Measurement of the Pearlite Content in Ductile Iron Microstructures AFS Ductile Iron Division Quality Control Committee (12-E) ‘AgsTRACT Difforont methods of making an exact deter mination of tho porcontage of pearlte present in a microstructure were compared bby mombors of the AFS Ductile Lron Quali ty Contral Committee. The point count method was found to offer an acceptable level of accuracy while requiring « minimum, ‘expenditure in de and equipment, Further ‘work is planned to develop a set of refer fenee photomicrographs as an aid in making rapid visual comparisons, Introduetion—Anyone who has been associ sted with the use of casting material spec fleations, in which one of the requirements is a limitation on the amount of a cortain constituent in the microstructure, will be aware of the problems that can arise in practical applicition of these epecifications. ‘The seeds of misunderstanding between customer and supplior may already be provent before the onder for cartings has ‘been formalized. Specifications often con tin vague terms such as “2s minimum of pearlte” or “essendally fersitic® that mean ifferat thins o different people ‘Attempts. to inchde 2 measusement smumber in the requirement, as with “10% ‘maximum pearlite,” may be based on a set ‘of reference photomicrographs, showing ‘ifferent measured levels ofthe constituent However, they are most often based on pettondl Interpretation by the man looking at the structure, Occasions when supplier interpretation and customer interpretation {09 not agree are innumerable, Even with control photomicrographs the actual structures and those viewing the structures vary widely enough so that there is often disigseement as to whether 2 now structure contains more or less of the const ‘ueat than is shown in the reference photo- micrograph. However, this approach does ‘offer an optimum in simplicity and speed that is desirable in the evaluation of large numbers of microstructures, When applied by experienced observers, it need not suffer from the inaccuracies usually associated with the simple visual estimation of const twents, Evaluation of a cast part for its planned application on the basis of microstructural {examination has certain obvious limitations ‘The area examined is very minute and thus ray mot be representative of other ares in the casting. This is especially true if dif- ferent parts of the casting have cooled at difteent rates ~ due to variation in section ‘thickness, locaton in the mold relative to risers of adjoining hesyy sections, thermal history through shakeout or heat trtment and similar process variables. Microstruc. tural examination at several pointe on a casting, a8 « means of overcoming this Timi tation, can become time consuming and costly. Finally, the fact that any exami- ration of structure requires a certain mount of time and mast often be destrac live of the part involved usually makes 100% inspection with this method impxac- tical, if not impossible {In many applications some other method of evaluation will better serve the purpose described. This might be a hardness test or some type of service test measuring harden- ability, machinability or strength under actual ‘conditions of manufacture and use Such tess overcome some ofthe difficulties mentioned above. Also, they can often pre- ict the performance to be expected of a casting in service on a direct bass, rather than inferentilly as would be the case when ‘examining microstructure, However, recognizing the limitations listed and adjusting test procedures to mini- mize them, the examination of microstruc. ture stil remains a timeshonored and popular method of determining the proper ties ofa casting or group of castings. In fact, certain sorvice properties ae sll ‘est controlled by placing limitations on the percentage of a_pven constituent in the microstructure, These might include a maxi mum lovel on ferrite to ensure good 20. sponse to surface hardening, a maximum level on pearlte to reduce cating growth at elevated temperatures or a maximum level fon peatlte to prevent brittle fracture at reduced temperatures. InvestigationRecognizing the fact that the numberof specifications containing a micro- structural requirement seemed 0. be in- ‘creasing, but hopeful that their application could be made less of 2 problem, the Qual ty Control Committee set out fo evaluate the different methods of measuring, rather than estimating, the percentage of peatite present ina ductile iron microstructure. Tho methods considered were: Weight Methodin which @ photomicro ‘gph of the structure is enlarged to common 8 x 10 in. The