300M USE LP? {(—1362)’ m 31 ‘9 mafia"? ABSTRACT THE PREFERENTIAL ACCEPTANCE OF CERTAIN IONS INTO THE FERRITE SPINEL LATTICE by Allen Vaughan Shaw It is well known that minerals of the spinel group crystallize with a variety of chemical formulas and that solid solution occurs freely among members of the various spinels series. Little work has been done, however, concerning preferential acceptance of ions into the spinel lattice. To fill this area to some degree, several members of the ferrite series of spinels were syn- thesized by heating mixtures of coprecipitated hy- droxides to temperatures of 6000 to 10000 C. and held there for not less than 12 hours. X-ray dif- fraction analysis showed that the order of acceptance at these temperatures and at one atmosphere of pres- sure was Zn, Mg, Ni, Cu, and Mn. Allen Vaughan Shaw It is concluded that the primary control of the acceptance is the electron configuration of the element and not the ionic radius or ionization poten- tial. These conclusions may be affected by an increase in temperature, pressure, or available constituents. THE PREFERENTIAL ACCEPTANCE OF CERTAIN IONS INTO THE FERRITE SPINEL LATTICE By Allen Vaughan Shaw A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Geology 1965 (”,9 r‘_ (in?) 21’ ACKNOWLEDGMENTS I would like to extend appreciation and thanks to Dr. Harold Stonehouse who suggested the problem, guided me over the course of the research and reviewed the manuscript. I would like to thank Dr. William Hinze, Dr. James Trow, and Dr. Justin Zinn for reading and cri- ticizing the manuscript. Appreciation is also due to my wife Gail, who encouraged me during this research. ii TABLE ACKNOWLEDGMENTS. . . . LIST OF TABLES . . . . LIST OF FIGURES. . . . INTRODUCTION . . . . . Review of Spinels. THEORY . . . . . . , . Predictions. . . . EXPERIMENTAL PROCEDURE DATA . . . . . . . . . DISCUSSION OF DATA . . CONCLUSIONS. . . . . . SUGGESTIONS FOR FUTURE BIBLIOGRAPHY . . . . . APPENDIX . . . . . . . OF CONTENTS iii Page ii iv vi 12 14 19 44 47 48 50 53 Table LIST OF TABLES Page (N) indicates normal structure is in- ferred, the cation distribution not experimentally established. Distribu- tions marked with asterisks are not en- tirely inverse; the number of divalent ions in the tetrahedral sites as tem- perature dependent.. . . . . . . . . . 3 Elements found in minerals with a spinel type structure, listed by valences. 4 All data except electronegativity after Green, 1959. Electronegativity after Ringwood, 1955. Asterisk indicates radii in four-fold coordination; all others in six-fold.. . . . . . . . . . . . . . . . 10 After Rankama & Sahama, 1950 Electron Configuration of the elements. . . . . . 10 Data from the ternary sample Cu.Mg.Fe compared with the binary samples Cu.Fe and Mg,Fe. . . . . . . . . . . . . . . . 22 Data from the ternary sample Cu,Mn,Fe compared with the binary sample Cu,Fe and anFeo o o o o o o o o o o o o o o o 25 Data from the ternary sample Cu,Ni.Fe compared with the binary sample Cu,Fe and Ni,Fe. o o o o o o o o o o o o o o o 28 Data from the ternary sample Cu,Zn,Fe compared with the binary samples Cu,Fe and Zn,Fe. iv List of Tables - continued. Table 10 11 12 l3 14 15 Data from the ternary sample Cu,Zn,Fe compared with the binary samples Cu,Fe and Zn,Fe. . . . . . . . . . . . . . . Data from the ternary sample Mg,Mn,Fe compared with the binary samples Mg,Fe and Mn,Fe. . . . . . . . . . . . . . . Data from the ternary sample Mg,Ni,Fe compared with the binary samples Mg,Fe and Ni I Fe O O O O O O O O O O O O O O 0 Data of the ternary sample Mg,Zn,Fe compared with the binary samples Mg,Fe and Zn,Fe. o o o o o o o o o o o o o 0 Data of the ternary sample Mn,N,Fe com- pared with the binary samples Mn,Fe and Ni,Fe. o o o o o o o o o o o o o o o 0 Data of ternary sample Mn,Zn,Fe com- pared with the binary samples Mn,Fe and anFeo o o o o o o o o o o o o o o o 0 Data of the ternary sample Ni,Zn,Fe com- pared with the binary samples Ni,Fe and Zn,Fe° . . . . . . . . . . . . . . . . D-spacings and intensities of the three most intense lines of compounds found in this Study. 0 O O O O O O O O O O O Page 30 32 34 36 38 40 42 43 LIST OF FIGURES Figure Page 1 Successive layers in the spinel structure showing the packing of the ions. (a) is the lowermost layer. After Bragg (1937). . 5 2 Structure of spinel ABZO . (a) is the lowermost layer. After Bragg (l937). . . . 5 vi THE PREFERENTIAL ACCEPTANCE OF CERTAIN IONS INTO THE FERRITE SPINEL LATTICE INTRODUCTION Geochemists have long studied various chemical systems in an attempt to develOp theories concerning the crystallization of the rock forming minerals and the geochemical distribution of the elements. Wash- ington and Clark (1924) determined that eight elements (0, Si, Al, Fe, Ca, Na, K, and Mg) were the most abun- dant in the earth's crust. These are termed the common elements by Rankama and Sahama (1950) and the remainder as trace elements. How these elements are distributed in the rocks and minerals of the earth is determined by a number of variables such as valence, ionization potential and ionic radius. In the following pages, a series of spinels of the ferrite group has been studied in the hope that some additional information might be gleaned concerning the function of these variables. Review of Spinels Minerals of the spiral type are among the earliest minerals to crystalize from a magma. Bowen (1926) places them just below the olivines in his re- action series, but from the interstitial occurrence of these minerals, it appears that they may continue to crystallize throughout the crystallization history of the magma. Spinels are quite ubiquitous, being found as accessories in metamorphic and igneous rocks and as heavy minerals in clastic sediments (Deer gt .