Equilibrium Grain Boundary Segregation of Antimony in Iron Base Alloys Ravnotežna segregacija antimona po mejah zrn v zlitinah železa in antimona r. Mast1, H. Viefhaus, M. Lucas, H. J. Grabke, Max-Planck-lnstitute, Dusseldorf, Ger-many Prejem rokopisa - received: 1996-10-01; sprejem za objavo - accepted for publication: 1996-11-04 The equilibrium grain boundary segregation of antimony ivas investigated in iron base aiioys (Fe-Sb, Fe-C-Sb, Fe-Ni-Sb) after anneaiing at temperatures betvveen 550°C and 750°C. Utiiizing Auger electron spectroscopy (AES) the concentration of antimony at intergranular fracture faces was determined as a function of buik concentration and equilibration temperature. The segregation of antimony in Fe-Sb alloys with 0,012 wt.% - 0,094 wt.% Sb was described by the Langmuir-McLean equation. The evaluation leads to the free enthalpy of segregation AGsegr. = -19 kJ/mol - T 28 J/mol K. For Fe-0,93 wt.% Sb and Fe-1,91 wt.% Sb a thermodynamic calculation is not possible because of intergranular antimonides had formed. Scanning electron micrographs (SEM) of fractured samples show that the percentage of intergranular fracture increases with an increasing coverage of antimony at the grain boundaries. The addition of carbon to Fe-Sb alloys results in a higher grain boundary cohesion which is caused by two effects of carbon, displacement of antimony from the grain boundaries by carbon and enhanced grain boundary cohesion. In the Fe-Ni-Sb alloys an additional segregation of nickel was found at the grain boundaries but no enhanced antimony segregation, as expected from previous models of other authors, assuming Ni-Sb cosegregation. Key words: grain boundary segregation, antimony equilibrium segregation, Fe-Sb alloys, Fe-C-Sb alloys, Fe-Ni-Sb alloys, segregation thermodynamics, Langmuir-McLean equation, Auger electron spectroscopy (AES), intergranular fracture, embrittlement, site competition, Charpy impact tests Ravnotežna segregacija antimona po mejah zrn v zlitinah z železnoosnovno (Fe-Sb, Fe-C-Sb, Fe-Ni-Sb) po žarjenju v temperaturnem področju od 550°C do 750°C. Z metodo spektroskopije Augerjevih elektronov (AES) je bila določena koncentracija antimona na interkristalnih prelomnih ploskvah kot funkcija vsebnosti antimona v osnovnem materialu in ravnotežne temperature. Segregacija antimona v Fe-Sb zlitinah z 0.012 ut.% - 0,094 ut.% Sb je opisana z Langmuir McLeanovo enačbo izračunana je bila prosta entalpija segregacije AGsegr. = -19 kJ/mol - T 28 J/mol K. Za zlitini Fe -0,93 ut.% Sb in Fe -0,91 ut.% Sb termodinamični izračuni niso mogoči zaradi tvorbe interkristalnih antimonidov. Posnetek z vrstičnim elektronskim mikroanalizatorjem (SEM) prelomljenih vzorcev kaže, da odstotek interkristalnega preloma narašča z naraščajočo segregirano plastjo antimona na mejah zrn. Dodatek ogljika v Fe-Sb zlitino povzroči večjo kohezijo med posameznimi zrni, ogljik namreč izrine antimon z mej zrn in zviša kohezijo kristalnih mej. V Fe-Ni-Sb zlitinah je bila določena še segregacija niklja na mejah zrn ne pa tudi povečana koncentracija antimona kot je bilo pričakovati po prejšnjih modelih nekaterih avtorjev, ki so predvideli skupno segregacijo Ni-Sb. Ključne besede: segregacija na mejah zrn, ravnotežna segregacija antimona, Fe-Sb zlitine, Fe-C-Sb zlitine, Fe-Ni-Sb zlitine, termodinamika segregacij, Langmuir McLeanova enačba, spektroskopija Augerjevih elektronov (AES), interkristalni prelom, krhkost, tekmovanje za prosta mesta na površini, Charpyjev udarni preizkus 1 Introduction The increased usage of low quality scrap in steel production will lead to a higher content of antimony in steels, which may have a deleterious effect on material properties. The presence of antimony (and/or other tramp elements such as P, Sn, S, As) induces temper embrittlement of low alloy ferritic steels by segregation to the grain boundaries during application at