<<

MECHANISMS OP CORROSION OP

310 STAINLESS BY

VANADIUM OIL ASHES

A DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

'P' Martin Henry LeipoId, B.S., M.S*

The Ohio State University 1958

Approved by:

Adviser DejoArtment of Engineering Acknowledgments

The research reported in this dissertation was con­ ducted under a research contract of The Ohio State Univer­ sity Research Foundation with the Department of the Navy,

Bureau of Ships* The contract was designated No. NObs 72X57,

Index No . NS 013-120.

The author wishes to thank his adviser, Dr. J. 0.

Everhart, for his assistance in the preparation of this dissertation and for his encouragement and confidence dur­ ing the course of the investigation. He also wishes to thank Drs. Thomas S. Shevlin and Wilfred R. Foster for their assistance, instruction, criticism, and encourage­ ment during the course of the research. To David W. Jones and George S. Sikora, the author wishes to extend his appreciation for technical assistance.

ii CONTENTS

INTRODUCTION 1 Background 4

Mode of Investigation 8

MATERIALS 1 1

EQUIPMENT 15 PROCEDURE 21 RESULTS 25 DISCUSSION 32 CONCLUSIONS 75 REFERENCES 77 APPENDIX 79 AUTOBIOGRAPHY 80

ill List of Tables

Table Page

1 Materials H 2 XHD Sectioning Data, 1600° P 26 3 XRD Sectioning Data, 1200° P 27 4 XHD Sectioning Data, 1600° P 27 5 Lattice Parameter for Spinel 28 6 Thermogravimetric Analysis Buns 29 7 Chemical Analysis of 310 Stainless 33 Steel and

8 Oxide Ratio for Corrosion Products 34 9 Statistical Analysis of Standard Run 53

iv~ List of Figures Figure Page 1 Automatio Weight Change Recording 17 Equipment 2 Automatic Weight Change Furnace 19 3 X-ray Diffraction Data 37 4 Possible Oxidation Rates for 42 5 Parabolic Oxidation Rates for Metal 43

6 Theoretical Weight Gain for Chemical 44 Reaction 7 Weight Gain for NVg on 310 Stainless 47 Steel at 1600° F

8 Weight Gain for at 1600° F 48 9 Weigat Gain for Nickel at 1600° and 49 1700° F 10 Iron Oxidation Rate Constant 50 11 Nickel Oxidation Rate Constant 51 12 Oxidation Rate Constant 52 13 Weight Gain for 310 55 at 1600° F 14 Weight Gain for NVg on 310 Stainless 56 Steel at Various 15 Weight Gain for NVg on 310 Stainless 57 Steel at Various Temperatures

16 310 Stainless Steel Rate Constant 58 17 Weight Gain for NVg on 310 Stainless 60 Steel with Oxide Additions

18 Weight Gain for NVg on 310 Stainless 61 Steel with Ash Additions

v Weight Grain for NVg on 33.0 Stainless Steel After Stripping Weight Gain fpr Iron Submerged in NVg Weight Gain for 310 Stainless Steel Submerged in NVg Weight Gain for NVg on Stainless Steel, Pinal Nate Weight Gain for NVg on at 1600° P INTBDDUCTION

The problem of corrosion of and refractories at high temperatures in the presence of various accelerat­ ing agdnts is frequently found in today*a advanced tech­ nology. One such problem which has long vexed combustion equipment designers and operators is that of deterioration of components in contact with the products of combustion in various types of oil-burning power equipment. Non-gaseous products collect in turbines and heat exchangers, causing severe material attack. The concentration of potential ash formers is particularly high since the types of oil burned in heavy power equipment are the residual fractions from which many of the higher hydrocarbon fractions have been removed. This results in marked concentration of impurities.

The advent of the Venezuelan crude oils brought a new order of magnitude to the severity of this problem. Com­ ponents were so severely attacked that failure often oc­ curred in a matter of weeks.^ It was found that not only

^Evans, C. T., "Oil Ash Corrosion of Materials at Elevated Temperatures ASTM Special Tech. Pub. 108, 1950. was the total inorganic Impurity content high®3? in these Venezualen oils, but the concentration of certain trouble­ some elements, in particular, , was high. 2

Investigations of the problem have long associated the corrosion with the presence of sodium, , and vanadium in the oils* However, the attack had been treated as a gross result of the action of all three# Recently, the compatabilities existing within the of these elements o and, hence, the specific corrodents, were determined* The

^Foster, W. R., Leipold, M, H* and Shevlin, T* S#, tfA Simple Phase Equilibrium. Approach to the Problem of Residual Oil- Ash Corrosion ,” Corrosion, 12 (1956). knowledge is, of course, necessary for any study of corrosive mechanisms, since attempts to study a system in which sev­ eral corrodents and their associated meohanisms were active would be almost impossible#

The problem of corrosion has not been neglected in the research laboratory, particularly by the power equipment manufacturers. However, because of the presenoe of several complex systems, e.g., ash composition, base material com­ position, operating conditions, the problem had not yielded. Further, many of the investigations were quite empirical In approach and with the large number of parameters available, little correlation among approaches could be found.

The existence of so many parameters naturally involves selection of optimum conditions among the existing variables on a practical basis. For example, the problem could be solved by expensive processes which would remove the 3 troublesome elements from the oil* This is not feasible on an economic basis* Consequently, to achieve the desired selection of optimum conditions, it is necessary to have an understanding of the mechanisms existing so that the effects of a change in conditions may be predictable*

The complexity of the problem which makes empirical investigation les3 fruitful indeed complicates theoretical analysis. A direct approach based on reported data for oxidation of metals is difficult since the applicability of simple oxidation studies to these employing a corrodent is questionable. It Is necessary first to determine the effect of specific corrodents on the simpler materials and extend these results with further investigations to the more com­ plex systems, such as stainless steel. Until the recent determination of the compatibilities existing within the ash system, selection of a specific corrodent would have presented a similar wide choice of compositions and unknown phases. However, it has been determined that the specific composition in the Na2 0 -V2(>5 system which has the greatest influence in the corrosive­ ness is Na2 0 *V2 0 4 *5 V2 0 5. This compound for which data are given in the appendix has been chosen as the specific corrodent for use in these investigations. For the purposes of simplicity, it Is referred to throughout this disserta­ tion as NVg. 4 Literature Survey The problem of corrosion of stainless steel by vanadium compounds at high temperatures is contiguous to several areas in investigation that are fraught with considerable diversity of opinion* They include the chemistry and com­ patibility of vanadium compounds and the mechanisms of oxi­ dation* A completely satisfactory mechanism for the unaided oxidation of stainless steel has not been obtained, much less for corrosion accelerated by various corrodents,

v2°5* Mo(>3 * Obviously, all of these investigations are related to the specific problem on hand, but many only in a very general way. Consequently, effort will be made to cite only those investigations which are more closely related to the mech­ anisms of corrosion and their resultant products rather than to the problems of corrodents and related internal chemistry.

It is necessary to include discussion of the simple oxidation of iron and iron-bearing alloys to obtain any background concerning the mechanism for the formation of an oxide film on a metal. For example, McCullough, Fontana and Beck^ studied the oxidation of stainless at various

^H. M. McCullough, M. G-. Fontana, and P. H. Beck, Formation of Oxide on Some Stainless Steels at High Temperatures, Trans* of the A.S.M. (1950).

conditions of time, , and * Their work indicated the presence of corundum and spinel oxide products 5 and also suggested that the resistant nature of certain oxides formed is related to their abnormally high chromium content. However, the films produced were far thinner than those of interest here.