different const- tuent areas aze outlined with a pencil or ‘other marker and 2 small manicure scissors used to cut the different areas out of the photograph. All. areas representing each Constituent’ are weighed on a laboratory balance and percentages are then calculate. nimetor Methed—in which an instrament, commonly used to measure isroularl) shaped areas in topography, thermo- ynamics and. other fields, is used. to ‘measure similar areas on an enlarged photo- rmictograph. Design of this instrument is ‘based on the measurement of area by inte: gration of linear dimensions. Different Constituent arcas in the photomicrograph are outlined and the stylus of the instru: ‘meat made 0 follow the outline of each area of interest in the constituent. The atea is then marked #8 counted, the instrament reading recorded and the instrument zeroed {to bezin measuring the next ares. Afterall, areas have beon meatured, instrament read ings are totalled and their percentage of the ‘otal area calculated. Point Count Method-in which a photo: ricrograph of the structure & enlarged to a common 8 x 10 in, The diferent constty lent areas are outlined with a pencil or other marker. A grid of evenly placed lines is placed on the photograph, The amber of sid line intersections that fall on the con stituent areas of interest ae counted, Then their percentage of the total number of points is calculated, Improved speed and reproducibility can be achieved with this method by construo- tion of a transparent plastic template with ‘tho grid of lines drawn on the plastic shee, ‘The template can then be laid on tha photo- micrograph and an ink marker used to ind cate each intersection counted, In this way intersections will not be missed or counted ‘twice, count can bo interrupted when necessary and there is no need to keep the count figure in the observer's head as he proceeds. A small manual counter may be Useful for keeping tho count figure intact during the counting process, Afterall the intersections have. been’ marked and counted, a percontago can be caloulated and the ink markings on tho template erased, For this study a field 7.5 x 9.5 in, was laid out on an 8 x 10 in, tomplate, assuming ai in. border on a standard size photo: ‘graph. Then inside this fc, stating 1/8 in, from each side, horizontal and vertical Hines were drawn at Y in, intervals, (See in- sett) Line Intoreopt Method—in which a number of evenly spaced calibrated lines are super imposed on the microstructure, Tae number ‘of calibration units that full on areas of the constituent in question are counted. Then the percentage of the total number of cal ‘bration units i calculated. This count is usualy made with the ricroscope directly on the polished sped ‘mea without the production of a photo imlerograph. The structure is slowly scanned at a constant rate along one axis, with the ‘observer operating some sort of counting Gevice whenever the scanning point is cror sing an area of the constituent in question. Several parillel scans are made at evenly spaced intervals and. perceniage of the ‘tl lineal travel is aloulated, ‘This method was considered but not in- cluded in the comparisons carried out in this investigation. It vas recognized that such measurement would be quite time con ‘suming and would require microscopic equipment not available in most labore tories ‘An altemate approach, using a photo: imerograph and a visible set of lines fs also considered acceptable. Using this approach, speed comparable to the point. count method could probably be achieved by con- struction of a transparent template, similar {o that described above. In this case, the template would contain a sct of evenly spaced parallel lines. Each Ine would be ‘uniformly calibrated slong Its length. A count Would be made of the number ofcalibration units falling on areas of the ‘constituent in question and a percentage ‘aleulated Jn this case, accuracy would depend on the proper ehoice of calfration lengihs and line spacings. However, since the mechanics fof this method did not seem to offer ad- vantages over the point count method, no attempt was made to evaluate the effect of these two variable, If @ metallograph with the capability of projecting an image of the structure on a {round plas scroen is available, both point ‘count and line interospt methods can be carried out by placing the transparent template on this sereen, This technique al- lows the observer to make a count at cach Of several locations on a specimen. These fan then be averaged and greater coverane fchioved in less time than would be required for the printing of several