Quantitative X-ray diffraction analysis of oxides formed on superalloys NASA Technical Memorandum I N i ; A i. CM. I Xv txv. <«• 2: X-RAY DIFFRACTION ANALYSIS^ >: OXIDE FORMEb o? NA1 5A AND ADMiHfSTIATlON • WASHIN6ION, Ore. * 1| 1. Report No. 2. Government Accession No. 3. Recipient's Catalog No. NASA TM X-2627 4. Title and Subtitle 5. Report Date QUANTITATIVE X-RAY DIFFRACTION ANALYSIS OF OXIDES September 1972 FORMED ON SUPERALLOYS 6. Performing Organization Code 7. Author(s) 8. Performing Organization Report No. Ralph G. Garlick E-6911 10. Work Unit No. 9. Performing Organization Name and Address 134-03 Lewis Research Center 11. Contract or Grant No. National Aeronautics and Space Administration Cleveland, Ohio 44135 13. Type of Report and Period Covered 12. Sponsoring Agency Name and Address Technical Memorandum National Aeronautics and Space Administration 14. Sponsoring Agency Code Washington, D. C. 20546 15. Supplementary Notes 16. Abstract K«X O ^ ^ The ratio of the intensity of a particular line for component 1 to a particular line for component 2 then is 11 K., p9 X1 X., -L = _L _£ _J = (constant) —1 1 2 K2 pl X2 X 2 that is, the ratio of the intensities of any two lines of the mixture is independent of the mass absorbtion of the mixture. Furthermore, since , n = (constant)(X0) _ T X if the weight fraction of phase 2 (X2) is held constant, then the weight fraction of component 1 is linearly proportional to the intensity ratio. Weight fraction X2 can be kept constant if phase 2 is an internal standard added to each sample in a constant proportion. For an actual system C is determined by plotting IiAo f°r synthetic mixtures of various weight fractions of phase 1. This constant can then be used to determine the weight fraction of phase 1 in any matrix. Quantitative analysis us- ing these principles has been effectively used by other investigators. For example, Copeland and Bragg determined calcium hydroxide in a matrix of hydrated calcium sil- icates using magnesium hydroxide as an internal standard (ref. 3). The peak intensity of a reflection from a phase is effected by many factors other than the weight fraction of the phase and absorbtion. Using an internal standard compen- sates only for absorbtion differences. To make the analysis accurate, the other factors must be either minimized or controlled. The factors, both instrumental and sample, are described in reference 4. To minimize sample factors, for example, sample particles should be small enough to make extinction and microbsorbtion effects negligible, and crystallites should be randomly oriented. Also, stressed particles should not be used. In this study the nature of the materials was such that stress and particle size were not problems. Preferred orientation is usually found when samples contain crystallites whose shape is acicular or tabular and the sample is compacted in the specimen holder. In this study the powders were not compacted, and in most cases the crystals were such that preferred orientation was not a problem. Methods were examined for using appropriate internal standards to determine weight fractions of aluminum oxide, nickel oxide, chromium oxide, cobalt oxide, cobalt chro- mite, and cobalt tungstate (A12O3, NiO, Cr 2 O 3 , CoO, CoCr 2 O 4 , and CoWO4). These are some of the most common phases found as products of oxidation on superalloys. Materials and Procedure Materials used as internal standards were chosen that had strong reflections well resolved from those of the oxides to be analyzed. The weight fractions of the internal standards used were chosen so intensity ratios expected for actual oxide samples could be most accurately measured. One reflection was chosen for each material that did not overlap any reflections from other materials to be combined with it. Three different internal standard materials were