§1., 1963). Palache et al. (1944) divide the spinel group of minerals into three isomorphic series based on the dominant trivalent ion; these are the aluminates, the magnetites, and the chromites. Two other geologic- ally important minerals within this structure type are maghemite (7'- hematite) and ulvospinel, the latter being a titaniferous variety. Table 1 illustrates the various spinel series as found by Goodenough and Loeb (1955). Ammma .nmoq can nmzocmpooov :3 z z z z oo H H :6 z z :6 z z 2 cm *H H :6 so 2 H z Himzw Hz H H 2 H z z 00 H z z 2 mm H E z z 2 as H H .H *H 2 .H z z z m: +~x +¢> +¢He +¢cm+vmo +maH +mmo +m.E +mwm A+mczV +muo +m> in:H +¢N no +mw .musuosuum Hmchm mem>cH paw HmEHo: ou Hmmmu H can 2 muwuuoa mnE .ucmocmm Imp onsumuwmamu mm mmuHm Hmnomsmuumu may CH mcoH usmam>HU mo Hones: 0:» “mmum>cH mamHHucm uoc mum mmeHmumm zuHB pwxnme mcoHpanHumHQ .UQOHHQmumm maamusmEHmexm no: :oHuanuumHU coHumo on» .UmHHoHcH mH musuosnum HmEHo: mmumoHocH AZVII.H mqméa As might be expected from the above, spinels are able to accept a wide variety of elements into their lattice without major structural variations. Gorter (1954) lists twenty-two different elements as being found in spinels (Table 2). TABLE 2.--Elements found in minerals with a spinel- type structure, listed by valences. After Gorter (1954) and Ringwood (1959). m Univalent: H, Li, Cu, Ag, Na Divalent: Mg, Ca, Mn, Fe, Co, Cu, Zn, Cd Trivalent: Al, Ti, V, Cr, Mn, Fe, Ga, Rh, In Tetravelent: Ti, V Mn, Ge, Sn, Mo, W The spinel structure was first determined by two independent workers, Bragg and Nishikawa, in 1915. Through x-ray diffraction methods, they found that the oxygen atoms were in approximate cubic closest packing (Fig. 1) and that the general formula was AB 04 (Bragg. 2 1915). The ions in position A were divalent and in four-fold coordination with oxygen, while those ions in the B positions were trivalent and in six-fold co- ordination (Fig. 2). Subsequently, Barth and Posnjak (1931) discovered a second spinel structural type with the general formula B(AB)O4. In this form, the A ion <:>..Oxygen O . .Aluminum \’60 O . .Magnesium A B Fig. l.—-Successive layers in the spinel structure showing the packing of the ions. (a) is the lower- most layer. After Bragg (1937). 0. .Oxygen @. .Metal in octahedral coordination o..Meta1 in tetrahedral coordination Fig. 2.—-Structure of spinel AB 0 . (a) is the lower- most layer. After Bragg (l937). is in octahedral coordination, as is one of the B ions, with oxygen, while the other B ion is in tetrahedral coordination. This lattice is said to have "variate atom equipoints” meaning that similar ions can be found in two different structural positions within the lattice. Table 1 shows that the majority of spinels are normal in structure and only the ferrites are commonly inverse. A great deal of work has been done in an at— tempt to explain the cause of the inverse structure which Goodenough and Loeb (1955) have approached from a theory of covalent bonding, Gorter (1954) has attacked from a consideration of the electron configuration of the ions, while Verwey and Heilman (1947) simply state that trivalent and tetravalent ions occupy the octahed- ral sites preferentially, with the exception of Fe, In, and Co which also occupy tetrahedral positions. Of the spinels, the ferrites have been studied most extensively, however in recent years interest has been shown in the entire series. In spite of this, little has been done concerning preferential composition of the minerals and it is hoped that the present work will contribute some knowledge of this area. THEORY The ability of one element to substitute for another is well known to mineralogists and petrolo- gists. If an element substitutes for another in the same structural position, the substitution is termed disdochic. Isomorphism is defined as the ability of two compounds of similar structural type to incorpor— ate more than 5 percent of each other in solid solu— tion (Rankama & Sahama, 1950; Wells, 1962). The spin- els are an isomorphous mineral group, with diadochic substitution occurring among the elements listed in Table 2. Goldschmidt (1937) was the first to offer a theory concerning geochemical distribution of the elements. On the assumption that ionic bonding is the dominant bond type in the majority of rock form- ing minerals, he postulated the following rules: 1. For two ions to substitute diadochically one for another in a crystal structure. the ionic radii must not differ by more than 15 percent. 2. When two ions possessing the same charge. but different radii, compete for a position in the crystal structure, the ion with the smaller radius is preferred. 3. Ions having similar radii, but different charges of the same sign may substitute diadochically in a crystal; in which case. the ion having the larger charge has pref— erence over the ion with the lesser charge. In 1951. Fyfe pointed out that mineral bonds are not completely ionic and suggested that the fol- lowing criteria be used when discussing diadochic substitutions: 1. Two ions will substitute for one another if their sizes are similar. 2. Two ions will substitute for one another if the number and directional prOperties of the bonds are similar. 