles (0.168%) tends to increasingly retard the accelerating effect of the calcium-gluconate complex, which is in proportion decreasing in quantity relative to the free gluconate in solution. The second approach to the explanation of the two main effects observed follows: The first of these effects to be explained is the marked increase in corrosion between the control (32 sodium hydroxide, hard water solution) and the lowest sodium gluconate concentration used (where the mole-to-mole ratio was one or 0.042% sodium gluconate). The second effect to be explained is the decrease in corrosion with increase in sodium gluconate concentration from 1 mole to 2 moles to 4 moles of sequestrant per mole of alkaline earth metal cations. The increase in corrosion found between the control and the lower gluconate concentration used may be due to the calcium-gluconate complex being intermingled with the corrosion products and affecting their nature and distribution in such a manner as to ultimately result in increased corrosion. One way this might occur is through incomplete sequestration of the complex or incomplete solubilization and maintaining of the complex in solution. Since at this point we are concerned with a mole-to-mole ratio of one, where each gluconate has a negative charge of one and each metal cation has a positive charge of two, if all the cations and anions combine one-to-one the possibility exists that the calcium- or magnesium- gluconate complex as formed may have a net positive charge, and as such, may act as a.molecular-complex cation and be attracted to the cathodes. At the cathodes this complex could modify the nature and distribution of corrosion products in such a manner as to increase the permeability to diffusion of oxygen. If oxygen diffusion is accelerated as a result, then the corrosion rate would be expected to increase to perhaps the 157 extent found. The second effect, namely that of the decreased corrosion. with increased sodium gluconate concentration may be due to the formation of a more soluble complex. At the 0.084% concentration there exists two gluconate anions (each having one negative charge) per alkaline earth metal ion (each having a positive charge of two). For this reason there should be many less complexes of some possible net positive charge than in the previous case (0.0421 sodium gluconate). The 16% decrease in corrosion from the 1 mole concentration (0.0422) to the 2 mole con- centration of gluconate per mole of metal cation, as compared to about a 81 decrease in corrosion from the 2 mole to 4 mole concentration- (0.168%), suggests that the sequestrant-to-metal ion ratio may be pre- dominately 2-to-l. The additional decrease in corrosion between the , 2 and 4 mole concentration levels could be the result of further improved sequestration. The stability of the 2 mole gluconate-to-l mole divalent metal cation complexes is probably such that the complexes are continually breaking down and recombining; as the concentration of gluconate is in- creased from 2 to 4 moles, the chances of recombining are increased, so that at any given time, more stable, water soluble, uncharged complexes are present in solution and free from the metal and its corrosion prod- ucts. In this manner, as the gluconate concentration increases over..