used for the determination of the oxides: silicon (Si) for AlgOg, Cr2Og and NiO, nickel (Ni) also for A^Og, Cr2O,, and NiO; and NiO for CoO, Cr2O,, CoCr 2 O 4 , and CoWO^. Materials were all acquired in reagent-grade powder form. X-ray diffraction showed all to be single phase except CoWO4 and CoCr^Ch. The CoWO4 contained a small amount of calcium tungstate (CaWO4). The CoCr2O4 was found to be chiefly Cr2O3 and CoO. This powder was heated at 2000° F (1366 K) for 91 hours to induce more complete formation of CoCr2O4. After treatment the powder was CoCr2O4 plus a small amount of Cr 2 Og. Table I(a) shows the reflections used to determine the calibration curves for AlJO^, NiO, and Cr 2 Oo using a 0. 25 weight fraction Si standard. Table n shows the mixtures used in these determinations. These combinations were mixed for 1 hour in a jar mill. The powder was also mixed immediately before diffraction by vibration in a vial for 4 minutes. Some of the loose powder was then dumped into a rectangular cavity 0. 3- millimeter (0. 020-in.) deep in a flat specimen holder. Excess powder was then removed with a spatula. After diffraction, the sample was returned to the vial and the process re- peated. In this way the intensities were determined 10 times for each mixture. For many of the mixtures both peak height intensity and integrated intensity were determined. A planimeter was used to determine integrated intensity. Integrated intensity gave ratios similar to those from peak heights. Peak height was therefore used in further measure- ments. Calibration curves were similarly determined for the same three oxides (Al2Oq, Cr 2 Og and NiO) using 0. 1 weight fraction nickel as the internal standard. The reflections used are shown in table I(b). The mixtures of oxides used in this case were 100 percent of each of the oxides and the possible 50-50 percent mixtures. A prototype sample con- taining known amounts of the three oxides was mixed to estimate the application of cali- bration curves made from a mixture of one or two of the oxides to a mixture of multiple phases, including phases not present in the calibration samples. The mixture was 30.0 percent NiO, 24. 0 percent AlgOg, 10. 0 percent Cr 2 O 3 , 16. 0 percent SiO2, 17. 6 percent WC, and 2.4 percent Co. Relatively large amounts of the unknown must be used for the loose powder methods described above. For example, about 0. 5 gram of NiO are needed to fill the sample holder cavity. For some oxidation tests, very small amounts of spall are collected, and it may be useful to know the proportions of the oxides in these. Therefore, another method was in- vestigated that allows for a smaller sample. This method of sample preparation is sim- ilar to that described by Munch and Pierron (ref. 5). In this method the sample (0. 2 g for the calibration samples) is put into a small mortar with enough mineral oil so that mulling yields a smooth paste. This paste is then scraped from the mortar onto a section of a glass slide and spread evenly. (Specimens as small as 0.05 g have been used.) Dry- ing the sample for 1 hour was enough so that it retained its shape and the individual par - tides apparently did not re-orient themselves with further time. This method was used for a 0. 1 weight fraction NiO internal standard for the following oxides: CoO, CoCrgO^, Cr 2 O 3 , and CoWO,. The reflections used for this analysis are given in table I(c). For these calibrations no mixtures of oxides were used, only 100 percent of each plus 10 per- cent NiO. Intensities were measured at least five times for each oxide. A prototype sample of these oxides was then made, the mixture being 5. 5 percent Cr^Oo, 39. 