Davies, Sinmad, and Birchenall^ investigated the mech-

^Davies, M, H., Sinmad, M. T., Birchenall, 0, E., On the Mechanism and Kinetics of Scaling of Iron, Trans, A,I,M,E, 191(1951),

anisms for oxide formation on iron in air. They noted the formation of the different amounts of the oxides of iron at different tempena1ur«s. They further suggested that the inflection in the In K vs l/T curve for iron was related to the appearance of small amounts of FegOg as a high tempera­ ture product, Pfiel® studied the diffusing in each of the oxides

^Pfiel, L, B., uThe Oxidation of Iron and Steel at High Temperature,” J, Iron and Steel Institute, 119 (1929),

of iron and found that the diffusing ion was Pe in FeO, Pe and 0 in FegO^, and 0 in FegOg. Birchenall et al, later confirmed these investigations by using radioactive silver 6 markers* These observations were again reported confirmed 6 by Dunnington*

®D* W. Dunnington, F. H* Beck, M. O r. Fontana, The Mechanism of Scale Formation On Iron at High Temperature, Corrosion. 8 , (1952).

With the addition of a corrodent to these simple oxi­ dation studies, work becomes more nearly applicable to this investigation; however, the work also tends to become more empirical in nature* Several studies which employed M0O3 as a corrodent have reported more than the gross observations. Leslie and Fontana1? found that the oxide product of

*?W* C. Leslie and M. G-. Fontana, Mechanism of the Bapid Oxidation of High Temperature, High Strength Alloys Con­ taining , Reprint #26. A.S.M* (1948).

a chromium-nickel-molybdenum stainless steel was a spinel with the a° equal to 8.52 A°*

Also with molybdenum, Brenner8 observed a concentration

8S. S. Brenner, Catastrophic Oxidation of Some Molybdenum- Containing Alloys, J. Eleotrochem. Soc.. 102 (1955), p 16.

of molybdenum at the interface and indicated a transfer of the material by the corrodent from the metal to the oxide 7

In the form of metal oxide dissolved in the corrodent# However, he observed a linear rate of oxidation# With the vanadium corrodents specifically, the data concerning the mechanisms of corrosion are meager. Bowder9

9A. T. Bowden, D. Draper, H# Rawling, wThe Problem of Fuel Ash Deposition In Open Cycle Turbines,* Proc. Inst# Mech. Engr. (1953).

reports , vanadates, and spinel, although he is obviously including the effects of sulfur and vanadium corrosion# Hall's^ work likewise includes both vanadium l^A# M. Hall, D# Douglas, J# H# Jackson, ’’Corrosion of Mer­ cury Boiler Tubes During Combustion of a Heavy Residual 011,” Trans. A.S.M.E., (1949).

and sulfur corrosion; however, he reports sulfates, sul­ furic , and VgOg.

Monkman and Grant^ studied the effect of V2 O5 on iron

■^F. C. Monkman and N. J. Grant, ,fArv Investigation of Accel­ erated Corrosion of Heat Resistant Metals Due to Vanadium” , Corrosion 9 (1953).

and found FegOg and Fe^O^ as the products# By means of tracers, they also determined that the concentration of 8 vanadium at the Interface was extremely high*

■Evans*1-2 gives much general information on the corrodents

V o ' ' . ■" | 1 1 ^Evans, C. T*, Corrosion*

and their origin; however, he offers little on the nature of the products or the controlling factors of corrosion* Mode of Investigation Basically, this investigation consists of the determin­ ation of the mechanisms of corrosion of 310 stainless steel by NVg. The specific corrodent, NVg, has been selected be­ cause of its pertinence in the vanadium corrosion and be­ cause it permits study with no more than one corrodent in operation* The base material, 310 stainless steel, has been selected from ai practical standpoint because of its wide application in high temperature component fabrication. Various methods of the study of reaction rate are pos­ sible* These are periodic chemical, X-ray, or petrographic analysis, atmosphere composition or pressure determination (oxygen is a reactant in the corrosion of stainless steel), or determination of corroding specimen weight changes re­ sulting from oxide formation* The analytical method involves the difficulties arising (i) from a lack of continuity, since the specimen must be disturbed for analysis, and (2 ) from time-consuming procedures. This is contracted with the 9 compensating ability to indicate reactions not involving oxygen or weight changes. Of the choices involving oxygen determination, weight change measurement at temperature or thermogravimetric analysis was chosen for its greater sen­ sitivity and accuracy. The knowledge of increases in weight of a specimen re­ sulting from the formation of oxides of the base material would be indicative of total oxygen reacted, and would not give information concerning intermediate steps, such as the formation of suboxides or intermediate compounds. To cir­ cumvent this difficulty, a second method of investigation sensitive to such possible changes in the system is neces­ sary. This information may be either chemical or structural in nature, or both. For various reasons, a physical evalu­ ation of the structure was used, specifically employing an X-ray diffraction analysis of the reaction products. Speed and simplicity of analysis techniques, reliable location of the sample under investigation, and availability contributed to the selection of X-ray spectrometer techniques. Examin­ ation of reaction products without removal from the specimen permitted accurate determination of the sample under investi­ gation plus evaluation of a very thin layer of material.

Studies of several series of specimens corroded for varied times at different temperature were made. These, coupled with occasional chemical analysis, yielded an 10 understanding of the final corrosion products*

The combination of these two principal modes of inves­ tigation, thermogravimetric analysis and X-ray diffraction analysis, coupled with some chemical analyses proved to be effective* The kinetic data could be correlated with rates for various physical phenomena, and then a selection of the proper phenomenon made. By means of X-ray diffraction analysis of the structure of the reaction products, knowl­ edge could be gained concerning the disposition of the var­ ious produots, changes in these products with respect to time and temperature, and, to some degree, changes within the specific products themselves*

T. MATERIALS The materials employed in these investigations were obtained either from normal chemical suppliers, or, where this was impossible, from synthesis in this laboratory. The materials are listed below with sources. Where labor­ atory synthesis was necessary, the technique employed is given at the end of this section.

Table 1 Materials

Materials Supplier Grade

J, T, Baker Chemical Baker Analyzed V2°5 Co,

NaV03 Vanadium Corp, of Vacoram, C, P, America (dried 48 hrs, at 100° C) NaVOg Synthesized at this laboratory NaOH Mallinckrodt Chemical Works

NagCOg Mallinckrodt Chemical Anhydrous, Ana­ Works lytical Reagent

NagO•Vg04 •5VgO 5 Synthesized at this laboratory

Na3 V0 4 Vanadium Corp. of C. P. America

Na3 V04 Synthesized at this laboratory

NagO • 3Vg0g Synthesized at this laboratory

11 12 Table 1 (contd.)

Materials Supplier Grade

PeV04 Synthesized at this laboratory

PeVs04 Synthesized at this laboratory

V>4 Vanadium Corp* of C* P. America Vanadium Corp. of C. P. V2°3 America

Pe2°3 J. T. Baker Chemical Baker Analytical Co* Be agent

Pe Plastic Metals Co. Stainless Steel, On hand Cr, 24*0-26.0$; Type 310 Ni, 19*0-22.0$ This material was Mn, 2*00$ max; available in sheet C, 0*25$ max., stock 0*050 thick Pe, balance* and in powder form -200 -t-325 mesh.

Iron On hand Pe, 99.6$ min; 0*040 thick

Nickel On hand Ni, 99.8$, 0*030 thick Chromium Electrolytic Cr, in approx. 3/4 chip form inches in dia.

Chromium On hand Electrolytic Cr, -325 mesh powder prepared in plates by press­ ing and sinter­ ing in Hg at 3050° P for one hour* 13

Table 1 (contd.)

Materials Supplier Grade

Altunina crucibles On hand Morganite No. x N 10 , 7/8 m $ O.D. x 1 3/8 in. high recrystal­ lized alumina crucibles.

Mullite crucibles On hand Coors No. 39 0 size No. 1 30 diameter mullite porcelain cru­ cibles.

The synthesis techniques employed for compounds which

could not be obtained through normal chemical supply channels are given below. All materials so prepared were checked by optical and/or x-ray methods to determine that a single phase existed or that only a minor amount of a second phase was present as an impurity.

Na2 0 *Vg0 4 *5 V2 0 5 ! This material was prepared from C.P.