- the range tested, less calcium-gluconate will be present in the corrosion products and the corrosion products will become a more homogeneous, less permeable oxygen diffusion film and thus corrosion will be retarded. In the case of some calcium-phosphate complexes their role in corrosion has been explained in a manner somewhat similar to the above. Putilova, Balezin, and Barannik (1960) have reported that the glassy phosphates migrate towards the cathodes in water containing calcium ions and that this differs sharply from the behavior of sodium phosphate dis- solved in distilled water where the phosphate ions move towards the anodes. They state that when a calcium salt is present, cathodic migration can take place if positively charged colloidal particles consisting of numerous diphosphate chains joined by calcium atoms are formed. They also point out that the positive charge on the colloidal particles may be due to adsorption of calcium ions by the colloidal calcium phosphate. Tetrasodium ethylenediaminetetraacetate in alkaline distilled water 158 solutions produced a significant increase in corrosion as the concentration 'was increased from 1 mole (0.078%) to 4 moles (0.3121Q.of sequesterant per mole of alkaline earth metal cations. The differences in amount of corro- sion between 1 mole (0.0781) and 2 moles (0.156%), and between 2 moles (0.0781) and 4 moles (0.312%) of sequestrant per mole of metal cations were not significant. A plot of the results gives an equation of the form ’3‘? = 3.057 + 1.7351: where 3.057 is the intercept (a) and 1.735 is the slope or linear re- gression coefficient (be) over the concentration range studied.- The linear regression coefficient is significant at the 401 level.. The.“ tendency toward some increase in corrosion with increase in concentration that manifests itself in the alkaline distilled water solutions, but not in the alkaline hard water solutions,appears to be due to the same causes as the increase noted with gluconate in alkaline distilled water solutions, despite the fact that tetrasodium ethylenediaminetetraacetate supposedly is not effective in ferrous and ferric ion sequestration above a pH of 12 ("Keys to ChelatiOn" (1959)) which existed in these solutions. Many instances of corrosion acceleration or retardation due to the use of a variety of sequestering agents under a variety of conditions have been found in these laboratory investigations. A number were un- expected and some were of academic interest. However,from the practical point of view the ultimate decision is whether to use a certain sequester- ing agent or not under these conditions. This decision can be based upon the laboratory information reported and discussed here, keeping in mind that these were laboratory and not field investigations, and that other advantages (e.g. detergency) to be gained from its incorporation may off- set certain corrosion disadvantages. Economically these advantages and disadvantages will need to be weighed against each other in order to facilitate this decision. SUMMARY AND CONCLUSIONS Significant acceleration or retardation of the corrosion of mild-g steel was found with sequestering agents used under a variety of labora- tory test conditions similar to those found in practice. The laboratory methods developed specifically for these investigations were responsible for the reliability and reproducibility of results obtained. The per- cent error of over-all means was generally less than plus or minus 7 per- cent and a difference of about 10 percent between over-all means was generally significant for samples of 20 replications. Values obtained with the test controls agreed with literature values, and the experi- mental procedure used was found satisfactory for use by individuals of differing technical training or experience. The corrosion weight loss results were not significantly different for cold rolled and hot rolled AISI No. C 1008 steel specimens immersed in distilled water solutions of alkali, alkali plus gluconate, and alkali plus ethylenediaminetetra- acetic acid. Cold rolled steel showed significantly more corrosion in distilled water than hot rolled steel. Distilled water_plus sequestering agents: The corrosivity of dis- tilled water was reduced 97.6 percent with either trisodium phosphate or tetrasodium pyrophosphate (inorganic phosphates) and about 75 percent with sodium gluconate (hydroxycarboxylate). Tetrasodium ethylenediamine- tetraacetate (aminopolycarboxylate) caused a 12 percent increase in corrosivity over that found with distilled water. Distilled water plus sodium hydroxide plus sequestering agents: Three percent sodium hydroxide solutions of tetrasodium ethylenediamine- tetraacetate, sodium.gluconate, and tetrasodiumypyrophosphate listed in order of decreasing corrosivity were each found significantly more corro- sive than the control (3 percent sodium hydroxide). Tetrasodium ethylene- diaminetetraacetate solution was also more corrosive than the trisodium phosphate solution. The differences in amount of corrosion between the individual agents in group one (tetrasodium ethylenediaminetetraacetate, sodium gluconate, and tetrasodium pyrophosphate), group two (sodium glu- _ 159 160 conate, tetrasodium pyrophosphate, and trisodium phosphate), and group three (trisodium phosphate and control) were not significant. Distilled water plus trisodium phosphate or tetrasodium pyrophosphate were not significantly more corrosive than sodium hydroxide plus tetrasodium pyro- phosphate. Significantly more corrosion was found at 170°F than at 1309F with sodium hydroxide, sodium hydroxide plus sodium gluconate and sodium hydroxide plus tetrasodium ethylenediaminetetraacetate solutions. Corro- sivity of sodium hydroxide plus sodium gluconate solutions was not signif- icantly altered by the addition of an amphoteric surface-active agent of the oxyethylated sodium salt type, containing both anionic and cationic centers. Amounts of corrosion found with 3, 5, and 7 day immersion periods (length of specimen exposure) were not significantly greater than that found with the 1 day immersion period of specimens in sodium hydroxide or sodium hydroxide plus sodium gluconate solutions. Significantly more corrosion was found with the 7 day immersion ofIapecimens in sodium hy- droxide plus tetrasodium ethylenediaminetetraacetate in their first immer- sion than with the l, 3, and 5 day immersions, which were not significantly different. Hard water plus sodium hydroxide plus seguesterigg agents: The sodium gluconate solution was much more corrosive than the control, which was significantly more corrosive than tetrasodium pyrophosphate, tetrasodium ethylenediaminetetraacetate, and trisodium phosphate which were themselves not significantly different in corrosivity. ' Distilled water and hard water studies combined and compared: Signif- icantly more corrosion was found with distilled water plus tetrasodium ethylenediaminetetraacetate than with distilled water or hard water (12 grains per gallon total hardness as calcium carbonate). Distilled water was significantly more corrosive than hard water. Comparison of the corrosivities of alkaline detergent, distilled water and hard water sol- utions showed that sodium hydroxide and sodium hydroxide plus sodium gluconate hard water solutions are markedly more corrosive than their distilled water counterparts. Differences in the amount of corrosion found with distilled water and hard water soltuions of trisodium phos- phate, tetrasodium pyrophosphate, and tetrasodium ethylenediaminetetra- acetate were not significant. Sequestering agent concentration effects 161 were assessed for sodium gluconate and tetrasodium ethylenediaminetetra- acetate in sodium hydroxide, distilled water and hard water solutions. Corrosion was found to increase in the distilled water solutions and to decrease in the hard water solutions as the concentration of sodium glu- conate increased from 1 mole to 4 moles of sequesterant per mole of alka- line earth metal cations. The predicting equat nus for sodium gluconate concentration effects on corrosion weight loss with sodium hydroxide, dis- tilled water and hard water solutions over the concentration range eval- uated were determined to be '3‘? = 3.065 + 2.601X and 0.00gggg X respectively. Significantly more corrosion was found at the 4 mole con- 10g 'i" -.