5 per- cent CoCr 2 O 4 , 40 percent CoO, and 15 percent CoWO,. This mixture was then used to estimate the accuracy of the method. RESULTS AND DISCUSSION Silicon Internal Standard Loose Powder Samples The calibration curves for NiO, AlnOg and CrgOo using a Si internal standard are shown in figures 1 to 3. The spread of values shown represents all data points (10 points at 1. 0 weight fraction, 20 points at other weight fractions). The curves, which were forced through the origin, can be represented by the equations: I A19O,(104) = (0.726±0.068) I Cr 9 O Q (104) ^ - (1.61±0.14) X, The errors reported represent one standard deviation calculated using the computer techniques described in reference 6. The pooled relative standard deviations were about 10. 2 percent of the weight fraction for NiO, 9. 4 percent of the weight fraction for A1dt0O,, O and 8. 7 percent of the weight fraction for Cr^O*. Since these errors are relatively high and similar for the three oxides, it was thought that they may be due to variations in the intensity of the Si standard. To check this, a sample of Si powder alone was placed in the holder, and the peak height measured at least 10 times for each of the following scan- ning speeds: 0. 2°2#/min, 0.4°26>/min and 2°20/min. Considerable variation in the peak height was found, the lowest relative standard deviation (RSD) being 4.5 percent. The cause of this variation is not obvious, since from counting statistics we do not expect this error, and since the variation was random, not either increasing or decreasing with time as would be expected due to an orientation change during settling of the powder. Nickel Internal Standard-Loose Powder Samples The calibration curves for NiO, AUOo, and Cr^Oo using a Ni internal standard are shown in figures 4 to 6. The curves, which were also forced through the origin, can be represented by the equatiqns: = (3.63±0.05) T = (1.26±0.03) X A1 0 £\.Ln\Jt} =(2.75±0.06)X These standard deviations are much lower than for the Si internal standard, the highest being 2.4 percent for the C^O,. The averaged concentrations of 10 replicate determinations of the prototype sample showed good agreement with the actual amount, as shown in table III. Nickel Oxide Internal Standard-Paste Samples The results of the standard specimens for the paste samples are shown in table IV. The CoWO4 showed a very high variation of individual peak intensities. In addition, it was observed that the ratio of two peaks of the CoWO^ was highly variable, indicating ex- tremely variable preferred orientation of the monoclinic CoWCh phase. However, meas- urement of the sum of two peak intensities did not improve the variability of the measure- ment of this oxide. The results for the prototype paste sample as shown in table IV showed good agree- ment between the actual weight fractions and those determined by intensity ratios except for CoWCK: Cr^O* was 3. 1 percent compared with 5. 5 percent actual, CoCr2O4 was 39. 7 percent compared with 39. 5 percent actual, CoO was 39. 6 percent compared with 40. 0 percent actual, and CoWO^ was 9. 0 percent compared with 15. 0 percent actual. The high error for Cr 0£tOotS is believed due to the inaccuracy in measuring the low signal from the small amount of Cr^Oo present. The paste method was used in the analysis of a series of spalls containing only these oxides. These spalls were produced during the cyclic oxidation of a cobalt base alloy (WI-52) at 2000° F. Because of the poor accuracy for the CoWO^ standard specimen, X-rays were not used for its determination in the WI-52 spall samples. Since the spalls consisted of only CrgO,, CoCrgO^, CoO, and CoWO4, the first three phases were deter- mined from intensity ratios, and the CoWO4 by difference. These results are also shown in table V. The analysis was valuable in this case in showing how the amount of and CoO in the spall depended on the surface preparation of the samples. CONCLUDING REMARKS The methods described in this report were adequate for quantitative analysis of the