higher temperatures1'2-1. The driving forces for such an enrichment in a range of a monolayer are the decrease of interfacial en-ergy and the release of elastic energy. Especially the lat-ter effect is important for antimony because of its large atom size compared to iron atoms. Many researches have been shown that the amount of antimony segregation de-pends on the total composition of the steel. However, there is no uniform evidence how other alloying components, especially nickel2-4'5, influence antimony segregation. Dr.Sc. Ralph MAST Max-Planck-Inslitut fiir Eisenforschung GmbH Postfach 140 444. 40074 Dusseldorf, Germany Therefore, the equilibrium grain boundary segregation of antimony and its effects on material properties vvere examined in simple iron base alloys to avoid the complex chemistry of multicomponent steels. The degree of coverage vvas determined by Auger electron spectros-copy (AES) on the intergranular fracture faces after fracture by impact inside the UHV chamber. The influence on the mechanical behaviour vvas studied by scanning electron microscopy (SEM) and Charpy impact tests. 2 Experimental procedure The alloys used in this study vvere melted in a vacuum induction furnace. The chemical compositions are listed in Table 1. Small amounts of manganese (0,02 wt.%) vvere added to each alloy to tie up sulfur, vvhich has a strong tendency for grain boundary segregation3 and may hinder antimony segregation. The ingots of the Fe-Sb, Fe-C-Sb and Fe-Ni-Sb al-loys vvere hot forged and then machined into rectangular specimens. The Fe-Sb and Fe-Ni-Sb samples vvere heat treated by austenitizing at 1060°C for 70-90 min, air Table 1: Chemical composition of the Fe-Sb. Fe-C-Sb and Fe-Ni-Sb alloys (wt.%) Alloy Sb C Mn P S Fe-Sb 1 0,012 0,005 0,027 0,0015 0,0013 Fe-Sb2 0,049 0,0048 0,027 0,0011 0,001 Fe-Sb3 0,094 0,0057 0,027 0,001 0,0011 Fe-Sb4 0,93 0,006 0,026 0,0013 0,0012 Fe-Sb5 1,91 0,0039 0,028 0,0014 0,0012 Fe-C-Sb 1 0,056 0,0043 0,025 <0,002 0,0013 Fe-C-Sb2 0,053 0,0085 0,023 <0,002 0,0013 Fe-C-Sb3 0,052 0,0144 0,023 <0,002 0,0014 Fe-C-Sb4 0,094 0,0057 0,027 0,001 0.0011 Alloy Sb Ni C Mn P Fe-Ni-Sb 1 0,049 0,53 0,0035 0,022 <0,002 Fe-Ni-Sb2 0,049 2,85 0,0069 0.024 <0,002 cooling, and then tempering at 780°C for 168 h and water quenching. These two heat treatments were per-formed in flowing wet hydrogen to decrease the bulk carbon concentration below 10 wt.-ppm. The Fe-C-Sb alloys were annealed in flowing dry argon to avoid carbon losses. The samples were homoge-nized at 1060°C for 70 min and air cooled. Afterwards they were recrystallized at 780°C for 2 h and water quenched. Then ali specimens were held at ageing temperatures of 550°C, 600°C, 650°C, 700°C and 750°C for different periods of time, to establish the equilibrium concentration of antimony at the grain boundaries. The time neces-sary for equilibration at each temperature can be as-sessed using an equation proposed by McLean6. AES measurements confirmed that the calculated time was long enough to reach equilibrium segregation; the condi-tions of each exposure are listed in Table 2. Table 2: Conditions for the establishment of segregation equilibria Ageing Temperature/ Ageing Time Exposure Conditions 550°C/600 h vacuum/quenched in vvater 600°C/140 h vacuum/quenched in vvater 650°C/ 50 h flovving argon/quenched in vvater 700°C/ 5 h flovving argon/quenched in vvater 750°C/ 2 h flovving argon/quenched in vvater The amount of grain boundary segregation was to be measured by AES, vvhich is conducted in UHV to avoid surface contamination. After cooling to about -120°C the cylindrical notched specimens were fractured by impact in the UHV chamber of the spectrometer. The fracture surface was then imaged by operating the electron beam in a scanning electron microscope (SEM) mode to distin-guish between intergranular and transgranular areas. Auger spectra were taken from at