VgOg and C.P. NaV0 3 by weighing, blending and fusing for one hour at 1600° P. in a raullite crucible. The material was chipped from the crucible and crushed. In some cases where high purity was desired, this crushed material was leached with hot water for 50 hours to remove any soluble Impurities.

NagOSSV^Og: This material was prepared in a like man­ ner to NagO *7 2 0 4 *5 7 2 0 5 . 14

NaV0 3 * This material was prepared by solution of VgOg in concentrated NaOH(aq). Proper proportions were obtained by mixing approximately 10 H NaOH (standardized with HC1) and weighed amounts of C.P. VgOg. The solution was then boiled until all VgOg was in solution (pH about 6 ) and de­ canted and evaporated# NagVO^ This material was prepared in the same manner as NaVOg. However, the pH in this case is greater than 10#

PeV0 4: This material was formed by reacting equimolar parts of C.P. V2O5 and C.P. FegOg at 1550° F for 18 hours in air. X-ray diffraction indicated complete reaction with no residual FegOg or VgOg. The pattern obtained checked that reported by Monkman and Grant for this compound

^Monkman, F. C., Corrosion. EQUIPMENT

Initial attempts at thermo-gravimetric analysis were made by an entirely manual method to determine the feasibility of the method* When it was found that much data were to be obtained and sharply time dependent variations existed, plans for a more sophisticated apparatus were made. When it was further learned that reactioh times would often be 48 hours or more, an automatic recording device was deemed necessary. The automatic recording balance constructed was based on a design by MaueP-4 and is shown in Figure 1. It was

14Mauer, F. A., w Analytical Balance for Recording Rapid Changes in Weight The Review of Scientific Instruments, 25 (1954).

selected because of Its wide range, simplicity of operation and adjustment, and low cost. The unit consists of a stand­ ard analytical balance with a magnet replacing one pan and weights. The magnet Is Inside a solenoid through which a controlled variable current is passed. The magnetic field Induced in the solenoid creates a force on the bar magnet which may be made to control the balance of the unit. De­ flection of the balance beam is detected In a circuit by a dual photo-cell energized by a light reflecting from a affixed to the top of the balance beam. The electri­ cal imbalance from the bridge is amplified and fed through

15 16 the solenoid to restore balance. A strip chart recorder, detecting the potential drop across a resistor In the solenoid circuit, Indicates -che magnitude and direction of this balanoing current and, hence, the change in weight which has occurred. The specimen is attached to a wire from the bar magnet and is suspended in a furnace below the balance. The fur- naoe Is a Kanthal-wound vertical tube furnace employing two concentric procelain tubes separated by ground refractory brick to increase thermal stability. The resistance wind­ ings and a Ft - Pt 10$ Bh control thermocouple are outside the tubes, while a Pt - Pt 10$ Bh indicating thermocouple is inside the tubes. The control couple is connected to an automatic Foxboro potentiometer controller, while the indi­ cating couple runs to a standard Leeds and Northrup poten­ tiometer. Control is such that normal variations from the Poxboro controller are not detectable by the indicating thermocouple. A water-cooled top helps maintain stable temperatures and also provides a means of introducing controlled atmos­ pheres. A stainless steel gas inlet tube (which also suf­ fices as a thermocouple support) runs to the bottom of the furnace. The controlled atmosphere, if employed, exhausts from the small hole in the top through which the balance wire passes. Thus, although the system is not completely 17

Pig. 1. Automatic Weight Change Reoording Equipment Showing Various Components. 1) Strip Chart Recorder 2) Illumination Source 3) Photocell 4) Control Solenoid 5) Range Adjustments 6 ) Potentiometer (Temperature Measurement) 7) Furnace 8 ) Furnace Controls 18 sealed from the atmosphere, a continuous one-way flow of gas through the hot zone is capable of maintaining a largely controlled atmosphere* Figure 8 shows the top of the furnace with various attached* The recorder is a standard Leeds and Northrup Micromas recorder with a scale of +, 1 0 millivolts* Through control panel adjustments, which may be made while the unit is in operation if required, a range of ^ 1*500 grams Is?, obtain­ able* Through changes in the bar magnet, additional exten­ sion of the range Is possible although unnecessary in this work. Provision has also been made to add automatically a small weight when the recorder reaches full scale to reset the balance and allow operation to continue. With this provision, continuous recordings of weight may be made over changes of 10 to 15 grams and periods of hundreds of hours without adjustment of any kihd. The automatic recorder re­ quires no auxiliary equipment other than an additional balance to determine initial specimen and ash weights* Subsequent to initial construction, several questions aroae concerning the preciseness of its operation. Conse­ quently. the following standardization techniques were em­ ployed. These were repeated as needed when some change was made in the operating system. The extent and location of temperature gradients within the furnace were checked wjLth a traveling thermocouple. The 19

Pig. 2 . View of Furnace Showing Various Water Cooling Leads, Atmosphere Entranoe Lines, Thermocouple Wires, and Power Leads. 20 specimen location was then determined accordingly. Absolute temperature calibration of the hot zone was made with a thermocouple which had been standarized against aluminum and having melting points certified by the National Bureau of Standards, Speolmen heat-up time was determined by imbedding the bead of a thermocouple in a small hole in the surface of a standard stainless steel specimen. The time to reach oper­ ating temperature was found to be two to three minutes. This value agrees quite closely with that indicated by the inter­ section of the rate curves with zero weight gain. Finally, possible oxygen starvation within the furnace was determined by running a standard specimen with and with­ out a forced air atmosphere. It was found that an arti- fically forced draft had no effect up to the amount which would bouy the specimen (about 4 liters/minute). A controlled atmosphere, however, was obtainable with only 1 / 2 liter/minute. The equipment used for the x-ray diffraction studies consisted of two Norelco x-ray diffraction units one of which was equipped with a spectrometer. Accessory equipment em­ ployed consisted of a metallographic polishing unit with diamong faced grinding and polishing wheels for the sec­ tioning of the products. A recessed metal clamp was used to hold the specimen by the edges for grinding. A copy of U. S. Dept, of Commerce Tables was used to translate direct­ ly for 2 0 values wdn spacings. PROCEDURE

The preparation for the metal specimens for kinetic analysis was maintained to a standard procedure. One inch by two inch specimens were cut from sheet stock, drilled with a l/8n inch drill at one end and ground to remove burrs. They were cleaned first in benzene, next in acetone and then weighed. The corrodent, where used, was applied from a water suspension with a small brush, A drop of 3# santomerse was used as a wetting agent. The corrodent was dried and reweighed to determine the amount applied. This process was repeated until the required amount had been applied. The standard amount of corrodent was approximately

0,1500 grams. The specimen was hung from the balance wire and lowered part of the way into the furnace and the water-cooled top lowered into place. When the furnace hot-zone had returned to temperature, the specimen was lowered into the hot-zone, and recording begun. When specimens were submerged in crucibles of ash, metal specimens were approximately 0.700 inches in diameter, A short piece of thermocouple protection tube was used to hold the specimen off the bottom of the crucible to permit circulation of the ash. Recrystallized alumina crucibles were used to contain the ash and the specimen, while a plat­ inum bail was used to attach the crucible to the balance wire. The introduction of the specimen into the furnace 21 22

was the same as for the small plates. When these speci­ mens were removed from the furnace, they were either broken or sectioned to permit examination of the metal specimen. No standard reaction time was employed for either type of specimen. Time varied from 20 to 150 hours. The gravi­ metric data were then read from the charts and replotted as required. The procedures for the x-ray diffraction studies of the products of corrosion employed two types of analyses. In a few cases, normal powder methods were employed for the Debye-Scherrer method. However, since the bulk of the x-ray

diffraction was directed toward the determination of the oxide structures without their having been removed from the specimen, some special procedures were used. For the pov/der methods, material was ground to -525 mesh and either extruded with Duco cement as a carrier and binder or placed in a Lindemann capillary. Exposure

times were 6 hours with Cu Koc radiation and a nickel filter. Intensities were estimated visually, while wd" spacings were read with overlay templates. For the sectioned specimens, where the reaction products were examined in place, a special technique was employed.