3. Two ions will substitute for one another if the bond types are similar. Thus, although the first of Fyfe's rules is similar to Goldschmidt's, a consideration of the electron configuration and tendency of coordination is added. Ahrens (1953) uses the ionization potential of an element as a key to its distributional tenden- cies. He states that when two ions of the same val- ence and similar radii compete for a structural site. the one with the greater ionization potential should arrive first, but may not remain at that site. The ability to remain at a site depends on the bond sta- bility of the anion and the directional properties of the cation toward the anion. Ringwood (1955a,b) states: 1. Whenever diadochy in a crystal is possible between two elements possessing appreciably different electronegativities, the element with the smaller electronegativity is pref- erentially incorporated. 2. When diadochy occurs between elements pos- sessing similar electronegativities, the Goldschmidt Rules are usually applicable. Table 3 lists the elements used in this in- vestigation, their radii, ionization potentials and electronegativity. Note the correlation between the latter two qualities, from which one may conclude that Ringwood simply restates Ahrens' results. The elements chosen for this study show varying degrees of natural association and diadochy. With the ex- ception of magnesium, they are all members of the first transition series. Table 4 shows the elec- tron configuration of the elements considered. 10 TABLE 3.--All data except electronegativity after Electronegativity after Ringwood, 1955. Asterisk indicates radii in four-fold coordination; all others in six-fold. W Green, 1959. Ion Ionic .Ionization Electronegativity Radius Potential v. Cu .96 7.223 1.8 Cu .72 20.28 2.0 Mg .66 14.97 1.2 Ni .69 18.13 1.7 Zn .7l*/.74 17.89 1.7 Fe .74 16.24 1.6 Fe .64 - 108 Mn .80 15.70 - Mn .57*/.60 52.00? - TABLE 4.-—After Rankama & Sahama. figuration of the elements. 1950 Electron con- Shell 1 .2 .3 4 . 5 6 Group ls 2%2p 3s de 4skfi4d4f sjsfiSde 6pdef Element Mg 2 2 6 2 Mn 2 2 6 2 6 5 2 Fe 2 2 6 2 6 6 2 Ni 2 2 6 2 6 8 2 Cu 2 2 6 2 6 lo 1 Zn 2 2 6 2 6 10 2 ca 2 2 6 2 6 10 2 6 10 2 Hg 2 2 6 2 6 10 2 p 1014 2 6 10 2 11 Copper is classified as a chalcophile element. meaning that it occurs as a sulfide more commonly than as an oxide (lithOphile) or as a free metal (siderophile). Unlike the elements which directly precede it in the periodic table, copper oxide (CuO) forms a structure where the metal is in tetrahedral coordination with oxygen, while the monoxides of Mn, Fe, Co, and Ni have the NaCl type of structure where the metal is in octa- hedral coordination. Zinc, the other chalcophile ele- ment studied, forms simple oxides with a wurtzite struc- ture, where the metal is in four—fold coordination with oxygen. Iron and nickel are siderophile elements, while manganese and magnesium are classified as lithophile. Magnesium, however, often diadochically substitutes for iron, as does nickel; and both iron and nickel readily form sulfides and iron oxides. It does not seem likely that these classifications would indicate a preference in the present study. 12 Predictions According to Wells (1962, p472) all manganese hydroxides form Mn304 when heated above 10000 C. Since this is the upper limit of the experimental temperatures used in this investigation, one might expect to find this form of manganese as a product. This structure might also be expected to accommodate other elements; therefore a general formula might be AMn204- There are a number of possible products which could be derived from these reactants. Since hydrox— ides change to oxides when heated, it is possible that a mixture of simple ox1des might be found. A second possibility is that a complex oxide (ferrite) plus a simple oxide will be formed. These are the desired products. A third possibility might be a complex oxide of the general formula (A,B)Fe where A and B are 0 . 2 4 the two divalent elements present in the reactants. and two simple oxides, A0 and B0, composed of the ions which did not enter the ferrite lattice. Half of the elements used in these experiments have only one oxidation state and are divalent. Fe, 13 Cu, and Mn can have several valencies. Heslop and Robinson (1960) state that tenorite (CuO) goes to cuprite (Cu20) at 9000 C., but is stable at normal temperatures. Thus, at the upper limit of the tem— peratures involved, copper would not have a valence that would allow the formation of a copper ferrite. On the basis of ionic radii, the expected order of substitution is: Mg, Ni, Zn, Cu, Mn. Thus, no manganese ferrite should be formed except as a control, copper ferrite only when paired.with manganese, etc. If Ahrens' theory is used, the order is: Cu, Ni, Zn, Mn, Mg. It is easily seen that nickel and zinc retain their positions, while copper becomes more favored and mag- nesium less so° Thus nickel ferrite should be formed in preference to that of zinc and zinc ferrite in preference to that of manganese. This is in keeping with the findings of Wager and Mitchell (1951) in their studies of the Skaergaard intrusion. EXPERIMENTAL PROCEDURE There are several methods of synthesizing spinel-type minerals. Spiroff (1938) precipitated magnetite by slowly dripping ferric sulfate and fer- rous chloride into a weak ammonia solution. Posnjak (1930) formed zinc ferrite by heating equal parts of zinc carbonate and ferrous oxide together. Mason (1947) created both manganese and zinc ferrites by heating a precipated mixture of hydroxides. The lat- ter method is the one followed in this work. The original chemicals (Baker Reagent grade) were