phases studied except for CoWO4. An accuracy within about 10 percent of the amount re- ported was obtained on oxide blends for weight fractions above about 0.1 for each of the phases. For lower weight fractions counting errors in determining the height of the peaks will cause more error, but even these lower values were good to about 1 percent absolute. The applicability of such methods to other systems depends chiefly on the following: 1. The phases present in the sample should be known and their X-ray patterns well characterized. This is necessary to insure that there is no superposition of other lines on those used in the analysis. 2. The individual phases used in making a standard should be the same as the phases to be determined. 3. A homogeneous mixture of the unknown with the correct weight fraction of the in- ternal standard should be achieved. 4. There is little lattice strain in the phase to be analyzed. 5. The crystallites should be large enough so that line broadening due to small crys- tallites is not a problem. 6. There is little preferred orientation in the phase to be analyzed. The analysis of the spalls from the oxidation of Co base alloy WI-52 suggests the ap- plication of such techniques to other superalloy spalls if the oxides formed are well char- acterized. For such oxides it should be possible to achieve an accuracy of ±10 percent of the weight fraction reported. At present there is no other method for quantitative analysis of such phases. Lewis Research Center, National Aeronautics and Space Administration, Cleveland, Ohio, June 7, 1972, 134-03. REFERENCES 1. Grisaffe, Salvatore J.; and Lowell, Carl E. : Examination of Oxide Scales on Heat Resisting Alloys. NASA TN D-5019, 1969. 2. Klug, Harold P. ; and Alexander, Leroy E. : X-Ray Diffraction Procedures for Poly- crystalline and Amorphous Materials. John Wiley & Sons, Inc., 1954, pp. 410-438. 3. Copeland, L. E. ; and Bragg, Robert H. : Quantitative X-Ray Diffraction Analysis. Anal. Chem. , vol. 30, no. 2, Feb. 1958, pp. 196-201. 4. Barrett, Charles S. ; and Massalski, T. B. : Structure of Metals. Third ed., McGraw-Hill Book Co., Inc., 1966, pp. 171-181. 5. Munch, R. H. ; and Pierron, E. D.: Practical Method of Powder Sample Preparation for X-Ray Diffraction. Norelco Reporter, vol. 10, no. 2, Apr. -June 1963, pp. 75-76. 6. Sidik, Steven M. ; and Henry, Bert: RAPIER: A FORTRAN IV Program for Multiple Linear Regression Analysis Providing Internally Evaluated Remodeling. NASA TN D-5656, 1970. 7. Deadmore, Daniel L. : Observations on the Cyclic and Isothermal Oxidation of WI-52 andIN-100. NASA TM X-2195, 1971. 8. Anon. : Powder Diffraction File. Spec. Tech. Publ. No. 48, ASTM. 8 TABLE I. - REFLECTIONS USED IN TABLE II. - MIXTURES USED FOR QUANTITATIVE ANALYSIS CALIBRATION CURVES FOR (a) Silicon internal standard - loose powder method DETERMINATION OF NiO, Phase Miller indices, Interplanar Intensity ratio, A12O3, AND Cr2O3 USING hkl spacing, [I/I/ SILICON INTERNAL m STANDARD Si (100) 3.14X10"10 100 [Standards are these mixtures plus NiO (111) 2.41 91 0. 25 weight fraction silicon. ] A1203 (104) 2.55 90 Cr (104) 2.67 100 Mixture Weight fraction 2°3 (b) Nickel internal standard - loose powder method NiO A1203 Cr 2°3 Ni (111) 2.03X10"10 100 1 1.00 0 0 NiO (111) 2.41 91 2 .75 .25 0 A1 (104) 2.55 90 3 .75 0 .25 2°3 Cr (104) 2.67 100 4 .50 .50 0 2°3 5 .50 0 .50 (c) Nickel oxide internal standard - paste method 6 .25 .75 0 7 .25 0 .75 NiO (200) 2.09X10" 10 100 8 0 1.00 0 CoO (200) 2.13 100 9 .75 .25 Cr 2°3 (104) 2.67 100 .50 .50 10 CoCr 2 O 4 (311) 2.51 100 11 .25 .75 CoWO4 (110) 3.73 30 12 1 ' 0 1.00 CoWO4 (031) 1.76 14 Ref. 8. TABLE III. - SUMMARY OF ANALYSIS OF PROTOTYPE OXIDE MIXTURES Method of Oxides Weight percentages Relative analysis present standard Added Determined deviation, percent 0 Nickel A1203 24 24 13 internal NiO 30 30.5 10 standard Cr ? O, 10 10 10 wca 17.6 -- Coa 2.4 -- Si °2a 16 -- Nickel CoWO4 15 9 -- oxide 39.5 39.7 internal Cr 5.5 3.1 -- 2°3 standard CoO 40 39.6 -- Extraneous phases not present in standards. Average of 10 replications. x 100. -RSD average percent TABLE IV. - INTENSITY RATIOS FOR STANDARD SPECIMENS [NiO internal standard-paste samples; each standard is phase listed +0. 