A 1 in. x 2 In. metal plate whioh would fit directly into the spectrometer sample holder was used as a specimen. The cleaned metal speoimens were coated on one side with a 23

+ .0000 quantity of NVg equal to 0.0750 _„o025 Srams 011(1 Were fired in air to the desired time-temperature relationship. when a series of speoimens were to be fired at varying times at the same temperature, they were introduced period­ ically into a heated furnace and removed simultaneously. Removal was after force-cooling the specimens in the fur­ nace to 400-500° P. This procedure was necessary to prevent spalling of the products from the surface and still permit rapid quenching of the reaction. This same cooling procedure was employed for all specimens in order to improve consis­ tency among various" runs .

A small recessed holding block was used to clamp a specimen by the edges so that its flat surfaces could be easily ground with a diamond impregnated grinding wheel. When one side of a specimen! had been ground down to a clean metal surface, the specimen was measured at each comer and at the center to obtain a reference thickness. Then the reverse side was repeatedly examined by x-ray analysis and repeatedly ground to remove reaction products in a thickness of several thousands of an inch. Record was made of the amount of material removed during each grinding operation so that the exact distance from the surface at which the speci­ men was being examined would be known. This procedure wa3 repeated until the side under examination exhibited metal.

'i 1 In this manner, a complete record of variations in structure of the reaction products from the Interface to the surface would be obtained. All x-ray diffraction examinations were made under fixed exposure conditions by using Cu k ot radiation with a scanning speed of 9 2 2°/min. All power settings, slit settings, etc., were held constant. Recordings of intensi­ ties were made over a range 15° < 29 < 165°. Calibration of the spectrometer with a standard sample was made intermittently to prevent long term drift of equipment. BESULTS

The results of the x-ray diffraction studies of the reaction products analyzed in place are listed according to the time-temperature relationship of their reaction as sev­ eral series of specimens were prepared. These include two series at 1600° P, with times varying from 0 to 30 min. and from 0 to 96 hrs., one series at 1200° P with times varying from 0 to 48 hrs., and one series for 2 hrs. with temper­ ature varying from 1300 to 1550° P. The sectioning data are listed in Tables 2, 3, and 4. The x-ray diffraction data are shown in Figure 3. The data shown in Figure 3 are rep­ resentative data for all specimens at each corresponding temperature. No significant variations were noted in the general nature of these results within a series, and so all individual patterns are not shown. The significant data for the back reflection portion of the spinel data are shown in Table 5. A list of the thermogravimetric determinations made is given in Table 6 . Listed in this table are the figures in which the data are presented. These figures are distributed throughout the discussion section for ease of reference.

25 26

Table 2 Amount of Material (Inches) Removed Prom Each Specimen Prior to X-ray Diffraction Study

Time at 16008 P (hrs.) X-ray Number 96 32 8 2 1/3 1 /1 2

1 .0 0 0 0 # .0 0 0 0 .0 0 0 0 •0 0 0 0 .0 0 0 0 .0 0 0 0

2 .0 0 1 0 # .0025 .0015 .0 0 2 .0 0 1 .0 0 1

3 .0 0 2 0 # .0055 .003 .0 0 2 .0 0 2 4 .004 .0105

5 .0075 .013

6 .0 1 0 7 .014

8 .017

^Reaction product mounted in resin. 27

Table 3 Amount of Material (lnohes) Removed Prom Each. Specimen Prior to X-ray Diffraction Study

X-ray Time at 120G0 P (hrs .) Number 48 24 6 1 /6

1 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 0

2 .0 0 2 0 .0015 .0 0 2 .0025 3 .003 .0025 4 .005 .004

Table 4 Amount of Material (Inches) Removed Prom Each Specimen Prior to X-ray Diffraction Study

X-ray Time at 1600° P (min.) Number 30 70 10 5 0 *H*

1 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 0

2 .0018 .0 0 2 2 .0 0 2 0 .0 0 1 0 .0 0 1 0 3* .0045 .0048 .0040 .0035 .0035

* . NVg just fused on surface. 28

Table 5 Lattice Parameter (a) for Spinel at Various Depths After Corrosion at 1600° by NV6

Position

Specimen Surface •0 0 2 down

#1 - 96 hrs. ,8317 .8307 #2 - 96 hrs. .8318 .8312 #2 - 32 hrs. .8321 .8309

# 1 - 8 hrs. .8320 .8310 Ave. all values .8319 .8310 29

Table 6 Thermogravimetric Analysis Buns

Bun Temp. Base Material Corrodent Pig* Bemarks

No. °P. Type Grams No.

1 1 1 0 0 310 nv6 0.1480 16

2 1200 310 nv6 0.1466 14,16

3 1300 310 nv6 0.1474 14, 16

4 1430 n o nv6 0.1454 14,16 * 5 1465 310 nv6 0.1488 14,16

6 1500 310 nv6 0.1450 14,16

7 1600 310 nv6 0.1448 14,16

8 1700 310 nv6 0.1485 14,16

9 1600 310 nv6 0.1495 - Preoxldiz< specimen

10 1600 310 nv6 0.1518 19 Borroded and strip]

11 1600 310 - - 13 Blank

12 1300 Fe NV6 0.1526 10

13 1400 Pe nv6 0.1494 10

14 1500 Pe nv6 0.1448 10

15 1600 Pe nv6 0.1479 8 ,1 0

16 1300 Pe - - 10

17 1400 Pe - - 10

18 1500 Pe — 10 30

Table 6 (contd.) Thermogravimetric Analysis Runs

Run Temp. Base Material Corrodent Pig. Remarks No* °P. Type G-rams No.

19 1600 Pe -- 8 ,1 0

20 1600 Ni nv6 0.1450 9,11

21 1700 Ni hv6 0.1492 9,11

22 1200 Ni - - 11

23 1400 Ni - mm 11 24 1600 Ni - 9,11 25 1700 Ni - - 9,11

26 1400 Cr NV6 0.1463 12 calc, to normal spec,

27 1600 : Gr nv6 0.1504 12 calc, to normal spec,

28 1400 Cr mm - 12

29 1600 Cr - - 12

30 1700 Cr - - 12

31 1600 310 nv6 0.1490 18 10# NV3 addi tion

32 1600 310 nv6 0.1500 18 25# NV3 addj tion 33 1600 310 nv6 0.1430 17 mixed with 1 gm oxide

34 1600 310 nv6 0.1460 17 reacted witt 1 gm oxide

35 1600 310 nv6 8.8026 20 crucible of ash 31

Table 6 (contd*) Thermogravimetric Analysis Huns

Run Temp. Base Material Corrodent Fig. Remarks No* op. Type Grams No.