the sulfates of the desired elements. Since Mason was not specific in his description of technique. the individual sulfates were tested with each of the precipitating agents (NaOH and Na C03) to determine 2 any variation in reactions. There was none. The sulfates desired for a particular mixture were weighed on a rough balance to get approximate amounts and then were transferred to Chain-o—matic balance for accurate measurements. The quantitites l4 15 were calculated to give a ration of one mole of dival— ent ion to two moles of trivalent ion. The sulfates were then added to enough boiling deionized water to make a .25 N solution. The solution was stirred until all of the sulfate had dissolved, then 250 mls. of .5 N sodium carbonate solution and an equal amount of l N sodium hydroxide solution were added to precipitate the mixed hydroxides according to the following equa- tion: 2BSO + 2ASO + 2Na C0 + 4NaOH -e 2A(OH)2 + 4 4 2 3 2B(OH)2 + 4Na SO + 2C0 . The precipitate was usually 2 4 2 dark in color and curdy in appearance. The mixture was stirred to insure completing the reaction and allowed to settle. A few drops of sodium carbonate solution were added to check the completeness of the reaction. If complete, the beaker was set aside to cool, if the reaction was incomplete, more of the pre- cipitating agents were added, the mixture again stirred and tested. As soon as the precipitate had settled after a completion of the reaction, the residual fluid was decanted and the precipitate washed by adding deion— ized water and stirring the mixture, followed by a 16 period of settling and decanting the wash water. This procedure was repeated several times and then a test for sulfate was made in the following manner: Two milliliters of the wash water were placed in a test tube and tested for extremes of pH with litmus paper. If highly acid or base, it was neu- tralized by adding ammonium hydroxide or hydrochloric acid. Then .5 milliliters of 5N hydrochloric acid were added followed by .5 milliliters of barium chloride solution. If any sulfate ion was present. SO formed (See Mc- 2 4 Alpine and Soule, 1949, p. 215). a fine white precipitate of Ba When a negative sulfate test had been obtained. the precipitate was placed in a Bucher funnel and at least one liter of deionized water filtered through it. A final sulfate test was run and the solid placed in a covered beaker in a drying oven for a period of twelve hours. The dried solid was crushed and portions of not less than a gram were weighed out. These were placed in high alumina crucibles (Made by Norton and purchased from the Fisher Company) which were then placed in an electric furnace and heated to the desired 17 temperature. This temperature was held for twelve or sixteen hours after which the crucibles were re- moved and allowed to cool. Since the transition of spinels from a high temperature is slow, no attempt was made to quench the products other than to leave them at room temperature. The cooled sample was ground under acetone in an agate morter to a fine powder and mounted on a fine glass rod using clear nail polish as a mount- ing medium. This method was chosen over other pos- sibilities because it seemed the quickest, easiest. and had given good results in the past. The mounts were placed in a standard Norelco powder camera and exposed for 2.6 hours to iron radiation. The films were developed in standard fashion and measured on a standard Norelco film-measuring device. For a review of x—ray procedures, see Azaroff and Buerger (1958). No attempt was made to correct for film shrinkage. The d-spacings were calculated on the CDC 3600 computer using the program appearing in Appendix 1. This method was chosen primarily for its speed (one line took less than a minute to put 18 on a data card). The d-spacings were then compared with those recorded in the ASTM X-ray Powder Data File. DATA The x—ray data is presented as visually esti- mated line intensities and the calculated d-spacings for each film. As each mixture was heated at four different periods, there are four films in a set and these are grouped with the control binary ferrites that have been treated in a similar fashion. Thus in Table 5, Set 5.1, the data in the left hand column is that of the control copper ferrite, that in the middle column is the ternary sample (Cu, Mg) and that in the right hand column is data for magnesioferrite all heated at 10000 C. for 16 hours. In correlating lines of each film, allowances had to be made for errors in measuring and in estimat- ing intensities. Therefore, if two lines had similar intensities and differed by .01 Angstroms they were correlated. In some cases, the trimetallic line fell between the dimetallic controls in value, so there might be .02 difference from the high line to the low 19 20 value. In those cases where the trimetallic line was within .005 of one dimetallic control and .009 of the other, the correlation was with the closest lines. As expected, the lines recorded in the ASTM powder data files for copper ferrite do not correlate with those found in the experimental control sample in Set 5.1. However, the experimental lines do not match those recorded for the expected copper oxide (CuZO) either, but there is a close correlation with the ASTM lines for magnetite. With the exception of metallic copper, none of the lines listed for possible copper products in the ASTM files can be found in Set 5.1. The line superscripted l, is the most intense line of metallic copper. Spectrochemical analysis shows that copper is indeed