photomicro- graphs. OF course, using. this approach a Permanent record is not kept of the areas sieasued. ‘Automatic Scanning Method~in which the specimen or photomicrograph is system cally scanned to evaluate several hundred ‘thousand individual points. The intensity of, Tisht at each point is measured as an indi cation of whether that point should be identified as one constituent or another. Signals from each point go to some type of computer where they are totalled and per centages calculated. Such systems require Considerable inital investment but offer reat speed and exact reproducibility of results. However, their ability to identify zeus as peat or ferrite is controlled by the judgment of the operator who deter ‘ines the intensity levels at which differen= tiation between the (wo structures wall be ‘made. ‘A problem arises too in the interpreta: tion of the microstructure of the various ‘ait irons, In many case, the ight intensity levels from peatite and graphite are so similar that iti difficult t0 distinguish be- tween thom. One technique often used is to scan the unetched specimen first to de- termine the graphite content, then etch the specimen and scan again. The results of the stcond scan are subtracted from those of the fist to determine the peasite content in the structure ‘When using any of these methods, the practice of outining all pearite areas before beginning to measure them seems to be a matter of personal preference. Those who favor the practice claim that it allows the observer to concentrate on the iwo distinct Fequirements of identification and measure ment separately, rather than simultaneously, resulting in. both greater accuracy and ‘greater spacd. If the results of a measure ‘nt are to be subject fo review by a second observer at a later date, outlining the areas measured, for the purpose of reference, ‘would seem to be a wise precaution. Comparison OF Methode—Six photomicro- raphe, exhibiting different levels of pearlte fontent, wero chosen for measurement Each photomicrograph was enlarged 10 8 x 10 in. The areas considered by one observer ‘to be peatlte wore outlined on cach photo- s1aph. Thea these marked photographs were ‘measured by the weight and point count methods, Two steuctazes were also tose checked by the planimeter method. The original photomicrographs are shovm in Fig 1-6. The marked enlargements of these photomicrographs are shown in Fig 7-12 ‘Values obtained by the three methods of ‘measurement are shown in Table 1. It can bbe seen that the deviation in percentage reported by the different methods for a sven structure was never greater tina one and one-quarter percent. Fig, 2. Microstructure @, nital etch, 100X. 3. Microstructure C, nital etch, 100X. . Sao Oe Fig. 6. Microstructure F, nital etch, 100X, ‘The weight method it considered by some investigators the most accurate of the three methods. However, it is time con suming and tedious — requiring a typical {two hours of cutting to measure the pearlte content in a Single structure, Lf one photo- micrograph is to be held as e record of the areas considered t0 be peatlite, two of the Same structure ate outlined. Finally, this method requires possesion of 2 balance capable of measuring infractions ofa gram, « an item not found in every quality control laboratory,Tobie 1. Pearlte pereentoge in structure, measured by ferent mathods on matkod photomicrographs. Microstructure ae) cd D E F Weigint Methoo TE Sy Ga ES Sa ha Point Count 4 Seer Planimeter 166 5 3 : 250 ‘The planimeter method should logically bbe as accurate as the weight method time required to make a measurement in the ‘order of 20 to 30 min per structure. Proper technique fer operation of the instrument «an be easily learned. ‘The point count method was i good agreement with the weight method in this series of comparisons. A point count can be completed in $ or 10 min and the only equipment required is the template de- feribed above and a felt tip ink marker. Variation Between Observers—In another comparison, the point count method was, sed by several observers to measure the percent pearite in each of the structures previously chosen, However, in this ease, the Identification of pearite areas was left to the judgment of the individual observers, Instead of marked in advance by one party Values reported by the different observers are shown in Table 2 Fig, @. Ductile iron microstructure B, nital fetch, 100X, onlarged and matked, ‘The wide variation between certain of ‘these measurements on the same structure points out a problem that will aways ze- main, regardless of the measurement ‘method