36 160G 310 nv6 8.4353 - crucible of ash in Ng

37 1600 Pe m e 8.7895 20 crucible of ash

38 1600 Pe nv6 8.7284 - crucible of ash in No DISCUSSION

The two methods of investigation employed in these in­ vestigations, thermogravimetric analysis and x-ray diffrac­ tion analysis, provided distinct although somewhat over­ lapping results. For more cogent presentation, the x-ray diffraction results will be summarized as a unit, and then introduced as required into a discussion oriented around the more productive thermogravimetric analysis. The x-ray diffraction results for the examination of

the reaction products of NV6 on 310 stainless steel at 1600°F were similar throughout the entire time range studied. In all cases, a mixture of corundum and spinel structures was found. Since both of these permit liberal substitution of the various metallic involved, little quantitative data concerning the disposition of the different ions into the two structure could be obtained. The relative quantities of the two structures also remained constant with time. Coup­ ling this with the change in composition of the product as reported by chemical analysis (See Table 7) and with careful measurements of the a value of the spinel, it is suggested that the choice of structure is not determined by the rela­ tive quantities of metal ions forming the structure but by conditions of formation, e.g., temperature. The fact that the spinel-corundum ratio charges with temperature as indi­ cated in Table 8 suggests that some other factor, perhaps thermodynamic relationships, are controlling. 32 33

Table 7 Chemical Analysis of 310 Stainless Steel and Oxidation Product

310 Stainless Oxidation product Steel During First Hour o o .

Fe(wt.$) act. 52.27^ .02 39.69 t • Ni(wt.$) act. 24.51 i .03 14*07 t .03

Gr(wt,#)act, 21.65 - .09 15.61 t .26 Fe fraction 53.1 57.24 Ni fraction 24.9 20.25

Cr fraction 2 2 .0 22.51 Table 8 Oxide Ratio for Corrosion Products.of 310 Stainless Steel Coated with HVg

mtenaity of Standard line Spinel Corundum Tempgrature for each Structure Corundum Orthovanadate Orthovanadate Corundum Spinel

1600 0 5 23 4.6 — 1550 0 12 37 3.1 1500 0 24 29 0 .8 — w 1450 40 32 0 —- 0.73 a* 1400 17 12 0 0.70

1300 36 3 0 -- 0.08

1200 29 0 00 35 It should be noted that the spinel-corundum ratio does not change from the surface to the b ase material. This is in contrast to the gradation in oxide composition found in the normal oxidation of iron in air. Table 5 also shows that the section of largest a value, and thus the highest iron content, is at the surface and remains at the surface. Since chemical analysis has shown that the more rapid iron oxidation takes place initially, this would suggest that some inward of oxygen occurs, leaving the surface layer unchanged. It has been shown that there is some inward diffusion of oxygen in iron oxide, thus indicating agreement in these

results. However, much of the oxygen diffusion in iron oxide takes plao© in the oxygen deficient corundum structure present as a surface layer. Sine e no surface layer differ­ ing in structure was found, a slightly different reasoning must be applied. However, an appreciable amount of oxygen diffusion (about 20 per cent) is reported in spinel. This amount would appear to satisfy requirements for oxygen transport. As is shown in Figure 3, when the temperature'of reac­ tion falls below approximately 1450° F, a marked change oc­ curs in the structure of the reaction products. Spinel ceases to appear as a product, while a very poorly crystal­ line material, resembling FeVO^. structurally, appears. The 36 very poor x-ray patterns obtained for the orthovanadate make a precise determination of this structure difficult, particularly with respect to substitution of other ions. Back reflection data which are needed for this type of analysis were not obtainable. The specific cause of the change in structure of the products occurring at 1450° P is important to practical cor­ rosion studies. The fact that the rate of corrosion is dependent on the structure of the oxide film suggests the possibility of reduction in corrosion rate by controlling the structure of the oxide, e.g., stabilization of the low temperature products. ^he difference in low temperature products, specifically the substitution of the orthovanadate for the spinel, markedly affects the diffusion rates as shown in the thermogravi­ metric analysis results. This lower rate would apparently dtem from the lower symmetry, the poor crystallinity (as indicated by the weak x-ray diffraction patterns), or both. The analysis of the kinetics of reaction is a powerful method of study. In the situation here, where the gravi­ metric determination of oxygen provides a simple determina­ tive method, it Is ideal. Howeger, the method does not provide information concerning individual reactions, or the occurrence of more than one reaction at a time. Such ef­ fects must be determined by either a secondary method or 'd"l 1200 1300 1400 1450 , ,1500 _ IfiSfi _ 1ROO _ ... J_L _ J £ Oh 0 Oh wo o £ TO ft u o 111 i L 1 il Figure 3. "d" Spacings for Reaction Products of NV of Products Reaction Spacings for "d" 3. Figure __ & CO 1 1 i L _ P =Sie; O. Crnu; RH. Orthovandate - ORTHO. Corundum; - COR. Spinel; = SP. 2Sssg _ L 1 J S5H i _ il i Li_ ___ O os i_ l . J l. J_ i . . ... O od o cn I 1 I i 1il 1 . J O 0d o 1 1 __ Z to L » 1 1 5 CJ o o c hH h Z > E K 1 il O O O Gj OS ..

. .... 0 n30 tils Sel t 1600°F at Steel on310 Stainless . 1 .... O od o ......

. .

..... I > Z CO .. 38 by inference from the effect of variation of parameters in additional weight gain studies. In these investigations a combination of both methods was used. The x-ray diffrac­ tion and chemical analyses were used as secondary methods and the following parameters were varied in the thermogravi­ metric analysis to determine their effects upon the kinetics of reaction: -

1 . The influence of each of the three major alloying elements in the metal

2 . The influence of ash composition 3. The effect of temperature of reaction 4. The effect of the reaction products on further corrosion This last parameter markedly overlaps the first two in that changes In effective metal composition or ash composi­ tion may occur as reaction progresses. However, unique effects are possible, e.g., reaction to form new compounds and changes in the physical or chemical composition of the old. Since these data obtained from the thermogravimetric determinations are, of course, the sum of any or all of the many possible reactions, it is necessary to look into the theory involved in simpler systems to learn at least some of the reactions and their associated rates which may be

expected. 39

The possible rates for the simple oxidation of a pure

metal with no corrodent are numerous; consequently, the addition of a metal and a corrodent would appear to make the problem insurmountable. However, the specific conditions of reaction and the degree of corrosion reduce the number considerably. For example, the high temperature

involved in these investigations (1100-1700° F) is consid­ ered to bring the problem into a range where either parabolic or linear rates of growth are to be expected. Further, the high degree of corrosion produces a thick film of oxide. This eliminates the existence of electric fields in the oxide as a result of- electronic™transfer from metal ions to surface oxygen ions. This condition exists in very thin films and produces inverse logarithmic oxidation rates.

■^Smeltzer, W. W., Principles Applicable to the Oxidation and Corrosion of Metal and Alloys National Assn. of Cor­ rosion Engineers. 11th Annual Conference, Chicago. 111. (1955).

With oxide films of the order found here, these and similar effects are nil. Consequently, under the experimental conditions imposed here, the oxidation of a metal may be considered to follow

one of two rate processes. The first is linear, in which the plot of weight gained versus time is a straight line with positive slope. This type stems from a non-proteotive 40 which does not affect further corrosion* The second type is parabolic, in which the plot of weight gained versus time is a parabola about the time axis. Here the oxide coating is protective and reaction is slowed by the neces­ sary diffusion of ions through the oxide reaction product. Thus the rate of oxidation is inversely proportional to the amount of coating accumulated, i.e.,

dW = K _l_ dt ' " 0 W

Integration yields W s Kx Vt .

A graphic representation of these two forms of oxida­ tion is shown in Figures 4 and 5. Rates of oxidation fall­ ing between these two limiting cases would normally result from physical cracking or spalling of a protective layer, effectively giving a series of parabolas. However, it would be expected that if this were the case, the experi­ mental weight gain curve would not be smoothly continuous but would exhibit small breaks. Some alteration of these two fundamental types of oxi­ dation would result in basic changes in the shape of the curve, thus requiring a change in the form of the equation representing it; while other variations would require only changes in the parameters of the function without changes in the form. An example in which only parameters change 41 would be represented by a parabolic rate in an oxygen defi­ cient atmosphere. This condition would W * Kg -/t as compared with W = *\{% for a normal atmosphere where Kg < K^. See Figure 4 and 5. Figure 5 shows a parabolic plot in which (time) forms one axis. This type of plot displays a parabola as a straight line and permits calcu­

lation of the oxidation rate constant from the slope of the line. It is used extensively in this discussion. When additional chemical entities are introduced into the system, as is the case here, by coating the specimen with a vanadium-bearing ash or by using a metal in which the composition may change as reaction occurs^ the analysis be­ comes more difficult, although basic physical laws must still apply. For example, in oheiaical reaction where there is a limited amount of one of the reactants, progress would —let be described by W = W0e , where W is the amount or con­

centration of the reactant remaining at time t and W0 is the initial amount or concentration of specified reactant. Here, where we are measuring the amount of reaction product, the form would be W * Wq (l-e~^) where W is the amount of reaction product at the time t, and Wq is the total amount of product capable of being formed. By multiplying out, re­

arranging, and taking logarithims, one obtains In (W0 - W) "