present, so the only con- clusion is that copper exists as a metal, rather than an oxide under the conditions found in this experiment. The copper-iron product of the control sample in Set 5.2 is similar to that found in the preceding set, but Set 5.3 has lines corresponding to the ASTM copper ferrite lines. In Set 5.4, some hematite 21 lines are found as well as some for tenorite (CuO) anc copper ferrite. The magnesioferrite sample in Set 5.1 has close correlation with the lines in the ASTM file. There is also a good match for magnetite, but since there is no evidence of metallic magnesium, it is felt that magnesioferrite is formed. A similar set of lines is found in Set 5.2 while at the lower tem- peratures of sets 5.3 and 5.4, some hematite lines and one or two weak lines which could be MgO are also found. The Cu, Mg sample in Set 5.1 has lines which have a strong correlation with both control samples. with the exception of the third line which is super- scripted 4. Although the d-spacing is .02 Angstroms shorter than that of the strongest line for Cu 0 (cu- 2 prite), this line is considered as indicating the presence of this mineral. This conclusion is sup- ported by the presence of lines which are definitely cuprite in Set 5.2. In sets 5.3 and 5.4, the copper oxide is tenortie as expected. Hematite is found in Set 5.4 in addition. The presence of magnesioferrite is found in all samples. 22. TABLE 5.—-Data from the ternary sample Cu,Mg,Fe compared with the binary samples Cu,Fe and Mg,Fe. 'ndicates metallic cop- per, 2hemetite, 3magnesium oxide, ‘tuprite. and Stenorite. CuJFe Cu,Mg,Fe Mg,Fe CuyFe Cuqu,Fe Mg,Fe I d I d I d I d I d I d 5 4.833 5 4.823 20 2.962 5 4.762 20 2.944 40 2.933 10 2.957 100 2.527 100 2.518 100 2.510 20 2.942 60 2.949 5 2.4514 100 2.515 100 2.526 100 2.518 3 2.424 10 2.4644 20 2.0971 20 2.088 40 2.084 3 2.411 2 2.418 10 1.716 10 1.708 30 1.708 3 2.1334 60 1.616 80 1.612 80 1.611 20 2.0851 20 2.096 50 2.092 60 1.486 90 1.482 90 1.480 10 1.708 10 1.714 20 1.710 80 1.611 80 1.616 80 1.614 3 1.5104 90 1.481 90 1.486 90 1.483 5 1.416 Set 5.1 16 hrs @ 10000 c. Set 5.2 12 hrs @ 10000 c. 5 4.807 3 4.007 3 4.818 15 4.762 3 4.732 20 2.958 30 2.947 30 2.951 3 3.659 3 2.7 3 3.599 3 2.683 3 2.953 10 2.945 20 2.936 20 2.554 40 2.684 5 2.6802 100 2.5135 100 2.519 100 2.5115 100 2.5155 100 2.5052 100 2.4933 3 2.415 10 2.410 3 2.421 40 2.3175 20 2.3115 5 2.302 30 2.3065 3 2.271 5 2.1352 10 2.202 30 2.092 30 2.093 20 2.0913 30 2.0863 20 2.067 10 2.056 10 1.861 3 1.891 3 1.860 5 1.862 3 1.731 15 1.838 5 1.835 20 1.712 5 1.702 5 1.708 10 1.699 3 1.682 30 1.691 5 1.6902 5 1.646 20 1.611 30 1.609 20 1.612 70 1.612 50 1.613 3 1.505 20 1.595 40 1. 7 95 1.481 100 1.480 5 1.503 40 1.454 5 1.452 100 1.4903 90 1.483 100 1.484 20 1.465 3 1.454 Set 5.3 12 hrs @ 8000 c. Set 5.4 12 hrs @_600°c. 23 The manganese ferrite in Set 6.1 appears to be magnetite with no evidence of manganese. Lines for the expected hausmannite (Mn 04) are not present, neither 3 are lines for any of the manganese oxides nor metallic manganese. Some manganese products have strong lines with d-spacings Similar to those of magnetite, but the correlation is not line for line, and there are no ex- traneous lines present in the set under consideration. Spectro—chemical analysis shows the presence of mangan- ese so one might conclude that the element is present in either an amorphous state or has substituted for iron in some fashion. In subsequent manganese-iron samples, the prod- uct is bixbyite (Mn,Fe O3) accompanied by hematite in 2 the lower temperature ranges. The lines of the Cu,Mn sample in Set 6.1 correlate well with those of the Manganese control. There is however no evidence of either copper or manganese products even though analy- sis shows that both elements are present. In set 6.2, there is no correlation with either of the controls, nor with any of the ASTM lines for the expected oxides. The lines are slightly low for magnetitie, but correlate to some degree with those for maghemite. Set 6.3 has 24 some lines which are correlatable with those recorded for copper ferrite, but there is not enough evidence to warrant calling this product copper ferrite. There is a similar situation in Set 6.4, with the exception of some lines which are definitely those of hematite. Some of the lines could be due to gamma manganese oxide. but there is no conclusive evidence for this. 25 TABLE 6.—-Data from the ternary sample Cu, Mn, Fe compared with the binary sample Cu, Fe and Mn, Fe. Cu,Fe Cu,Mn,Fe Mn,Fe Cu,Fe Cu,Mn,Fe Mn,Fe I d I d I d I d I d I d 20 4.869 10 4.787 5 4.833 5 4.762 20 2.983 10 2.972 20 3.824 20 2.962 20 3.656 100 2.546 100 2.538 20 2.958 100 2.527 20 2.942 3 2.424 5 2.438 100 2.703 20 2.112 30 2.109 100 2.531 20 2.097 100 2.515 30 2.509 10 1.729 20 1.727 10 2.424 10 1.716 3 2.411 70 1.630 100 1.627 30 2.347 60 1.616 20 2.205 70 1.498 100 1.497 45 2.103 60 1.486 20 2.085 30 2.004 60 1.842 30 1.723 10 1.708 40 1.693 80 1.664 85 1.643 80 1.611 5 1.526 90 1.481 90 1.493 5 1.488 10 l 453 5 1.416 60 1.424 Set 6.1 16 hrs e 10000 c. Set 6.2 10000 c. 15 4.762 3 4 807 7 3.8 5 4.747 |I|IIIH'f_________I""""""""""IllllIlIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 2 if] 1 TABLE 6 -—Data from the ternary sample Cu, Mn, Fe compared With the binary sample Cu, Fe and Mn, Fe. Csife . Cu.MnhEe -Mnttsi i Miéurtsmi. . Cu,MULEsi_iMQIEsMIII ' - ii .......u T’ _--.