least 10 individual grain boundary facets using a cylindrical mirror analyzer (CMA) and the results were averaged. The peak-to-peak heights of antimony (454 eV), nickel (848 eV) and carbon (271 eV) were related to the iron peak at 651 eV. The entire analysis of each fracture face had to be completed within approximately 3 h to prevent contamination ef-fects. The operating conditions were as follows: primary beam energy 5 kV, primary beam current 3 x 10"6 A, and primary beam size 10 pm. To estimate the degree of coverage of antimony at the grain boundaries, it can be assumed that antimony is uni-formly distributed on both fracture faces. This supposi-tion was verifted by some AES measurements in which opposite fracture facets vvere investigated7. From LEED studies of surface segregation on Fe-Sb single crystals a calibration factor had been obtained vvhich converts the peak-to-peak height ratio to the degree of coverage8. Supplementary surface analytical methods vvere em-ployed. The binding state of core electrons of segregated antimony was determined by X-ray photoelectron spec-troscopy (XPS), vvhile scanning Auger microscopy (SAM) vvas applied to examine the distribution of segregated elements on grain boundary facets. The fracture type and the mechanical properties vvere investigated using SEM (accelerating voltage 20 kV) and Charpy impact tests (DIN 50115). 3 Results and discussion 3.1 Fe-Sb alloys Typical Auger spectra of transgranular and intergranular areas are represented in Figure 1. On transgranular fracture surfaces of the Fe-0,094 wt.% Sb alloy, no antimony peak vvas observed, since the bulk concentration is belovv the detection limit of the AES method. The oxygen peak is due to adsorption from the residual atmosphere after breaking the sample. The spectrum taken on a grain surface of the same alloy clearly indi-cates the enrichment of antimony vvhich is caused by grain boundary segregation. Figure 2 illustrates that the average coverage of anti-mony at the grain boundaries increases vvith increasing bulk concentration and decreasing equilibration temperature. The scatter of the data indicated by the error bars in one curve is rather large (25% - 30% of the mean value) due mainly to the follovving reasons: a) The segregation of antimony may be strongly de-pendent on grain orientation as indicated by surface segregation studies on Fe-Sb single crystals8. b) The examined areas have different distances and different surface normals to the cylindrical mirror ana-lyzer (CMA). c) The degree of coverage is calculated from measurements on only one side of the intergranular fracture face. It vvas verified by some AES measurements in vvhich opposite fracture facets vvere investigated that the average grain boundary antimony concentration is nearly the same on both fracture faces7. The assumption that an- 100 200 300 400 500 Kinetic Energy [eV] 600 700 100 200 300 400 500 Kinetic Energy [eVj 600 700 .6 AH, ln-- lnxoh = - 1-0 Sb RT a cexs segr ^ _ '-'segr R (D which expresses the relationships between bulk concentration (mole fraction) xsb, temperature T, and degree of coverage 0, at the grain boundaries. The results according to the Langmuir-McLean equation are plotted in Figure 3. The estimation yields the segregation en-thalpy AHsegr. = -19 kJ/mol ± 5 kJ/mol and the segregation entropy ASsegr. = 28 J/mol K ± 6 J/mol K. The free enthalpy of segregation in a-iron can be expressed as follows: AG«gr = -(19 kJ/mol±5kJ/mol) - T(28 J/mol K±6 J/mol K) o\ 18-16-14-12- 10-8 -6 -4 -2 - 0.094.% Sb Fe-Sb 600 650 700 Ageing Temperature [°C] 750 Figure 2: Grain boundary concentration of antimony plotted as a funetion of equilibration temperature for the alloys Fe - 0,012 wt.% Sb, Fe - 0,049 wt.% Sb and Fe - 0,094 wt.% Sb Slika 2: Koncentracija antimona na kristalni meji kot funkcija ravnotežne temperature za zlitine Fe - 0,012 ut.% Sb, Fe - 0,049 ut.