-. 0.7782 + centration of tetrasodium ethylenediaminetetraacetate than at the 1 mole concentration in the sodium hydroxide, distilled water solutions. Corro- sion was not significantly increased or decreased with increase in tetra- sodium ethylenediaminetetraacetate concentratiOn in the sodium hydroxide, hard water solutions. The apparent discrepancy between the effects of sodium gluconate con- centration on corrosion in these distilled water and hard water solutions may be explained as follows: In the first case gluconate anions are free to sequester ferrous ions of the ferrous hydroxide protective film of corrosion products, and as the concentration of gluconate is increased more ferrous ions can be sequestered. As the'film is diminished in such a manner, less resistance is afforded to the diffusion of oxygen to the cathodes, andicorrosion is thereby promoted. In the second case alka- line earth metal cations are present in solution and the gluconate anions would normally be expected to sequester them, thus forming a stable water soluble complex. However, it is suggested that the complex is not adequately solubilized at the lower concentration and that inter- mingling of this complex with the corrosion products decreases their pro- tectivity and results in increased corrosion. Addition of more gluconate enhances the solubility of the complex leaving less to alter the protect- ive effect of the corrosion products, resulting in less corrosion with increasing gluconate concentration. 10. ll. 12. 13. 14. 15. 16. REFERENCES Corrosion Studies Related References Aiken, J. K., and C. Garnett. 1957. Chelating agents in metal cleaning and de-rusting. Electroplatigg and Metal Finishin , lg, 31. . Bablik, Heinz and H. Belohlavy. 1957. Modern detergents and hot galvanization. (in German) Werkstoffe‘g. Korrosion, Q, 742. Bacon, L. R. and E. G. Nutting. 1952. Polyphosphate detergents in mechanical dishwashing: Solubilizing action of polyphosphates on certain metals. Ind. Eng. Chem., 44, 146. Bukowiecki, A. 1959. Investigations of the corrosive effect of washing media. (in German) Schweiz.yArcht.angewrf isrfigf 2529' g 24,.355. Chaberek, Stanley and A. E. Martell. 1959. Organic sequestering agents. John.Wiley & Sons., New York, p. 616. . . . . Clements, P. G. and T. Kennedy. 1952. The use of alkali in clean- ing processes. lag. 0hem.,‘2§, 152, 301. . . COOper, William. 1960. 1Private communication. Geo. J. Meyer Manufacturing Co., Cudahy, Wisconsin. Dixon W.~Js and F. J. Massey, Jr. 1957. Introduction to statis- tical analysis. .McGraw-Hill 00., Inc., New York, p. 488. Duncan, D. B. 1955. Mpltiple range and multiple F test. Biometrics, .13.. 1. Dvorkovitz, Vladimir and T. G. Hawley, Jr. 1952‘. Washing composi- tion. U. 8. Patent 2,5843017r. (to Diversey Corp.) Jan. 29. Dvorkovitz, Vladimir and T. G. Hawley, Jr. 1952 . Washing composi- tion. 0. 8. Patent 2,615,846; (to Diversey Corp.) Oct. 28. Gegner, P. J. 1960. Corrosion in alkaline environments. Proc. Short Course on Process Ind. Corrosion. p31. NACE Houston, Texas. Getty, Raymond, N. H. McCready, and William“ S'tericker. 1955. Silicates as corrosion inhibitors in synthetic detergent mixtures. ‘59. Soc. Testigg Materials Bull. No. 205, TP 76, p.50. Haller, H. S., F. M. Grant, and C. J. Babcock. 1941. The corrosive effect of chlorine and lye solutions on metals used in dairy equipment. .2; S. Dept. é§5' Tech. Bull. No. 756, p.27. Harter, H. L. 1960. Critical values for Duncan's new multiple range test. .Biometrics, 16, 671. Hunziker, 0. F., W. A. Cordes, and B. H. Nissan. 1929. Metals in dairy equipment: Corrosion caused by washing powders, chemical sterilizers, and refrigerating brines. 'g. Dairy Sci.,'12, 252. 162 17. 18. 