-_-600° c. 35 The apparent products of Mg,Zn sample in Set 11.1 are zinc ferrite and magnesium oxide. Similar lines are found in the subequent sets with good cor- relation between the ternary samples and zinc ferrite maintained throughout. This is somewhat different than the results of the Mg,Ni sets, something which might not be expected in view of similarities between the zinc ferrite and nickel ferrite controls. Zinc must be much more acceptable to all temperatures to the ferrite lattice than nickel. 36 TABLE ll.--Data of the ternary sample Mg,Zn,Fe compared with the binary samples Mg,Fe and Zn,Fe. indicates MgO.* - m L r , Mg,Fe Mg,Zn,Fe Zn,Fe Mg,Fe Mg,Zn,Fe Zn,Fe I d I d I d I d I d I d 5 2.958 7 2.953 60 2.949 30 2.955 20 2.960 40 2.933 100 2.527 100 2.526 80 2.523 80 2.526 100 2.518 100 2.510 1 5 2.421 1 2 2.435 50 2.092 30 2.07 10 2.101 20 2.099 5 2.098 20 1.710 30 1.718 10 1.720 40 2.084 80 1.622 10 1.719 10 1.717 80 1.614 80 1.612 30 1.708 90 1.483 90 1.4881 90 1.491 80 1.621 80 1.619 80 1.611 . 5 1.500 90 1.480 100 1.4891 100 1.488 Set 11.1 16 hrs @ 1000° c. Set 11.2 12 hrs @ 1000° c. 5 4.807 3 4.818 30 2.951 40 2.953 20 2.958 5 4.792 50 2.799 3 4.732 3 2.772 3 3.599 3 2.683 50 2.958 20 2.964 100 2.527 100 2.527 20 2.936 100 2.519 10 2.790 5 2.470 5 2.680 3 2.421 5 2.428 10 2.420 100 2.523 100 2.534 30 2.093 40 2.1011 100 2.505 3 1.891 10 2.461 20 1.712 20 1.718 10 1.719 5 2.435 3 1.682 3 2.271 80 1.621 80 1.621 40 2.102 10 2.106 50 1.613 30 2.086 100 1.484 90 1.489 90 1.489 5 1.905 3 1.454 5 1.835 5 1.785 5 1.708 10 1.715 10 1.720 5 1.690 30 1.609 80 1.617 80 1.622 10 1.497 100 1.480 95 1.487 90 1.492 5 1.452 Set 11.3 12 hrs @ 800° c. Set 11.4 12 hrs @ 600° c. 37 The Mn,Ni products are apparently maghemite and nickel oxide in Set 12.1. The d-spacings are somewhat low for magnetite but match ASTM lines for maghemite with the exception of 2.406 which can be found in the recorded lines for nickel oxide. In Set 12.2 however there is good correlation between the ternary sample and the nickel ferrite control. Again, there is no sign of manganese and it must be assumed that it is either substituting for iron in some way or is present in an amorphous form as spec— trochemical analysis indicates that it is present. In Set 12.3, there is a good match with the ASTM file and experimental data for nickel ferrite with a possibility of gamma manganese dioxide. The increased intensity of the line 2.090 might be at- tributed to the latter substance. This is extremely tenuous as the other intense lines for this product are not evident. Set 12.4 also shows an increased intensity for a line in this area, which might be considered support for the hypothesis. In addition, a strong line of hematite is found. !llEIlI!IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 38 TABLE 12.-—Data of the ternary sample Mn, N, Fe compared with the binary samples Mn, Fe, and Ni, Fe. lindicates NiO, 2indicates hemetite, and 3indicates gamme MnOZ. Mn,Fe Mn,Ni,Fe NiLFe :Mh,Fe MnLNi,Fe Ni,Fe I d I d I d i I d I d I d 3 4.767 5 4.797 10 2.972 20 3.824 15 2.927 20 3.656 5 2.891 30 2.966 100 2.538 10 2.925 100 2.510 100 2.703 60 2.491 30 2.509 80 2.505 100 2.506 5 2.4063 5 2.398 30 2.109 1 30 2.347 20 2.0903 20 2.205 10 2.072 20 2.085 30 2.079 20 1.727 30 2.005 10 1.708 10 1.697 3 1.997 100 1.627 60 1.842 70 1.611 60 1.601 20 1.706 10 1.700 100 1.497 1 40 1.693 80 1.481 80 1.664 100 1.472 80 1.610 10 1.603 3 1.416 5 1.526 5 1.488 100 1.481 80 1.474 10 1.453 60 1.424 5 1.415 5 1 408 Set 12.1 16 hrs @ 1000° c. Set 12.2 12 hrs @ 1000° 0 10 4.747 7 4.777 7 3. 3 3.783 10 3.633 10 3.633 30 2.936 15 2.944 5 2.918 5 2.927 100 2.697 100 2.689 3 2.6912 100 2.515 100 2.515 10 2.500 90 2.493 50 2.498 40 2.492 10 2.4043 10 2.410 5 2.394 3 2.396 10 2.340 3 2.335 10 2.201 20 2.195 50 2.0903 40 2.0863 10 2.076 10 2.071 10 1.999 3 1.999 20 1.839 50 1.835 5 1.715 20 1.708 20 1.707 50 1.690 5 1.694 10 1.689 7 1.697 40 1.659 80 1.610 80 1.600 80 1.612 1.480 100 100 1.480 3 1.418 3 1. 12 hrs @ 800° c. 12 hrs @ 600° c. ( //”. 39 Although the d-spacings in sets 13.1, 13.2, and 13.3 correlate well with those of zinc ferrite. they also correlate with lines recorded for magnetite, and manganese iron zinc oxide. For this can be concluded concerning the order of of zinc and manganese. In Set 13.4, one lines of hematite amongst those of "zinc This does not allow any conclusion to be cerning the question at hand however. reason, little acceptance also finds ferrite." drawn con- qj 40 TABLE l3.——Data of ternary sample Mn,Zn,Fe compared with the binary samples Mn,Fe, and Zn,Fe. indicates hematite lines. Mn,Fe Mn,Zn,Fe Zn,Fe Mn,Fe Mn,Zn,Fe Zn,Fe I d I d I d I d I d I d 210 2.972 20 3.824 10 2.960 7 2.953 20 3.656 100 2.538 100 2.531 80 2.536 20 2.975 5 2.429 2 2.435 20 2.960 30 2.109 10 2.103 5 2.098 100 2.703 20 1.721 10 1.721 10 1.717 100 2.539 100 1.627 80 1.624 80 1.619 100 2.526 100 1.497 90 1.493 100 1.488 30 2.509 5 2.441 5 2.421 1 30 2.347 * 20 2.205 S 10 2.113 10 2.101 i 30 2.004 g 60 1.842 g 10 1.726 10 1.720 40 1.693 80 1.664 70 1.627 80 1.622 5 1.526 5 1.488 80 1.495 90 1.491 10 1.453 60 1.424 Set 13.1 16 hrs @ 1000° c. ,Set 13.2 12 hrs @ 1000° c 7 3.8 % 5 4.828 3 4.816 10 3.633 g 3 3.783 20 2 968 20 2.958 g 5 3.659 100 2.697 5 10 3.633 100 2.530 100 2.527 75 2.962 20 2.964 10 2.500 3 2.795 5 2.430 10 2.420 100 2.689 10 2.340 25 2.679 10 2.201 100 2.531 100 2.534 10 2.110 3 2.335 10 1.999 20 2.195 5 2.195 20 1.839 3 1.999 25 2.104 10 2.106 5 1.715 10 1.724 10 1.719 3 1.999 10 1.689 50 1.835 10 1.839 80 1.659 25 1.721 10 1.720 70 1.627 80 1.621 50 1.690 15 1.6911 7 1.525 3 1.659 20 1.486 90 1.494 90 1.489 75 1.623 80 1.622 50 1.453 3 1.597 ‘3 1.522 40 1.485 75 1.489 90 1.492 1.452 12 hrs 8 800° c. 13.4 12 hrs 8 600° c. 41 The Ni,Zn products are zinc ferrite and nickel oxide in Set 14.1. The match between the ASTM files and nickel oxide is not without question, but the ob- servation that the ternary samples are closer to the zinc ferrite control lends credence to the above con- clusions. These products are found in all subsequent sets and are unaccompanied by hematite in Set 14.4. This is to be expected since neither of the control samples were found to have hematite. 