Journal of Microscopy,Vol.233,Pt32009,pp.417–422
诗韵整容Received28April2008;accepted17June2008
Five-parameter grain boundary analysis of a grain boundary–engineered austenitic stainless steel
R.JONES∗,V.RANDLE∗,D.ENGELBERG†&T.J.MARROW†
∗Materials Research Centre,School of Engineering,Swansea University,Swansea,SA28PP,U.K.
海清老公资料
†Materials Performance Centre,School of Materials,The University of Manchester,Manchester,
M17HS,U.K.
Key words.Annealing twin,EBSD,grain boundary engineering,
grain boundary plane.
Summary
Two different grain boundary engineering processing routes for type304austenitic stainless steel hav
e been compared. The processing routes involve the application of a small level of strain(5%)through either cold rolling or uni-axial tensile straining followed by high-temperature annealing.Electron backscatter diffraction and orientation mapping have been used to measure the proportions of 3n boundary types (in coincidence site lattice notation)and degree of random boundary break-up,in order to gain a measure of the success of the two types of grain boundary engineering treatments. The distribution of grain boundary plane crystallography has also been measured and analyzed in detail using the five-parameter stereological method.There were significant differences between the grain boundary population profiles depending on the type of deformation applied.
Introduction
The application of carefully designed thermo-mechanical processes can create grain boundary–engineered(GBE) microstructures in austenitic stainless steels resulting in improved resistance to intergranular corrosion and stress corrosion cracking(Shimada et al.,2002;Michiuchi et al., 2006;Jin et al.,2007).Processing typically involves the application of cold deformation followed by annealing,often applying several sequential deformation/annealing cycles (Randle,2004).However,it has been shown that variations in processing routes and parameters can have a dramatic effect on t
he grain boundary misorientation distribution and the distribution of grain boundary plane indices(Randle et al.,2008a).Consequently,the success of GBE and associated property improvement can be greatly affected by quite small changes in processing.美国留学考什么
Correspondence to:V.Randle.Tel:+441792295841;fax:+441792295676; e-mail:v.randle@swansea.ac.uk
Processing for GBE requires primarily that the proportion of  3boundaries,in coincidence site lattice(CSL)nomenclature, is increased(Schwartz et al.,2006).The 3category includes coherent annealing twins,incoherent annealing twins and other interfaces that have the 3misorientation.These different 3types all necessarily have the same misorientation, that is,60◦/<111>,but are distinguished on the basis of their boundary planes:coherent twins have{111}planes, incoherent twins are either symmetrical tilt boundaries with {112}planes or asymmetrical tilt boundaries having a range of low-index planes such as{110}/{114},and other 3 interfaces have general boundary planes.The properties of  3interfaces are dependent on the boundary plane type. For example,coherent twins are almost immobile,whereas incoherent twins are very mobile.Coherent and incoherent twins also have different but complementary roles to play in the development of a GBE microstructure(Randle&Owen, 2006).Forthesereasons,itisclearlyofgreatvaluetoaccessthe crystallogr
aphy of the boundary plane distribution in a GBE microstructure in addition to the misorientation distribution. Until recently,electron backscatter diffraction(EBSD)and orientation mapping delivered routinely only misorientation statistics,with measurement of boundary plane geometry being a specialized application requiring laborious experimental techniques(Randle,1997).In the past few years,a stereological method has been developed to measure the‘five-parameter distribution’,based on EBSD measurements.The five parameters are the five independent variables required to specify the geometry of a grain boundary,that is,the misorientation(three parameters) and the boundary plane(two parameters).The principles and practice of the five-parameter analysis are described in detail elsewhere(Saylor et al.,2003,2004).So far,the five-parameter analysis has been applied to measure the boundary plane and misorientation distribution in several minerals and metallic-based materials,including GBE copper and nickel(Saylor et al.,2003;Randle et al.,2008b).
C 2009The Authors
孟茜个人资料
Journal compilation C 2009The Royal Microscopical Society
苹果系统ios14418R.JONES ET AL.
In the present work,application of the five-parameter analysis to a GBE austenitic stainless steel is re
ported.The main focus of the investigation is to compare the effect of deformation mode,namely,uni-axial(tensile)deformation versus cold rolling,on the final microstructure and grain boundary distribution when all other processing parameters are kept constant.