19. 20. 21.‘ 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 163 Jelinek, R. V.‘ 1958, Corrosion refresher on cause and cure. Chem. as. 9.6.. 114. -- I Jensen, J. M. and G. A. Claybaugh. 1951. Effectiveness of modern detergents in can washing. 44th Ann. Conv. Proc. Milk Ind. Foundation, Detroit, Plant Sect., Vol. 2, 36-42. Detroit, Michigan. Keys to Chelation. 1959. Published by The Dow Chemical Company, Midland, Michigan, p. 26. . . Levina, S. L. 1956. Effect of inhibitors on the dissolution of iron in alkaline solutions. g, Appl. Chem.‘y.‘§.'§.‘§. (Engl. transl.) 22, 1457. Lewandowaki, Thaddeus. .1952. Factors involved in corrosion.of tinned and stainless steel by mild acidic solutions used in cleaning dairy equipment. .g. Dairy Sci., 22, 449. McCune, H. W. 1958. Corrosion of aluminum by alkaline sequestering solutions. lpg.'§pg. Chem., 29, 67. . . . - Mehltretter, 0. L., B. H. Alexander, and C. E. Hist. 1953. Sequest- ration by sugar acids. Ind. Eng. Chem., 42, 2782. .. Mehltretter, C. L. and P. R. Watson. 1959. New low cost sequest- rants. Soap Chem. Specialties, 2;, 8. Moore, E. W. and F. E. Smith. 1942. Effect of sodium hexameta- phosphate on the solution of lead. J, 59. Water Works Assoc., .24, 1415. Nieland, W. L., J. J. Maguire, 0. B. George, and H. L. Kahler. 19505' .Corrosion and tuberculation inhibition in water systems. U. 8. Patent 2,529,177. (to W. H. and L. D. Betz) Nov. 7. Nieland,.W..L., J. J. Maguire, 0. B. George, and H. L. Kahler. 1950 . Methods for obtaining corrosion and tuberculation inhibi- tion in water systems. U. S. Patent 2,529,178.. (to W. H. and L. D. Bets) Nov. 7. Parker, M. E. 1942. Corrosion tests on acid cleansers used in dairy sanitation. g, Milk Techno1.,‘§, 37. Putilova, I. N., S. A. Balezin, and V. P. Barannik. 1960. Metallic corrosion inhibitors. (Translated from the Russian by G. Ryback and edited by E. Bishop.) Pergamon Press, Ltd, p. 238. London. Schmitt, R. J. 1960. Use of iron, carbon steel, and alloy steel, in the chemical industry. Proc. Short Course on Process Ind. Corrosion, p. 31. Nat. Assoc. Corr. Eng., Houston, Texas. Schwartz, A. M., J. W. Perry,.and Julian Berch. 31958; ‘Surfiace active agents and~detergentsa.Interscience.Publishers, Inc., New York, p. 839. Shaw, J. K. and J. J. McCallion. 1959. Gluconate-caustic systems, corrosion and other considerations. Proc. Sixth Ann. Meeting Soc. Soft Drink Techno1., Washington D. C. Skaggs, S. R. and D. D. Miller. 1958. ”The influence of hard water on dairy detergents; New.Mexico Agr. Exp. Sta. Bull. No. 427, . p. 28. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. . 164 Smith, R. L. 1959. The sequestration of metals. Chapman &'Hall Ltd., London, p. 251. Snedecor, G. W. 1956. Statistical methods. 5th ed., The Iowa State College Press, Ames, Iowa, p. 534. ' Speller, F. N. 1951. Corrosion causes and prevention. 3rd ed., McGraw-Hill Book Co.) New York, p. 721. - Staff Feature. 1961. Corrosion damage on automobile exhaust sys- tems. Corrosion, ll, 18. . . - , Steel, R. G. D. and J. H. Torrie. 1960. Principles and'prOcedures of statistics. MeGraw-Hill.Book Co., Inc., New York, p. 481. Stupel, H. and F. Rock. 1959. Corrosion by washing compounds,“ with special references to "Syndets". (In German) Seifen-Ole- Fette-Wachse, 85, 311. . . ~ . Sundararajan, J. and T. L. Rama Char. 1961.} Inhibition of corro- sion of commercial aluminum in alkaline solutions. Corrosion, $1, 112. Technical Service Reports for the Bottled Carbonated Beverage Indé ustry. 1956. Information Bulletin on Bottle Washing. American Bottlers of Carbonated Beverages, Washington 6, D. 0. ~ Uhlig, H. H. 1949. -The cost of corrosion to the U. S. Chem.‘§pg. News,‘gz, 2764. , . de Vleeschauwer, A., H. Hendrickx, and J. Moulaert. 1951. Studie Over Enkele Eigenschappen Van Reinigingsmiddelen. Medel. Landbouwhogeschool Opezoekings stas. Staat Gent,‘l§, 76. de Vleeschauwer, A., H. Hendrickx, and E. Wallez. 