42 TABLE 14.-—Data of the ternary sample Ni,Zn,Fe compared with the binary samples Ni,Fe and Zn,Fe. indicates NHL 2indicates metallite zinc. = l m Ni,Fe Ni,Zn,Fe Zn,Fe Ni,Fe NiLZn,Fe Zn,Fe I d I d I d I d I d I d 7 2.953 5 4.797 5 4.797 20 2.936 30 2.966 5 2.891 20 2.951 20 2.960 80 2.526 100 2.522 100 2.526 100 2.513 100 2.506 1 60 2.491 20 2.434 20 2.4281 2 2.435 5 2.421 70 2.108 5 2.098 5 2.398 20 2.0891 40 2.110 10 2.072 30 2.0941 10 2.101 70 1.709 10 1.717 30 2.079 10 1.691 3 1.997 80 1.615 80 1.619 10 1.714 10 1.720 60 1.601 10 1.700 90 1.494 100 1.488 75 1.617 80 1.622 100 1.472 100 1.4841 70 1.603 50 1.497 90 1.491 80 1.474 80 1.4841 5 1.408 Set 14.1 16 hrs @ 1000° c. Set 14.2 12 hrs @ 1000° c. 20 2.958 3 4.818 10 2.938 2 4.650 5 2.918 20 2.968 20 2.964 100 2.527 5 2.927 90 2.513 2 2.854 90 2.493 100 2.526 100 2.534 5 2.413 10 2.420 40 2.492 5 2.394 5 2.421 5 2.435 70 2.0902 3 2.396 10 2.076 30 2.0962 10 2.106 7 1.697 10 1.709 10 1.719 10 2.071 80 1.600 80 1.612 80 1.621 5 1.714 10 1.720 100 1.472 100 1.4822 90 1.489 5 1.694 30 1.615 80 1.622 30 1.598 100 1.471 90 1.483 80 1.492 Set 14.3 12 hrs @ 800° c. Set 14.4 12 hrs @ 600° c. 43 TABLE 15.--D-spacings and intensities of the three most intense lines of compounds found in this study. After the ASTM Index to the X-ray Powder Data File (1959). Compound D-spacings IntenSities Copper 2.09 1.81 1.28 100 46 20 Copper iron oxide 2.49 1.49 2.60 100 100 50 Copper oxide 2.52 2.32 2.53 100 96 49 Copper oxide 2.47 2.14 1.51 100 37 27 Alpha iron oxide 2.69 2.51 1.69 100 80 8O Magnetite 2.53 1.48 2.97 100 70 60 Gamma iron oxide 2.52 1.48 2.95 100 53 34 Magnesium 2.45 2.61 2.78 100 41 35 Magnesium ferrite 2.52 1.48 1.61 100 90 70 Magnesium oxide 2.11 1.49 0.94 100 52 17 Manganese ferrite 2.56 2.12 1.64 100 60 60 Bixbyite 2.72 1.66 1.42 100 90 80 Gamma manganese oxide 2.48 2.74 3.08 100 70 60 Hausmannite 2.47 2.74 1.54 100 63 50 Manganese oxide 2.72 1.66 1.42 100 90 60 Manganese iron zinc oxide 2.55 1.49 1.62 100 80 70 Nickel ferrite 2.51 1.48 1.60 100 53 33 Nickel oxide 2.09 2.41 1.48 100 91 57 Zinc 2.09 2.47 2.31 100 53 40 Zinc ferrite 2.53 1.49 1.62 100 80 70 Zinc oxide 2.48 2.82 2.60 100 71 56 DUSCUSSION OF DATA The general order of acceptance into the ferrite structure seems to be Zn, Mg, Ni, Cu, Mn, especially at temperatures of 10000 C. At lower tem- peratures, Nickel and magnesium change places as dis- cussed above. This sequence is somewhat different than that predicted on the basis of ionic radii and significantly different from that predicted according to Ahrens' theory. Looking at Table 4, it is observed that both zinc and magnesium have full orbitals, while the re- maining four elements have but partially filled d- orbitals. Notice also that nickel is the next element below zinc to have a full s-orbiatal. Thus it appears that electron configuration is the controlling factor in the acceptance of these elements. These results generally disprove Ringwood's theory (1955) as zinc has a somewhat larger electro- negativity than that of magnesium and has a somewhat 44 45 larger ionic radius than the other ions. This latter quantity is not too significant as all the radii of the ions under examination lie within 15 percent of the radius of zinc, thereby conforming to the Goldschmidt rule. One possible answer for zinc's preference is that the normal configuration of the bonds about the ion is a tetrahedron which might allow the formation of a ferrite lattice with less difficulty than shift- ing an iron ion into the tetrahedral position. The samples containing manganese are somewhat of an anomally since they did not form the expected ferrite products. This can perhaps be answered by noting that manganese can substitute readily for any of the other elements due to the variable valence. Palache §t_§1. (1944) have several analyses which show that manganese is commonly present with other elements in a ferrite. Mason (1947) found that at 11000 C., a cubic product was obtained with equal amounts of iron and manganese, while at 10000 C., a similar product with a somewhat smaller lattice dimension was obtained with iron, manganese, and zinc in a ratio of about 2.5-2.5—1.0. With this in mind, 46 there seems little doubt that with a ratio of 2-1-1, similar substitutions could occur. The formation of ferrites in these experiments shows that minerals can be formed at temperatures well below melting points recorded for the mineral. One must be careful however not to forget that these are closed systems and that geologic processes are open systems. Zinc, nickel and iron form common sulfides as does copper, and one should not conclude that fer- rites will be formed in preference to sulfides from the data presented above. Temperatures and pressures found in the earth will no doubt effect the conclu- sions stated above. CONCLUSIONS Ferrites can be formed at normal pressures and at temperatures of 10000 C. to as low as 6000 C. in an oxidizing atmosphere. The substitution of the ferrites will be in the following order: magnem \AM Zinc, manganese, nickel, copper, manganese. The prime factor controlling this order seems to be the electron configuration of the elements. 