Experimental procedure
Specimens of as-received(AR)mill annealed type304 austenitic stainless steel(18.18%Cr,8.61%Ni,0.05%C)were subjected to two different single-step strain anneal processes in order to produce GBE microstructures.One set of specimens underwent deformation by cold rolling in the form of a5% reduction in thickness via a single rolling pass.A second, identical set of specimens were subjected to uni-axial tensile deformation involving a5%elongation of the original gauge length(20mm)at an extension rate of1mm min–1.Both sets of specimens were subsequently annealed at1050◦C.The cold-rolled specimens were annealed for time periods ranging from 5min to24h,and the tensile specimens were annealed from 5min to5h.Orientation maps of all samples were obtained using an EBSD system from HKL Technology with step sizes ranging from1μm to2μm.
Microstructure and misorientation data from the AR specimen and all the deformed and annealed sp
ecimens were analyzed using both standard HKL software(Channel5) and also by VMAP software(Humphreys,2001).Number and length proportions of 3n(n≤3)CSL boundaries were collated,with boundaries being categorized according to the Brandon criterion.Automated mean linear intercept grain size measurements were obtained,both including and excluding twin boundaries.Microstructures from the specimens subjected to5%deformation by both tensile deformation and cold rolling and annealing at1050◦C for 30min,plus the AR specimen,were fully characterized using the five-parameter analysis software obtained from Carnegie Mellon University,Pittsburgh,USA,in order to investigate variation in the distribution of grain boundary planes.Full details are available elsewhere(Saylor et al.,2003, 2004).For the five-parameter analysis orientation data from approximately40000grains was collected at a step size of 2μm from all three samples.Data from the five-parameter analysis were displayed as density distribution in‘multiples of a random distribution’,MRD,in stereographic projection with a10◦resolution.
Results and discussion
Effect of thermo-mechanical processing parameters
on GBE microstructure development
The proportion of 3n boundaries was calculated both in terms of number frequency and proportion of
total Fig.1.Comparison of tensile(TEN)and cold-rolled(CR)GBE processing routes in terms of 3n boundary number fractions and grain size (including twin boundaries). 3n boundary number fractions are displayed on the chart using bars and grain sizes are displayed using symbols.
grain boundary length to gain a measure of the success of the thermo-mechanical treatments in producing GBE-modified microstructures.Figure1compares the 3n number frequencies produced by both the tensile processing route and the cold-rolling processing route together with grain size(including 3boundaries)measurements.For comparison,the number of frequencies of 3n boundar
ies in the microstructure of the AR specimen were23.6%and4.2% for 3and (9+27)boundaries,respectively.The average grain size of the AR specimen including twin boundaries was14.4μm.It was observed that both processing regimes produced similar trends in 3n proportions as a function of annealing time.However,for annealing times greater than 10min,the cold-rolling treatments produced noticeably larger proportions of 3boundaries and also 9and 27boundaries (i.e.second-and third-order twins).Maximum proportions of 3n boundaries were reached after an annealing time of60min.Longer annealing times resulted in reduced levels of 3n boundaries.These observations support previous reports about the development of grain boundary network characteristics after low-strain processing(Engelberg et al., 2008).Similar trends were observed when length proportions of 3n boundaries were considered.The increases in 3n boundaries were accompanied by significant grain growth when cold-rolling processes were used.Considerably less grain growth occurred for the processes involving the application of a tensile strain.
The success of the GBE treatments was also studied by monitoring how the 3n boundaries disrupted the random grain boundary network. 3n boundaries were removed from EBSD orientation maps so that only random grain boundaries remained(Fig.2).It was observed that the microstructures produced by cold rolling were more successful in breaking up the random grain boundary network th
an those produced by tensile straining.This was attributed to the increased levels of 3n boundaries present in the cold-rolled microstructure,
C 2009The Authors
Journal compilation C 2009The Royal Microscopical Society,Journal of Microscopy,233,417–422
FIVE-PARAMETER GRAIN BOUNDARY ANALYSIS
419
Fig.2.Distribution of 3n boundaries and break-up of random boundary network after GBE processing.(a),(b)and(c): 3boundaries red, 9boundaries blue, 27boundaries yellow,random boundaries black.(d),(e)and(f): 3n boundaries removed.(a)and(d)are from the as-received(AR)specimen,(b) and(e)are from the tensile(TEN)deformed specimen and(c)and(f)are from the cold-rolled(CR)specimen.