% Sb in Fe - 0,094 ut.% Sb Figure 1: Auger spectra of fracture surfaces of Fe - 0,094 wt.% Sb alloy after annealing at 650°C. a) cleavage facet, b) intergranular fracture surface Slika 1: AES spekter prelomnih površin Fe - 0,094 ut.% Sb po žarjenju pri 650°C. a) prelomna ploskev, b) interkristalna prrlomna površina timony is equally distributed is probably not true for each single intergranular area. In spite of the large scatter of the data, a thermody-namic calculation was attempted, applying the Langmuir-McLean equation 6,8 6,6 X 6,4 C 6,2 — ■Zl 6,0 s. 5,8 CD 5,6 5,4 5,2 ■ 0.012 %Sb • 0.049 %Sb * 0.094 %Sb 0,95 1,00 1,05 1,10 l/T [K 'xl0'] 1,15 1,20 1,25 Figure 3: Langmuir-McLean plot of the data in Figure 2 Slika 3: Langmuir-McLeanov diagram podatkov iz slike 2 The segregation enthalpy value is low compared to values for phosphorus (AHsegr = -34 kJ/mol)9 or tin (AHsegr = -23 kJ/mol)10, this indicates the low tendency for grain boundary segregation of antimony in iron. It would be unreasonable in the present thermody-namic calculations to include the AES data for the Fe -0,93 wt.% Sb and Fe - 1,91 wt.% Sb alloys, since un-known antimonides had formed at the grain boundaries. In Figure 4, a typical scanning electron micrograph and the corresponding elemental map for antimony on the same intergranular area of the Fe - 0,93 wt.% Sb alloy indicate star shaped antimonides. In spite of the low tendency for grain boundary segregation, antimony has a strongly embrittling effect. The relationship betvveen the percentage of intergranular fracture and the grain boundary coverage of antimony is demonstrated in Figure 5. With increasing enrichment of antimony at the grain boundaries the fracture mode at 100 <=\ 80 i 1-1 PH S 3 fe 40 (D Figure 4: Intergranular antimonides observed in Fe - 0.93 wt.% Sb after annealing at 650°C; a) scanning electron micrograph. b) corresponding scanning Auger image of Sb Slika 4: Interkristalni antimonidi opaženi v Fe - 0,93 ut.% Sb po žarjenju na 650°C; a) posnetek z vrstičnim elektronskim mikroanalizatorjem, b) vrstični Augerjev posnetek low temperatures (about -120°C) changes from trans-granular to intergranular already at rather low grain boundary concentrations. The influence of antimony segregation on the mechanical properties vvas also studied by Charpy impact testing. The transition temperature determined Tt is a measure of the embrittlement of iron base alloys. Tt is defined as the temperature vvhere half of the difference value betvveen the impact vvork necessary for ductile fracture and the impact vvork for brittle fracture is reached. For the Fe-Sb alloys a shift of the impact transition temperature to higher values is expected vvith in-creasing antimony concentration at the grain boundaries. This supposition is verifted in Figure 6. For each of the tvvo investigated alloys a higher transition temperature is obtained vvith increasing coverage of antimony at the grain boundaries. Hovvever, the Fe - 0,094 wt.% Sb alloy tempered at 750°C has a lovver transition temperature than the Fe - 0,049 wt.% Sb alloy annealed at the same temperature. The observed phenomenon can be ex-plained by the different average grain size of these materials (Fe - 0,049% Sb: 0,21 mm; Fe - 0,094% Sb: 0,08 mm), vvith increasing antimony concentration the grain 0,1 0,2 0,3 0,4 "0,5 Peak-to-Peak Height Ratio [I(Sb)/I(Fe)] Figure 5: Percentage of intergranular fracture versus peak-to-peak height ratio I(Sb)/I(Fe) Slika 5: Odstotek interkristalnega preloma v odvisnosti od razmerja višine vrhov I(Sb)/I(Fe) size decrease vvhich leads to a higher strength of the material. One possible way to explain the embrittling behav-iour of antimoy is to apply quantum mechanical mod-els"12. The main conclusions of these calculations can be summarized as follovvs: The segregated antimony atoms are electronegative vvith respect to the host metal iron. Consequently electronic charge is transferred from iron to antimony. This charge transfer leaves fevver electrons to participate in 100 p ai l— ■D 03 d) O. E C o +-J tn C ni 80 60 40 20 0.049 wt.