In W0 - kt. Since In WD is constant, plots of In (WD - W)

against t may be used as shown in Figure 6 . Weight Gain iue . osbe xdto ae o Metal a of Rates Oxidation Possible 4. Figure W = k = W Time 2

t V Weight Gain iue . aaoi OiainRts fa Metal a of Rates Oxidation Parabolic S. Figure w w 2 r k2y - Maximum Gain Minus Actual Gain, 70 ioa_ 0 Figure Figure 6. 1 hoeia Gi fr hmcl Reaction Chemical for Gain Theoretical Time 44 2 3 4 45

With this background, investigation of the corrosion reactions can be begun. To determine the success of various calibration techniques employed in the control of the thermo- gravimetric analysis equipment, a determination of NVg on 310 stainless steel at 1600° P was periodically made as a standard run throughout the course of this investigation. Table 9 shows a statistical analysis of these results and Figure 7 shows a plot of these results, Indicating the de­ gree of reproducibility obtained for this standard test. Further, Figure 10 indicates close experimental agreement with published data for the oxidation of pure iron, indi­ cating high accuracy. This includes confirmation of the break in this curve as reported by Davies.^ 7

^Davies, M. H., Trans. A.I.M.E.

As has been stated, early investigations indicated that the corrosion of 310 stainless steel by high vanadium ashes was extremely complex. Since the stainless steel employed was type 310, containing iron, nickel, and chromium, an attempt was made to reduce this complexity by investigation of the effect of NVg on each of the constituent metals.

Figures 8 through 12 show that the effect of this corrodent was quite small when compared with the normal oxidation rates. With iron, no difference was noted; with nickel, a 46 very small difference. With chromium, considerable diffi­ culty was encountered in obtaining reproducible results under any conditions. Even without corrodents, results varied by many orders of magnitude. It is believed that the difficulty stemmed from the very low oxidation rate exhibited by pure chromium. Consequently, minor variations in porosity, surface texture, preparation and composition of the metal, and variations in atmosphere would result in great increases in oxidation. However, it can be stated from Figure 12 that the addition of corrodent did not produce any variations which were greater than those observed with these samples of the pure metal. For iron and nickel Figure 9 shows the parabolic rate obtained and 10 and 11 show the excellent agreement with published rate constants. The rate constants are obtained from the slope of the parabolic plots with proper adjustment of units. Therefore, for at least iron and nickel and probably for chromium, the mechanism of cor­ rosion is not changed by the addition of a corrodent. The same diffusion controlled rate obtains. That the diffusion rate could remain unaffected with the presence of considerable amounts of corrodent requires some explanation. Monkman and Grant-*-® have shown that with

-*-®Monkman, F. C., Corrosion. Weight Gala (grams) 0.0 0.5 2.0 1.0 1.5 0.0 iue . egt an o Ng n 1 Sanes te a 1600°F at Steel Stainless 310 on NVg for Gain Weight 7. Figure 1.0 Tm)/ (hours (Time)1/2 2.0 )1/2 3.0 ttsia Average Statistical $ Individual Determination Individual 4.0

Weight Gain (grams) 0.0 0.5 1.0 2.5 2.0 1.5 0.0 Figure Figure 8 Wih Gl fr rn t 1600°F at Iron for Gala Weight . 1.0 (Time) 1/2 (hours) (hours) 1/2 1/2 (Time) 2.0

ihu NVg without 3.0 Weight Gain (grams) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.0 1.0 iue . egt an o Nickel for Gain Weight 9. Figure 2.0 Tm) ^ (hours)^2 *^2 (Time) 49 3.0

o Ash No 4.0 .0 5 in K -19 V -23 -21 -17 -15 -13 1800 701600 1700 o Without NV, Without o Birchenall V Wt NV With D iue 0 Oiain ae osatfr Iron for Constant Rate Oxidation 10. Figure 6

Temperature t 14001500 10.0 1300 10.5 1200 11.0 -23

-25

-27

* Gulbransen2A ® Brenner * 3 -29 O Without NVfl

-31

8.0 10.0 10.5 11.0 11. 1800 1700 1600 1500 1400 1300 1200 Temperature

Figure 11. Oxidation Rate Constant for Nickel -24

% Gulbransen** 2 -28 + Without NVg (limit)

With NV6 (limit)

-32

-36

-40U ______1 /T k x104 18.0 10.0 10.5 11.0 11.5

°F 1800 1700 1600 .1500 1400 1300 1200 Temperature

Figure 12. Oxidation Rate Constant for Chromium 53

Table 9 Statistical Data for Six Runs of NV5 on 310 Stainless Steel at 1600° P

Time Average Weight Gain Standard Hours Grams Deviation

0,5 0.5329 0.0177

1 .0 0.5137 0.0213

2 .0 0.7150 0.0160 3.0 0.8312 0.0167 4.0 0,9166 0.0172

6 .0 1.0549 0.0140

8 .0 1.1675 0 .0 2 1 2

1 2 .0 1.3456 0.0130 16.0 1.4839 0.0240

2 0 .0 1.6198 0.0400 Average 0,0202 54

VgOg on stainless steel there exists a high concentration of vanadium at the metal-metal oxide interface. This would allow the relatively unchanged metal oxide to remain as the controlling factor. It is surmised that the same conditions obtain with NVg. This supposition is supported by the con­ tinuous corrosive action of NVg after extremely long periods of time, which precludes concentration at the surface, and by the failure of the corrodent to affect the corrosion rate with pure materials, which precludes distribution through the scale. The situation with alloys, specifically 310 stainless steel, was considerably different. Figure 13 shows the comparative oxidation for 310 stainless steel with and with­ out corrodent. Note the continued oxidation even at 150 hours when corrodent is present. However, when the rate curves for stainless steel are plotted parabolically, Figure 14 is obtained. The parabolic oxidation rate indioated Is more clearly exhibited in Figure 15 which magnifies the initial data. Further, when the rate constants for this initial portion of the data for 310 stAInless steel are com­ pared with those published for various metals, excellent agreement is obtained with those for iron (see Figure 16).

This indicates that the same diffusion mechanism Is control­ ling and hence the diffusion media are the same. X-ray examination reported previously collaborated this similarity of diffusion media. Weight Gain grams) 0 1.0 . 0 0 | -O. 30 iue 3 Wih an o 30 tils Sela 1600°F at Steel Stainless 310 for Gain Weight 13. Figure 60 ie (hours) Time Without NV, Without 90 ih NV, With 120 150 Weight gain (grams) 0.0 0.25 0.5C- 1.00 1.25 0.0 iue 4 Weight NVgGain onfor 310 Stainless Temperatures Various Steelat 14. Figure 2.0 (Time)*^ (hours)^ 4.0 6.0 1300°F 0.00 Weight Gain (grams) 0.75 1.00 0.50 0.0 iue 5 Wih anfr V o 30 tils Sel t aiu Temperatures Various at Steel Stainless 310 on NVg for Gain Weight 15. Figure 0.5 (Time) ^ (hours) ^ (hours) ^ (Time) 1.0 -o- 1.5 2.0 1430°F °F 1800 1700 1600 1500 1400 1300 1200 1100 Temperature

Figure 16. Oxidation Rate Constant for NVg on 310 Stainless Steel 59

After the first hour, however, the rate of oxidation of 310 stainless steel with NVg no longer remains parabolic. The rate begins to decrease markedly. Investigations of pure metals indicated that no change occurred within the ash. However, it was possible that the presence of the oxide formed during the first hour interfered with the accelerat­ ing action of the ash, thus reducing the rate. To determine the effect of this oxide on the accelerating properties of NVg, the investigations reported in Figure 17 were conducted.