sod. That is the original determi- nation on the part of the observer of what dteas he considers to be pealite, Differences in interpretation between two observers — for example, customer and supplicr ~ must be resolved before agreement can be reached fon measured pereentares. Outlining, the Dearlte areas ‘before beginning a point count, then holding the marked photograph for future reference, can help to bring such differences to ight, From data in Table 2, the mean value reported for each structure was calculated. Also, the confidence limits ofthis mean at ‘the 95% confidence level. This figure sug- gests where the mean might be expected 10 fall, if 4 much larger number of observers were surveyed, These figures are listed in Table 3. The table also contains the sian- Table 2. Pearlite perc Microstructure A Observer 1 154 2 200 3 208, 4 187 5 24 6 as 1 150 8 183 9 26.0 10 17 n 232 2 260 a 210 4 22 15 200 16 iss 7 174 18 172 9 230 20 168 2 172 2 179 3 216 m4 192 25, 204 25 190 Py 18.7 28 20.0 29 194 30 1538. 31 164 32 167 3 164 34 176 35 218 B dard deviation figure for each set of data, to Indicate the variation of individual observa: tions about each mean, Thus, the mean value reported for microstructure A by a larger number of observers would be ex: pected to fall between 18.5 and 20.3%, (19.39 + 0.95). However, valves reported by 95% of the individual observers could only bbe expected to Fall within a wider range of = two standard deviations ofthe mean; that i between 13.9 and 24.9%, teh, 100X, enlarged and marked. ntage in structure measured by different Ubservere on unmarked photographs using paint count method. © Deere ca eigen saieenis rea ste 12 168 145 103 179 160 a 16 mi 140 Wag euieieee 12a) 39 174151 74 180 127 53 8 100 Gabe pass ene: 83206 167 66 6S 125, ean UI 126: 13 163 62 13075 aes tesa 12 168 118 ie 180") 25.0 93 158 89 a4 26 119 66 11070 BS 16098 84 170109. hay 1860 uni07, BO 18010260 83 29 128 270 a1 197 «149s 96 24 «130 29.7 69 1ST 89244 89 158 10247 81 162 94233 6§ 13997235 Tans tome haa io emeat pert 3 Ra25 observer declined to identify pearlite in this structure because of the poor aualty of photomicrographFig. 11, Ductile ron microstructure E,nital fetch, 100X, enlarged and marked. Fig. 12, Ductile iron microstructure F,nital fetch, 100X, enlarged and marked. It should be emphasized that these wide variations in pezcentage reported for the same structure point aut « problem that wll always remain regardless of the measure: ‘ment method used — the erisinal determi nation on the part of the observer of what areas he considers to he pealite. Differences in interpretation between two observers must be resolved before agreement can be reached on measured percentages. Effect of Metallographie Technique—As individual observers reported the per ‘centages they had determined on the photo- micrographs sent to them, it became ob- vious that the quality of the metallographic nd photographic techniques used to pto- Guce’ these photomicrographs could be seriously questioned. Commenis from dif: ferent observers were “1 Found it difficult to achieve satis factory identification of peat areas in structure E. The peaite inthis structure appeared fuzzy, a8 though excessive nickel or manganese might have resulted fn a poor response to anneal. Structure B posed ‘similar problems bat was more {typical of commercial feritic ion.” “interpretation of structure Edit Ficalt, due to poor clarity of photomsiexo- graph.” “Some of the photomicrographs were difficult to interpret If poate isto be determined on an enact basis, I suggest that in the preparation of future speci- mens the final procedures include atleast ‘two light etching and polishing steps before final etch, in order to remove all disturbed metal, which might mask proper identification from a picture.” rican’s inability to trim pealite areas with complete accuracy could have a greater ef fect on the final result, if he were cutting ‘out very sinall pieces from 2 small photo: raph instead of lager pieces from a large photograph. With any method, variations ln the widih of lines drawn with a felt tip marker on a small photograph might affect the result reported. Outlining areas. and counting points, when extremely smal areas ate invoWed, can cause problems of mental foncentration and aye stain, "The accuracy of the point count method varies with the total number of points used. To have the number of points on 4 x Sin. photograph equal the number present on an 8 x 10 photograph, where a 44 in. grid spacing is used, would require 1/8 in. grid spacing on the smaller photograph. Again eye strsin