% Sb 0.094 wt.% Sb 82 °C 52 °C 25 °C 18 °C 0.024 0.035 0.042 0.064 Peak-to-Peak Height Ratio [I(Sb)/I(Fe)] Figure 6: Dependence of the transition temperature on the grain boundary antimony concentration for Fe - 0,049 wt.% Sb and Fe - 0,094 wt.% Sb alloys Slika 6: Odvisnost koncentracije antimona na mejah zm od prehodne temperature za zlitine Fe - 0,049 ut.% Sb in Fe - 0,094 ut.% Sb Figure 8: SEM of a faceted grain boundary in Fe - 0,094 wt.% Sb after annealing at 650°C Slika 8: SEM posnetek facetirane meje v zlitini Fe - 0.094 ut.% po žarjenju na temperaturi 650°C Figure 9: Pore at a grain boundary facet of Fe - 0,094 wt.% Sb after annealing at 600°C; a) SEM. b) corresponding scanning Auger image of Sb Slika 9: Razpoka v kristalni meji Fe - 0,094 ut.% po žarjenju na temperaturi 600°C: a) SEM posnetek, b) odgovarjajoči SAM posnetek Sb surface energy and such pores will intensity the observed embrittlement of the material. 2,5x1 o6 — 2x1o6 e jz I l,5xl06 Sb (segregated) 5x10'" 542 540 538 536 534 Binding Energy [eV] Figure 7: Photolines of pure Sb and segregated Sb in Fe-Sb alIoys Slika 7: XPS krivulje čistega Sb in segregiranega Sb v Fe-Sb zlitini Mg K,, Sb 3d3n Sb (elcmcntal) 537,40 eV the iron-iron bonding and these bonds at the grain boundary will be weakened. XPS measurements on a large area of intergranular fracture of Fe - 0,93 wt.% Sb alloy after annealing at 600°C show that the energies of the Sb 3d electron levels of segregated and pure antimony are distinctly different (Figure 7). The energy shift of about -0,5 eV in com-parision to pure antimony indicates an electron transfer to segregated antimony, as expected in the above model. It is also possible to explain the embrittling behaviour of antimony in another way by taking into consideration that the grain boundaries often are facetted, as illustrated in Figure 8. The segregation of antimony induces a re-construction of the grain faces vvhich results in a de-crease of grain boundary cohesion. On some intergranular areas pores were detected with an average diameter of 2 pm as can be seen in Figure 9. An antimony map recorded for the same area, shovvs an-tirnony enrichment vvithin this pore. Segregated anti-mony certainly favours the formation of such pores since its surface segregation causes a pronounced decrease of 3.2 Fe-C-Sb alloys Samples vvith different antimony and carbon contents vvere investigated to study the effect of carbon on anti-mony grain boundary segregation. The fracture faces of the Fe-C-Sb alloys vvith 0,049 wt.% Sb shovv transgranu-lar fracture caused by the carbon content. The higher cohesion of these materials compared vvith corresponding Fe-Sb alloys is due to the fact that antimony is displaced from the grain boundaries by carbon, according to the equation C(dissolved) + Sb(segregated) = C(segregated) + Sb(dissolved) (2) The mutual displacement of these tvvo elements corresponding to the displacement equilibria in the systems Fe-C-P9 and Fe-C-Sn10, vvas proven for the Fe-C-Sb al-loy vvith 0,094 wt.% Sb, as shovvn in Figure 10. The average grain boundary concentration of antimony de-creases vvith increasing grain boundary and bulk concentration of carbon. Simultaneously the percentage 0.14 | 0.12 .g" 0.10 "H (2 £ 0.08 nb '5 SE 0.06 ■S U 0.041 o "I 0.02 0 ▲ i / C/Fe ............ • ■ / ' /s......................... ■ 700°C --•-- 650°C —i— 600°C /s r ■'•»■'■ ' 1 ■ 40 10 20 30 40 50 Carbon Content [wt.-ppm] 60 Figure 10: Dependence of the Sb and C grain boundary concentrations on the bulk concentration of carbon in Fe - 0,094 wt.% Sb Slika 10: Odvisnost koncentracij Sb in C na mejah zrn od koncentracije ogljika v osnovnem materialu Fe - 0,094 ut.% Sb 20 O 0