Here, 310 stainless steel powder was oxidized and checked by x-ray analysis and mixed with NVg ratio of 1 part oxide to 0.15 part NVg. A portion of this mixture was pre­ reacted at 1600° F for one hour. This mixture, both as mixed and pre-reacted, was painted on standard 310 stainless steel speoimens in amounts such that the NVg present would by the standard 0.1500 grams. The curves shown in Figure 17 indicate that no significant dilution or Inhibition effect occurs from the presence of oxide. Further, the effect of surface oxide film was further determined by pre-oxidizing a specimen to constant weight and then treating it with NVg under standard conditions. No significant difference in oxidation rate was noted*

Effect of minor variation in ash composition was de­ termined by the addition of small amounts of NagO'SVgOg to the NVg. Results shown in Figure 18 indicate the effect to be negligible. Weight Gain (grams) 2.0 0.0 1.0 0.0 iue 7 Wih Gi o V o 1 Sanes Steel Stainless on 310 NVg for Gain Weight 17. Figure 2.0 (Time )1/2 (hours ih r-ece Oxide Pre-Reacted With ih Oxide With 1 2 )1/ 4.0

Normal 6.0 Weight Gain (grams) 0.5 0.0 1.0 iue 8 Wih Gi fr V o 30 tils Steel Stainless on 310 NVg for Gain Weight 18. Figure 2 . 0 (Time)l/2 (hours)^ 3.0 5 NV, Addition , V 25% N . 5.0 4.0 Addition 0 0 Addition , V 10%N 6.0 62

Since these results indicated that the reduction In rate was not a result of changes in the ash or of effects brought about by the addition of the oxide product, it was then necessary to determine any changes occurring at the surface of the metal. Such effects might be expected from the variation in reactivity of the constituents of the alloy at the temperature of interest. Under normal conditions, nickel and chromium exhibit low oxidation because of the protective nature of the oxide, not the low driving energy. Actually, the free energy of formation of chromium oxide is higher (negatively) than that of iron. Consequently some degree of preferential oxidation may be expected. The existence of preferential oxidation is indicated by Figure 19. Here a specimen of 310 stainless steel was corroded for 15 hours and stripped of its ash products by air-quenching and stripping in fused Na©H for 24 hours. Controls were used to insure that the effects of the NaOH bath were nil. The stripped specimen was then exposed to a second corrosion cycle. A substantial change in the reaction rate was observed, indicating the presence of a surface layer ojf different reactivity# The most likely cause of this change in oorrosion rate would be composition of the metal in that the x-ray did not indicate any change in structure in the ash or in the base metal at any time during the corrosion cycle. Further, the Weight Gain (grams) 0.000 1.000 1. 500 __500 1. 500 Figure 19. Weight Gain for for Gain Weight 19. Figure 65 eplcto o Ash and of Products of Reapplication Stripping fter A r l Corrosion al orm N NV q n 1 Sanes te a 1600°F at Steel Stainless 310 on

10.0 64 chemical analysis conducted on specimens of scale collected from 310 stainless steel oorroded by NVg for one hour indi­ cated an increase in the iron content of this initial scale over the iron content in the metal. This is in contrast to the increase in chromium content normally found in the oxide of 310 stainless steel.These results are shown in Table 7.

19 McCullough, H. M., Trans, of the A.S.M.

That iron is the depleted metal might seem strange in the light of the lower free energy of formation of the oxide. However, the situation is not simple oxidation. The free energy difference among the metals for simple oxidation are meaningful only to indicate the existence of differences. With the addition of the NVg* the solubility of the various ions becomes a factor. Further, the free energy of forma­ tion may differ with the formation oxide from the state of the ions in the liquid corrodent. Consequently, the order of reactivity is apparently changed, with the iron being the highest. Apparently, after about one hour at 1600° F, the avail­ ability of ions of all types decreases (apparently largely because of lesser amounts of iron) below that amount which is capable of being diffused outward thru the oxide. Con­ sequently, the oxidation,rate falls below that of a purely parabolic form. 65

It does not seem that the decrease in rate can be at­ tributed to a ohange in the diffusion coefficient for the oxide products. The x-ray findings indicate no major changes in amounts of various oxides or in their structure. It is not believed that the small difference In composition indi­ cated by the chemical analysis and lattioe parameter de­ terminations could sufficiently affect the diffusion rates. Further, the data obtained from the stripped specimens indi­ cate that the difference lies in the metal. To obtain information concerning the effect of large quantities of corrodent, several investigations were made employing crucibles of corrodent containing submerged discs of metal. These tests were first conducted in a normal air atmosphere, and the results shown in Figures 20 and 21 were obtained. As shown, iron exhibits a normal first order chemioal reaction. This is as would be expected, since dif­ fusion is not the controlling mechanism. Stainless steel, on the other hand, does not exhibit a logarithmic rate be­ cause of the oomplex composition of the reactant. Since oxidation of iron is a first order chemical reaction eradi­ cating a concentration dependent reaction), it is suggested that oxidation does not occur at the metal-slag interface.

If this were the case, then, the reaction would be propor­ tional to the surface area of the metal, and this quantity remains relatively constant over a long period of time. iue 0 Mxmm an iu Ata Gi fr rn umre i Ng t 1600°F at NVg in Submerged Iron for Gain Actual Minus Gain Maximum 20. Figure Maximum Gain Minus Actual Gain (grams) 0.01 0.02 0.03 0. 05 0. 0.07 0 0.20 0.30 0.50 0. 70 0. 1.00, . 10 0 10 ie (hours) Time 20 30 050 40 Maximum Gain Minus Actual Gain (grams) iue 1 Mxmm an iu Ata Gi fr 1 Sanes te Submerged Steel Stainless 310 for Gain Actual Minus Gain Maximum 21. Figure 0. 03j 0.05 0.07 0.70 1.00 0.30 0.50 0.2C_ 0 in NVg at 1600°F at NVg in 10 20 ie (hours) Time 67 04 50 40 30

68

This suggests that the ions are dissolved in the corrodent prior to their oxidation# In an attempt to indicate the solubility of the metal ions in the cr.rrcsive ashes prior to their oxidation, sev­ eral runs fere made employing submerged samples of metal in the ash under a nitrogen atmosphere. The atmosphere was capable of reducing weight gain with iron to less than 0.007 grams over 48 hours. Severe attack of the metal was noted, although less than that noted when extensive oxidation was permitted# Examination of the metal specimen after having been submerged in NVg an % atmosphere revealed a hard dense layer at its surface which could be broken off for further examination. X-ray diffraction examination of this product formed at metal-corrodent interface revealed a very weak unknown x-ray pattern. It did not correlate with any of the known oxides of iron or any of the ternary iron-vanadium- * Failure to observe even the lowest oxide of iron, FeO, is taken as a further indication of the existence of metal In some state between solid iron and iron associated with the oxide, i.e.,

Fe^M —>• Fe^ ^ • NVg —>- Fe"**1 • 0 •

The oxidation of Fe may take place at the metal-corrodent interface or at the corrodent-oxide Interface. Further 69 investigation., probably involving cell E.M.P. determina­ tions, would be necessary. As may be noted in numerous parts of the results, e.g., Figures 3 and 16, a change in mechanism occurs at below about 1450° F. The initial diffusion controlled rate of reaction no longer agrees in rate constat with that for iron. Further, the wkneew in the curve is not so pronounced, while the parabolic rate exists for a longer period of time. These changes in kinetics of reaction appear to stem from the presence of different reaction products which provide a different diffusion media. The x-ray diffraction data indi­ cate that the reaction products at these lower temperatures are orthovanadate and corundum structures with no spinel present. Little information is available concerning the structure of the orthovanadate other than the fact that it is more complex than the spinel. Further, x-ray diffrac­ tion patterns for it indicate poor crystallinlty. » ...... • The longer agreement with the parabolic rate would stem from (1) lower corrosion rate, since amount of metal removed ■ . j would be the cause of deviation, and1 (2) greater time permit­ ting removal of larger amounts of the more resistant ions from the surface. Both factors are suggested by the gradual increase in the amount of weight gained at the point at which the curves deviate from the parabolic. As has been stated, the literature reports a similar break in the In K vs l/T curve for iron at about 1475° F, 70 With iron, this break has been attributed to the presence of corundum structure above this temperature. The specific cause of the comparable structure change at this temperature in these investigations was not deter­ mined. The agreement in temperature between the break ob­ served, in the iron curve and that observed here suggests a similar cause. However, the x-ray diffraction indicated that the change in structure here is not the same as that for iron in that here, corundum was detected at all temp- atures; however, the change from spinel to orthovandate may stem from the same cause. However, it has also been reported that the orthovana- o 20 date is -unstable at temperatures above 816 C (1500° F).