in particular the high levels of 9and 27boundaries produced by twin–twin interactions.The largest break-up of the random GB network was observed in the cold-rolled specimens annealed for30and60min corresponding to the greatest proportions in 3n boundaries.The largest break-up observed in the tensile samples was observed after a30-min anneal.However,there remained noticeably more connectivity of the random GB network than in the two aforementioned cold-rolled specimens.Figure2displays orientation maps(including and excluding 3n boundaries) of the microstructures produced by5%cold rolling and tensile straining followed by annealing at1050◦C for30min, together with the AR microstructure.The increased boundary network break-up in the specimen processed by cold rolling (Fig.2(f))is clearly seen,even though there is an increase in grain size.Both the 3n boundaries proportions and the break-up of random GB network connectivity suggest that the cold-rolling treatments are more successful in producing GBE-modified microstructu
res than tensile treatments.
Five-parameter analysis
One specimen from each processing regime was selected for five-parameter analysis to investigate the distribution of GB planes within the GBE-modified microstructures.For both deformation routes,the specimen that had been deformed by 5%and annealed at1050◦C for30min was selected.These represented microstructures that produced a combination of high 3n boundary proportions and large break-up of random GB network connectivity within both processing regimes.An AR sample was also analyzed in order to observe the changes in grain boundary plane distribution brought about by GBE processing.The three specimens analyzed will be referred to as AR,TEN(tensile deformed specimen)and CR(cold-rolled deformed specimen).
经典幽默
Figure3shows the grain boundary plane distribution for the entire interface population,for each specimen.There are maxima at{111}in each case,which correspond mainly to coherent twin boundaries and also to non- 3boundaries, which have a{111}on one or both sides of the interface.The density of{111}planes,in terms of MRD,has increased for the GBE CR specimen but has decreased slightly for the GBE TEN specimen.
The distribution of grain boundary planes from each specimen was plotted stereographically in units of‘multiples of a random distribution’(MRD)for the<110>and<111> low index misorientation axes.The five-parameter data were analyzed by sectioning the data in10◦increments of misorientation angle for each misorientation axis,thus allowing the presence of tilt and twist type boundaries to be identified.For the purposes of convenient data representation, the entire misorientation axis distribution for a particular family of directions is folded into a single direction,that is,here <111>is represented by[111]and<110>is represented by [110].Table1displays the 3n boundary length proportions of the three samples selected for five-parameter analysis to aid comparisons with the plots.
Figure4shows the distribution of grain boundary planes for misorientations of60◦/[111],that is, 3,for each specimen. All specimens displayed strong maxima at(111)planes corresponding to{111}coherent twin boundaries,due to the large proportions of 3boundaries in all specimens.The
C 2009The Authors
Journal compilation C 2009The Royal Microscopical Society,Journal of Microscopy,233,417–422
420
R.JONES ET AL
.
Fig.3.Distribution of grain boundary planes in the entire sample population,expressed as ‘multiples of a random distribution’(MRD)in stereographic projection for (a)the AR specimen,(b)the TEN specimen and (c)the CR specimen.Table 1.Proportions of  3n boundaries in the as-received specimen (AR)and specimens deformed 5%by either tensile deformation (TEN)or cold rolling (CR)and annealing at 1050◦C for 30min.
AR
TEN CR  3Length proportion %40.453.262.1 9Length proportion %  2.03  2.93  6.03 27Length propo
rtion %
0.589
0.712
3.4
spread around (111)corresponds to non-coherent and other,general  3s.There is more spread around the peak in the TEN sample than in the other two.It can be seen in Fig.4that the TEN sample displayed a peak MRD value at (111)of 475,which is approximately half that of the AR sample despite an increase of 13%in  3length fraction after TEN processing (Table 1).The CR sample,however,displayed a peak approximately 1.25times that of the AR sample (corresponding to a change in 22%in  3length fraction).The CR sample also had a peak that was more than 2.5times that of the TEN sample.This represents a very significant difference in the twinning behaviour of the two GBE processing routes.Whereas both routes resulted in increased levels of  3boundaries (53%and 62%for TEN and CR,respectively),there were more coherent twin boundaries when cold-rolled deformation was used,as evidenced by the increase in the
{111}planes peak in Figs 3(c)and 4(c).The additional coherent twins in the CR specimen indicate that more new twinning occurred,which was facilitated by the extra increase in grain size,that is,sustained grain boundary migration,in the CR sample compared with the TEN sample (Fig.1).
Other misorientation angle classes,that is,10◦–50◦,on the (111)misorientation axis showed near random plane densities,with maximum MRD peak values of less than two in all cases.There was no consistent pattern for the distribution of boundary planes,although in all specimens there was a small peak representing {111}twist boundaries for some angle ranges.