would bea problem, If the wider spacing were maintained on the smaller photograph, only about one fourth as many points of line intersection would be produced, Such a educed number Of points could also be produced by using 2 Yin. grid spacing on an 8 x 10 in, photo raph. A comparison between this approach and the original % in. spacing was carried ‘ut, producing the values shown in Table 4 Variations 2s great as 4% pearite were found when the reduced number of points was ued, Table 3. Moan value and standard deviation, pearlte parcantages reported by diffrent obearvers Microstructure AB @ Deen mean of reported values 19.39 391-778 170611252531 confidence limits 095 034 036 «= 083087082 standard deviation 276 098 108242297238 “The quality ofthe photomicrographs supplied was not as good as desired Tor this type of study. I would expect this factor to contribute to the seater in the results from the individuals patielpating in the study.” “Photomicrograph Eas too mach contrast and some of the graphite shapes are degenerate. Those factors make it Impossible to tell pealite from graphite, in many ares.” “Samples A and E are sather poor from a photogrphie viewpoint. 1 ques tion the use of these samples for any thing other than a preliminary roport on this procodure.” Rebuttal has been made by some com mittee members that these photomicro- ‘aphs are representative of metallographic work produced throughout the industry — Whether by customer, supplier or com mercial lboratory. However, they agree that photomicrographs of 2 higher level of quality should be obtsined for publication 35 AFS reference photomicrograph. Enlargement Of Photomicrogeaphs~The need for enlargement of standard size Photomicrographs to the larger 8 x 10 in Size for these purposes might be questioned by some readers because of the resultant fnereased cost. However, such enlargement reduces the level of inaccuracy caused by limitations in technique or equipment. For Instance, with the weight method, the tec Percent of Matsie—The percentages included in this report up to this point have all been the percentage of peatite contained in the complete structure. Standard. practice in ‘mote stietly scientific circles is to report the percentage of pearlite present in the matrix alone, excluding the presence of free waphite. When visually estimating per~ centages such a requirement tends t9 con fuse the problem, but when exact measure- ‘ment methods ate used the mount of ad- tional work involved depends on the ‘method wre. In general, the time required to measure percentage in the matrix will be twice that ‘equited to measure percentage in the struc- ture. Since the point count method is the fastort of the three methods under consider- ation, its convenience will be magnified ‘when measurement of the percent pearlto in the matrix must be made, It 1 only necessary to make a second count of the points that fall on graphite areas and then Subtract these from the total number of Points in the structure before dividing to get ‘percentage figure, Measurement of percent peatlte In the matsix and percent free graphite were made fon the structures originally shown, using Doth the weight and point count methods ‘Values obizined are shown in Table STable 4. Pearlite percentage in ructure ‘measured by point count ‘method on marked photomicrographs, using different grid spacings. Microstructure ea Spacing 4 in 15439 Spacing in, 196 36 Conclusions—Several methods of measuzing rather than estimating the pearlte content fof a ductile iron microstructure are aval- able, In general, these methods offer com parable. accuracy but vary. widely in the txpendituze of time and equipment. The Point count method requires les time and ‘equipment than any of the other methods considered. Therefore, it would seem to offer a practical tool whereby customer and supplier, having agreed on which areas in a given photomicrograph they are going Identity as paste, could sit down with two ‘copies of the photograph and each ative at 1 nearly identical figure Tor the pearlte Content of the structure “Table 5 Pealito percentage in ms lifforent mothods on marke Microstructure eat) peatlte-woight 1g 49 peaslte-pt count 11945 exaphite-weight 139 188 sxaphite-pt count 139132 Future Plans~This study should be looked tapon as a progress report of the work done ‘yy the Committee to date, It is published {for the purpose of promoting discussion as ‘well 2s providing information, Tn the near future the Committee will supply sets of unmarked photomicrographs toa greater number of observers for ‘meagurement by the point count method. It has