1 NiO. ] Phase Ratio measured Average ratio for Range of apparent 1 . 0 weight fraction weight fractions (5 values) I Cr2O3(104) Cr 2 0 3 3.75 0.98 - 1.04 : NiO(200) T CoO CoO(200) 2.99 0.98 - 1.02 T NiO(200) ^oCrgO^Sll) CoCr 2 O 4 1.81 0.94 - 1.04 ! NiO(200) r CoWO4(100) + (031) CoW04 1.72 0.70 - 1.47 T NiO(200) 10 TABLE V. - QUANTITATIVE X-RAY ANALYSIS OF SPALLED SCALES FROM GROUND-LAPPED AND GROUND-POLISHED WI-52 FOR 20 HOUR CYCLES AT 2000° F (1266° C)a [Analysis used NiO internal standard - paste method. ] Cycle Phase content of spall, wt. % Cr 2°3 CoCr 2 O 4 CoO CoWO4 (by diff.) Ground -lapped 1 3 50 34 13 2 2 50 37 11 3 4 50 35 11 4 3 45 45 7 5 2 45 36 17 6 3 47 39 11 Ground -polished 1 11 22 54 13 2 3 29 59 9 3 3 21 56 20 4 4 23 51 22 From ref. 6. 11 1.0 S .6 S .2 0 .2 .4 .6 .2 .4 .6 .8 1.0 Amount of NiO, weight fraction Amount of A weight fraction Figure 1. - Calibration curve for quantitative determination Figure 2. - Calibration curve for quantitative determination of nickel oxide (NiO) using 0.25-weight-fraction silicon in- of aluminum oxide'(Al203> using 0.25 weight fraction sili- ternal standard. Slope, 2.16; number of data points, 20, con internal standard. Slope, 0.726; number of data except 10 at 1.0 weight fraction; pooled standard deviation; points, 20, except 10 at 1.0 weight fraction; pooled stand- 10.2 percent. Ranges on curve are one standard deviation. ard deviation, 9.4 percent. Ranges on curve are one standard deviation. 2.0 1.6 S 1.2 .4 .2 .4 .6 .8 1.0 .2 .4 .6 .1 1.0 Amount of weight fraction Amount of NiO, weight fraction Figure 3. - Calibration curve for quantitative determination Figure 4. - Calibration curve for quantitative determination of chromium oxide (C^Cty using 0.25 weight fraction sili- of nickel oxide (NiO) using 0.1 weight fraction nickel inter- con internal standard. Slope, 1.61; number of data points, nal standard. Slope, 3.63; number of data points, 20 at 20, except 10 at 1.0 weight fraction; pooled standard devia- 0.5 weight fraction, 10 at 1.0 weight fraction; pooled stand- tion, 8.7 percent. Ranges on curve are one standard ard deviation, 1.4 percent. Ranges on curves are one deviation. standard deviation. 12 .2 .4 Amount of Figure 5. - Calibration curve for quantitative determination Figure 6. - Calibration curve for quantitative determination of aluminum oxide (Al203> using 0.1 weight fraction nickel of chromium oxide (Cr203> using 0. 1 weight fraction nickel internal standard. Slope, 1.26; number of data points, 20 internal standard. Slope, 2.75; number of data points, 20 a t O . 5 weight fraction, 10 at 1.0 weight fraction; pooled at 0. 5 weight fraction, 10 at 1. 0 weight fraction; pooled standard deviation,- 2.4 percent. Ranges on curves are standard deviation; 2. 2 percent. Ranges on curves are one standard deviation. one standard deviation. NASA-Langley, 1972 18 E-6911 13 : IRST CLASS , ' ...••.:•!.' ' • . 1 : . - §i 6j the' ! i. :• , ' | | : : t$. C6K; j 2 ? • ; . : / , H I -»owl- r : , . . I | | and space, 'ft,*- atim r • • • Wie m •. | | I an4 app-, . • \ | | •wl/'off : : '. . ?fl • ;/ AN;.: : ' • •^ I : ' 1958 :, t i.: .:HNICAL I ..ILICA: ,INS |TECHNI| .MSLATIONS: H p I ; ||| I | ill ; .- .: '- - <, ' t K rit NASA d ! don In English,; I PUBLF 'MS: Information :r ! i ' - ' " '. : • ' , •.-•;••-. ols I i , I I CXJY UTL . ,||. j - 'interest iercial and other not ice applications. Publications include Tec ization Reports an«|s \ INFORMATION OFFICE ) SPACE ADMINISTRATION