Grain boundary planes in the <110>misorientation category showed a strong tendency to inhabit the [110]tilt zone,especially in the 30◦and 40◦sections.Figures 5and 6display the distribution of grain boundary planes for misorientations of 30◦and 40◦,respectively,about [110]for the three specimens.The 30◦/[110]plots (Fig.5)indicated a preference for asymmetrical tilt boundaries along the [110]tilt zone in all specimens.Increased MRD values were observed after GBE treatments.However,the CR produced a considerably higher peak than the TEN sample (26MRD compared with 10MRD).Some of these extra asymmetrical tilt boundaries added as a consequence of GBE are attributed to the increase in  27a (31.6◦/<110>)boundaries present in the GBE CR
sample.
Fig.4.Distribution of grain boundary planes for misorientations within 10◦of 60◦/<111>,expressed as ‘multiples of a random distribution’(MRD)in stereographic projection for (a)the AR specimen,(b)the TEN specimen and (c)the CR specimen.The [111]axis is marked on one of the plots with a triangle.
C  2009The Authors
Journal compilation C
2009The Royal Microscopical Society,Journal of Microscopy ,233,417–422
FIVE-PARAMETER GRAIN BOUNDARY ANALYSIS
421
Fig.5.Distribution of grain boundary planes for misorientations within10◦of30◦/<110>,expressed as‘multiples of a random distribution’(MRD)in
stereographic projection for(a)the AR specimen,(b)the TEN specimen and(c)the CR specimen.The[110]axis is marked on one of the plots with a
circle.
Fig.6.Distribution of grain boundary planes for misorientations within10◦of40◦/<110>,expressed as‘multiples of a random distribution’(MRD)in stereographic projection for(a)the AR specimen,(b)the TEN specimen and(c)the CR specimen.The[110]axis is marked on one of the plots with a circle.
The40◦/[110]plots(Fig.6)indicate a striking difference between the two GBE specimens.The TEN specimen displayed a distribution similar to the AR specimen,suggesting the favouring of asymmetrical tilt boundaries along the[110] tilt zone but with a slightly higher peak after GBE processing (MRD=8compared with MRD=6).This is similar to the planes distribution for the30◦/[110]section shown in Figs5(a)and(b).By contrast,the CR specimen has a strong maximum(MRD=16)at(114),which indicates a preponderance for{114}symmetrical tilt boundaries associated with the 9misorientation.A diffuse distribution remains along the[110]zone indicating that some asymmetrical tilt boundaries are also still present.The higher maximum MRD value for the GBE CR specimen compared with that for the TEN specimen is attributed to the extra proportions of 9(38.6◦/<110>)boundaries present in the CR specimen (6.0%CR,2.9%TEN).The doubling in length fraction of  9boundaries does not,however,account for the strong appearance of{114}symmetrical tilt boundaries.Since the energyofthe{114}symmetricaltiltboundaryissimilartothat of some asymmetrical t
ilt boundaries in the 9system(Merkle &Wolf,1992;Gokon&Kajihara,2008),the appearance of the{114}peak is not directly associated with energy minimization.We can speculate that 3variant selection, which in turn affects the crystallography of the 9produced when two 9s meet(Forwood&Clareborough,1984),might operate differently depending on the deformation mode.It may also be the case that some 9asymmetrical tilt boundaries, but not{114}symmetrical tilt boundaries,have dissociated to give two 3boundaries(Gokon&Kajihara,2008).A further analysis will be carried out in the future.
Finally,the findings from the five-parameter analysis indicate that CSL boundary statistics alone do not provide a satisfactory measure and understanding of GBE processes and emphasize the importance of examining the grain boundary plane distribution in addition to the misorientation distribution and microstructure characteristics. Conclusions
Two GBE processing routes involving the application of a small (5%)strain either by cold rolling or uni-axial tensile straining followed by high-temperature annealing were applied to specimens of type304austenitic stainless steel.The effects on the microstructure,particularly in terms of the five-parameter grain boundary analysis,from the two processing routes were compared and the following conclusions were drawn:
1.Cold rolling was more successful than tensile straining in
developing a GBE microstructure in terms of both increased proportions of 3n boundaries and also the disruption of the random grain boundary network.
C 2009The Authors
Journal compilation C 2009The Royal Microscopical Society,Journal of Microscopy,233,417–422