been recognized by Commitice members that agreement between several observers on 44 marked photomicrograph should be quite good. Tho point count technique i simple fo understand and apply. Serious disagroe- ment should only arise when different ‘obsorvrs exorcise their individual judement, regarding the identity of the differont con” stituents in the structure. By distbuting ‘unmarked photomicrographs, itis hoped 10 determine the magnitude of such disagree ment for the structures in question. The final reslt of this study should be produc tion of aset of reference photomicrographs. Exch photomicrograph would be identified 25 exhibiting a certain most probable per centage of pealite and the statistical dovie- tion of this percentage stated. Further, the Committee recently ar ranged to have the pearlite content of these structures measured by one of the optical Scanning devices. Such data should provide nother set of guidelines for establishing a most probable percentage figure for the structures under consideration, c D F Bl aks 62 7s ering ated aie 26:1 ‘The members of the Committee feel that reference photomicrographs are nocesary {or rapid evaluation of structures in produc on. They look upon the point count method ai a practical referee technique ‘whon particular questions conceming. the accuracy of the reference photo method ‘Acknowledgment—The Committee wishes to express its appreciation to the following frganizations that made time, material and facilities available to carry out this invest gation: American Cast Iron Pipe Co. fon Deere & Co., H. P, Deuscher Co., Ductile Iron Society, Dusizon Co., Inc, Ford Motor Company, Grede Foundries, Inc., Hamilton c Disaeeeee fo al 204164 300 90 206 L313 ios 92187187 162 162 Foundry Division and Kuhas Bros. Co. The committe also wishes to gratefully acknow- ledge the guidance of Paul R. Gouwens, AFS ViooPresident ~ Technology and Evra L. Kotzin, Manager, Technical Services. APPENDIX Point Count Method Percent Pearlta in Total Structure 4, Using a felt tip marker with a fine point, ‘outline all the areas on the photograph that you consider to be pearite 2. Center the gid layout of the template on the plate of the photograph. Because of differences in the way individual photo- graphs may be cut, this may not always bbe the same as lining up the outside edges ‘of the template with the outside edges of the photograph. 3. Use papor clips, another dovice to keep ‘the template from slipping while you ae counting. 4. Count all points where a fll intersection ‘of two lines falls on a peasitic area. Do not include the partial intersections of lines around the edge ofthe gsi. 5. Ifa point falls on the edge ofa peatlte area, count it as peat. Such points right only be counted at half value in a more exact method, but since there area total of more then 1100 intersections on the grid, such a refinement in this ease ‘does not reslt ina significant increase in accuracy. 6. Mark all peatitic points on the template with a felt tip marker, then go back and ‘count them, In this way a line of points ‘will not be skipped and you will not lose Count if interrupted. After the count is Complete the points can be. removed from the template with a damp cloth. 7. The plates of some photographs may be smaller than the geid layout of the over: lay. This will cause the total number of points falling on the plate to vary. The complete grid layout contains 30 inter- sections along the 8 in. dimension and 38 intersections along the 10 in. dimension for a total of 1140 points. Peints not used, becluse they are off the plate or only’ on the edge of the plate, should be subtracted from this total 8, Percent peatlte in the total structure is determined by dividing the number of ppoinis falling on pearlte arcas by the fotal number of points, then multipiving bby 100. For rapid approximation this can become a simple matter of dividing the sunber of peu points counted by Note: to determine percent pearlte in the matre, the number of Points falling on graphite reat ‘must be counted and subtrac. ted from the total number of Points. Then the number of earlitic points can be divided by this reduced total and again ‘multiplied by 100 Calculations Involved in Determining Mean Values, Confidence. Limits and. Standerd Deviations: individual values reported sum of individual values humber of observer: sean of values Gifferonce between indivi dual values and mean sum of differences squared confidence limit of mean 2.034 = A constant value for 955; confidence Standard Deviation Confidence Limit © Example (microstructure A) een eee ag fie vom anmam =o