47 SUGGESTIONS FOR FUTURE WORK More work should be done with the manganese spinels, since they were the most enigmatic of those studied. It is suggested that accurate quantities be combined and the point at which two difinite pro— ducts are first formed located. This would be most easily accomplished using a ternary system so that quantity versus crystal form could be plotted (see Mason, 1947). It is also suggested that cadmium and mercury be used in experiements of this type since they are the next elements with full d-shells. Observation of acceptability into the lattice with respect to zinc and mangensium. Another approach might be to heat the samples under a non-oxidizing atmosphere and/or under pressures of more than one atmosphere to observe any change in the acceptability table found above. 48 49 Magnetic susceptability should be measured to observe variation with composition. In any future work it is suggested that the hydroxides be synthesized by using chlorides of the desired elements rather than sulfates as the reaction AC12+2NaOH-——§ A(OH)2+2NaCl gives a water soluble and easily removed by-product rather than the sometimes difficult sulfate. BIBLIOGRAPHY Ahrens, L. H., 1953, The Use of Ionization Potentials, part 2 Anion affini tv and qeochemigtrv, Geo- chemica et Cosmochimica Acta, v.3, pp. 1-29. Azaroff, V. & M. J. Buerger, 1958, The Powder Method in X—ray¥Crystalloqraphv, McGraw—Hill Book Company, Inc., N. Y. Barth, T. F. W. & E. Posnjak, 1931, The Spinel Struc- ture: an example of Variateggtom gguipointg, Journal of the Washington Academy of Science. v. 21, p. 255. Bowen, N. L., 1928, The Evolution of the aneo g Rocks, Dover Publications, Inc., 1956, by special permission of the Princeton University Press. Bragg, W. H., 1915, The Structure of Magnetite and the Spinels, Nature, V. 95, p. 561. Bragg, W. L., 1937, The Atomic Structure of Minerals. Cornell University Press, Ithaca, N. Y. Deer, W. A., Ra. A. Howie, and J. Zussman, 1962, Rock- forminq Minerals v. 5, The Non-silicates, John Wiley and Sons, Inc., New York, N. Y. Fyfe, W. S., 1951, Isomorphism and Bond Type, American Mineralogist, v. 36, p. 538. Goldschmidt, V. M., 1937, The Principles of Distribu- tion of Chemical Elements in Minerals and Rocks, Journa1 of the Chemical Society, Lon- don, p. 655. 50 51 Goodenough, J. B. and A. L. Loeb, 1955, Theory of Ionic Ordering, Crystal Distortion and Mag- netic Exchange Due to Covalent Forces in Spinels, Physical Review, v. 98, p. 391. Gorter, E. W., 1954, Saturation Magnetization and Crystal Chemistry of Ferrimagnetic Oxides. Phillips Research Reports, v. 9, pp. 295-320. Green, Jack, 1959, Geochemical Table of the Elements for 1959, Bulletin, Geological Society of America, v. 70. Helep, R. B. and P. L. Robinson, 1960, Inorganic Chemistry, Elsevier Publishing Company. Mason, Brian, 1947, The System Fe 0 -Mn 0 -ZnMn204 -ZnFe O , American Minera oqist, v. 32, pp. 426-441? McAlpine, Roy K. and Byron A. Soule, 1949, Fundamen- talsiof Qpalitative Chemcial Analysis, D. Van Nostrand Company, Inc., Princeton, N. J. Palache, Charles, Harry Berman, and Clifford Frondel, 1944, The System of Mineralogy of James Dwight Dana and Edward Aslispury Dana, Seventh edition, v. 1, John Wiley and Sons, Inc., New York. Posnjak, E., 1930, American Journal of Science, v. 19, p. 67. Rankama, K. and Th. G. Sahama, 1950, Geochemistry, The University of Chicago Press, Chicago, Ill. Ringwood, A. E., 1955, Principles Governing Trace Ele- ment Distrution During Magmatic Crystalization, Part I, Geochimiqg_et Cosmochimica Acta, v. 7, p. 189. , 1955. Principles Governing Trace Element Distribution During Magmatic Crystallization, Part II, Geochimica et Cosmochimica Acta, v. 7. p. 242. 52 . 1959. The Constitution of the Mantle--III Consequences of the Olivine-spinel Transition, Geochimica et Cosmochimica Acta, v. 15, pp. 195—212. Smith, Joseph V. (ed), 1959, Index to the X-ray Powder Date File, American Society for Testing Mater- ials, Philadelphia, Pa. . 1959. X-ray Powder Date File, American Society for Testing Materials, Philadelphia, Pa. Spiroff, Kiril, 1938, Magneitite Crystals from Mete- oric Solutions, Economic Geology, v. 33, p. 818. Verwey, E. J. W. and E. L. Heilmann, 1947, Physical Properties and Cation Arrangement of Oxides with Spinel Structures, The Journal of Chemi- cal Physics, v. 15, pp. 174-180. Wager, L. R., and R. L. Mitchell, 1951, The Distribu- tion of Trace Elements During Strong Fraction— ation of Basic Magma--A Further Study of the Skaergaard Intrustion, East Greenland, Geo- chimica et Cosmochimica Acta, v. 1, pp. 129- 208. Washington, H. S., and W. Clarke, 1924, The Composi- tion of the Earth's Crust, U. S. Geological Survey, Professional Paper 127. Wells, A. F., 1962, Structural Inorganic Chemistry, Third Edition, The Clarendon Press, Oxford. England. APPENDIX I Program to calculate d-spacings of minerals from their powder x-ray films. Adapted from Azaroff and Buerger. PROGRAM INDEX 10 READ ll,X2,X1,NUM,LINE,I ll FORMAT (2F10.0,A5,215) 12 A = x2 + x1 s = (x2 - x1) * 10, B = (S/4,) * .01745 D = .96865 / SINF (B) Q = 1./(D*D) PRINT 12, NUM,LINE,I,X2,X1,A,S,D,Q FORMAT (A7, 217, 6E15.8) GO TO 10 END END DATA A is a check on the accuracy of measurement of the lines. If done correctly, the column in the print out should not differ except in the third decimal place. S gives the arc length, B the angle of re- fraction, D the d-spacing. 53 "I7'711111111111111I