1953. Corrosion effect of detergents. (in German) Medel. Landbouwhogeschool Opezoekings stas. Staat Gent,‘1§, 544. Walker, H. M. and Joseph Lev. 1953. Statistical inference. Henry Holt and Co., New York, p. 510. White, J. 0. 1962. Washing Bottler: A Challenge. .52. Milk'gpy., a, 310 Wright, W. A. 1956. Cleansers for aluminum dairy equipment. Dairy .EEK':.Z§: 167' Experimental Procedure Related References- Am. Soc. Testing Materials, Standards. 1952. Tentative method of total immersion corrosion test of non-ferrous metals. B l85-43T. Part II. American Society for Testing Materials, Philadelphia, p. 1327. Am. Soc. Testing Materials, Standards. 1958. Standard method of total immersion corrosion test for soak tank metal cleaners. D 1280-57. Part X. American Society for Testing Materials, Philadelphia, p. 1498. 50. 57. S8. 59. 60. 61. 62. 63. 64. 67. 165 Borgmann, C. W.‘ 1948. General discussion of laboratory corrosion testing. A paper, pp. 953-955, in H. H. Uhlig, ed. The corro- sion handbook. John Wiley and Sons, Inc., New York, p. 1188. Calcott, W. 8., and J. C. Whetzel. 1923. Laboratory corrosion tests. Trans.‘§g. Inst. Chem. Eng., 12, 16. Calcott, W. S., J. C. Whetzel and H. F. Whittaker. -1923. Corro- sion tests and materials of construction for chemical engineering apparatus. 0. VanNostrand Co., New York, p. 391. Champion, F. A. 1952. Corrosion testing procedures. John Wiley and Sons, Inc., New York, p. 355. . .« Corrosion. 1956. Published by the International Nickel Company, Inc., New York, p. 51. . Darrin, M. 1946. Corrosion criteria--their visual evaluation. Am. Soc. Testing Materials Bull. No. 138, p. 37, No. 142, p. 48. Evans, U. R. 1960. The corrosion and oxidation of metals. Butler & Tanner Ltd., London, p. 1094.; ’ Fontana, M. G. ‘1960. Corrosion testing. Proc. Short Course on Process Ind. Corrosion. p. 319. NACE, Houston, Texas. Hoar, T. P. 1948. Action of organic inhibitors in the acid attack of mild steel. Pittsburgh Internet. Conf. on Surface Reactions. p. 127. Carnegie Inst. of Tech., Pittsburgh, Pennsylvania. Hume8,'C. H., R. F. Passano, and A. Hayes. 1930. A study of the error of averages and its application to corrosion tests. ‘52. Soc. Testing Materials, Proc., Vol. 30, Pt. II, p. 448. Jelinek, R. V. 1959. Correlate corrosion testing methods. Chem. EB}, fiy'loss ‘ Jelinek, R. V. 1958. How oxidation corrosion occurs. Chem.'§gg., ‘62,.125’ - .. .. a Johnson, C. G. 1950. Metallurgy. American Technical Society, Chicago, p. 418. Knapp, B. B. 1948. Preparation and cleaning of specimens. A paper, pp. 1077-1083, in H. H. Uhlig, ed. The corrosion handbook. John' Wiley and Sons, Inc., New York, p. 1188. Linden, J. J. 1960. Private communication. Amchem Products, Inc., Ambler, Pennsylvania. ' McCallion, J. J. 1959. Private communication. Chas. Pfizer & Co., InC., Brooklyn, Newiflbgk4-. McCollam, C. H. and_D. L. Warrick. '1948. The pickling of iron and steel products. A paper, PP: 725-729, in Taylor Lyman, ed. Metals handbook. The American Society For Metals, Cleveland, p. 1444. Meyer, W. R. 1948. Solvent and vapor cleaning of metals. A paper, pp. 299-300, in Taylor Lyman, ed. Metals handbook. The American Society For Metals, Cleveland, p. 1444. ‘ m 68. 69. 70. 71. 166 Speller, F. N. '1935. Corrosion causes and prevention. McGraw-Hill Book Co., Inc., New York, p. 694. Uhlig, H. H. 1948. Iron and steel. A paper, pp. 125-143, in H. H. Uhlig, ed. The corrosion handbook. John Wiley and Sons, Inc., New York, p. 1188. a . . Wesley, W. A. 1948. Total immersion tests. A paper, pp. 959-965, in H. H. Uhlig, ed. The Corrosion handbook. John Wiley and Sons, Inc., New York, p. 1188. . . Wesley, W. A. 1953. The total-immersion corrosion test. Ag, Soc. Testing Materials, Proc., Vol. 43, p. 649. u I “il’umflnmmmungrinrgignmgmm“