20 ; ' : Monkman, F. C., Corrosion.

It has been formed at this laboratory by reaction at 1550° F <*• however, the crystallization of the orthovanadate may have occurred upon cooling. It is possible that with the presence of other phases, as is the case here with the oxide products of stainless steel, this crystallization is prevented and hence no orthovanadate is formed upon cooling when the forma­ tion temperature is 1500° F or greater. Investigation would be required to determine which of these two factors is controlling. In this investigation, it 71 should be noted that this change in mechanism at lower temper­ atures is a change in degree rather than in kind. The in­ itial periods of corrosion are both diffusion controlled, although the diffusion media are different. The subsequent corrosion period is then concentration controlled by the iron concentration in the metal, although the period differs at which this rate factor becomes controlling. With these differences in mind, the mechanisms which control corrosion may be considered the same in principle. The initial rate is diffusion controlled. Apparently the action of the corrodent is to remain at the metal surface, dissolving the metal and carrying it out to the oxide layer through which diffusion takes place. The thin protective oxide which normally forms on 310 stainless steel is either rapidly dissolved or never permitted to form. The corrodent continues to carry metal ions, preferentially iron, from the surface of the metal to the oxide film where they diffuse outward. Since the number of ions available is diffusion is controlling. At such time as the quantity of metal ions available for diffusion becomes less than can be diffused through the oxide, the rate deviates from parabolic, and the concentration of iron becomes controlling. Although the iron concentration is apparently rate controlling, it is not the only metal being oxidized. Chromium and nickel are also involved, but to a lesser extent. Eventually an equilibrium 72 concentration should obtain which is fixed by the rate of attack of the more resistant metals. At this point the rate should again become parabolic, but with a lower rate constant than the initial period. This final parabolic rate is shown quite well in Figure 22. The implication of the results when applied to practical corrosion problems is not heartening. Since the action of the ash is to dissolve metal at the surface and do so pref­ erentially, the properties of the oxide formed are related to those of the poorest alloying element rather than the best.

Thus to eliminate the high diffusion rates associated with iron, it is neo®saury to eliminate iron from the alloy. Note the oxidation rate with Inconel x containing about 7 per cent iron shown in Figure 23. This increases cost considerably. The possibility of destroying the diffusing properties of the oxide does not appear to be very fruitful. Assuming that the alloy composition being corroded is comparable, the structure of the oxide, because its dependency on the poorer elements, would be comparable. Possibilities of packing the lattice would appear to be effective only to a degree. The most fruitful area for possible reduction in attack would seem to be through the structure of the corrodent. If some method of destroying or packing its structure could be found, corrosion should be reduced. However, since all vanadium compounds appear to be corrosive to some extent, corrosion would probably not be eliminated. Weight Gain (grams) . _ 1.5 2.5 3.0 8 iue 2 Fnl egt an o Ng n 1 Sanes te t 1600°F at Steel Stainless on 310 NVg for Gain Weight Final 22. Figure 9 Tme12 (hours//2 e)1/2 (Tim 10

11 12 Weight Gain (grams) 0.00 0.05 0.10 0 0.15 . 20 0.0 | _ iue 3 Wih an o Ng n noe t 1600°F at x Inconel on NVg for Gain Weight 23. Figure 1.2 ie (hours) Time) { 2.0 3.0 4.0 CONCLUSIONS

1, The rate controlling factors in the corrosion of 310 stainless steel by NVg progress through three time- dependent changes* 2* There is an initial diffusion controlled period during which the formation of an oxide layer is rate-control­ ling. During this period iron is depleted from the surface of metal.

3. Second, a period exists during which the concen­ tration of iron in the surface of the metal becomes rate- controlling, and the rate of corrosion falls below that pre­ dicted by diffusion alone. 4. Finally, an equilibrium concentration gradient in the metal obtains, and the rate again becomes diffusion- controlled with a lower diffusion constant than for the Initial period.

5. The nature of the reaction products which form on these conditions varies according to temperature of forma­ tion. 6. Below 1450° F., the reaction products consist of orthovanadate and corundum structures•

7. Above 1450° F., the reaction products consist of spinel and corundum structures•

8. The corrosion rates of the pure constituents of 310 stainless steel, viz., iron, nickel, and chromium, are

75 ' 76 not affected by the addition of the corrodent, The same diffusion-controlled mechanisms are controlling with or without the corrodent, 9, The preferential attack on the iron in the corrosion of 310 stainless steels suggests little probability for suc­ cess in alloying ferrous alloys to obtain a more resistant material. REFERENCES

1. Leipold, M. H., Shevlin, T. S., et al.# "Cemet Coating Development,” The Ohio State University Research Foundation Reports No. 1-30, Contract NObs 62161 (1954-1956).

2. Leipold. M. H., Shevlin, T. S., et all, "Effect of Specific Corrodents on 310 Stainless Steel,” The Ohio State University Research Foundation Reports No. 1-12 Con­ tract No. NObs 72157 (1956-1957). 3. Tables for Conversion of X-ray Diffraction Angles to Interplaner Spacing. U. S. Dept, of Commerce, National Bureau of Standards Applied Mathematics Series No. 10 (1950). 4. Handbook of Chemistry and Physics, Chemical Rubber

Publishing Co. Cleveland (1955). 5. Metals Handbook, American Society for Metals (1948). 6. Klug, H. P. and Alexander, L. E., X-ray Diffrac­ tion Procedures, John Wiley and Sons', New York (1954).

7. Evand, R. C., An Introduction to Crystal Chemistry» Cambridge University Press, London (1952). 8. Glasstone, S., Textbook of Physical Chemistry, D. Van No3trand Co., New York. (1946). 9. Barrett, C. S., Structure of Metals, McGraw-Hill

Book Company, New York (1943). 10. Dekker, A. J., Solid State Physics. Prentice Hall,

Englewood Cliffs, New Jersey (1957).

77 78

11. Smeltzer, W. W., "Principles Applicable to the

Oxidation and Corrosion of Metals and Alloys,” National Association of Corrosion Engineers, 11th Annual Conference,

Chicago, Illinois (March 7-11, 1955). APPENDIX

Physical Properties of NVg are as follows; Specific Gravity (picnometer) - 336 Color - blue-black

Streak - green-brown Indices of Befraction - 1.80 Birefringence - very high Color (transmitted light) - green-brown Melting Point - 646° G. X-ray Diffraction. Data

ud t» I

9.5 5 7.4 3 7.2 9 6.8 2 4.7 5 3.85 4 3.60 4 3.47 3 3.37 7 3.05 10 2.92 5 2.72 4 2.52 3 2.38 2 2.18 10 1.98 5 1.87 3 1.80 4 1.67 2 1.58 3 1.49 4 1.39 4 1.33 3 1.30' / 2 1.23 3

79 Jersey, ».**' w. i» from Cliff*14* X received W * <*m

neering » m V * *•«**<* ‘ search Fell** *• W W W periment t ■ * * » * * % in 1955* S u b W W ^ # * , * *■##**# WwNH«i* in the KngiM^V a * * * - 1* * .a ##» h e ld t h is pO*WAlj%.'»0 m . ,%$%§; requirements «»**.. M M **