摘要:

Electron Microscope Annealing Study of QuenchedUranium DioxideBY K. H. G. ASJBEE*Berkeley Nuclear LaboratoriesC.E.G.B. Glos.Received 3rd June, 1964Thin foils of uranium dioxide prepared from melts quickly cooled in an arc-furnace containextensive stacking faults and defect tetrahedra. These foils have been annealed during observationby transmission electron microscopy to produce hexagonal dislocation networks and the breaking-upof the tetrahedra. Analysis of the networks show that they are two-dimensional and are comprisedof dislocation lines lying in (21 1) planes. The dislocations are extended on planes containing theirrespective stacking fault displacement vectors, the stacking fault planes at each node being joinedby stair-rod dislocations. The defect tetrahedra are crystallographically identical to those observedin quenched gold 1,2 and NiCo.3 The annealing treatment causes these tetrahedra to break-upby the nucleation and gIide of dislocations.The possible effect of dislocation networks and tetra-hedra on reactor behaviour has been discussed.Uranium dioxide is face-centred cubic (fluorite structure with lattice parametera = 5.4704A) and has a melting point of about 2,800"C. The slip systems, reportedby Rapperport and Huntress,4 are { 1 lo> directions on {loo}, (1 10) and { 11 l } planes.An undesirable feature of its behaviour as a nuclear fuel is that UO2 has a poorresistance to thermal shock ; in pellets, radial cracks are propagated during start-upand shut-down. On a sub-microscopic scale, transmission electron microscopestudies 5 ~ 6 have shown that UO2 which has been quickly cooled contains intrinsicstacking faults (with displacement vector R = 116 (211)) and defect tetrahedra.Acharacteristic and important feature of many stacking faults in U02 is that they arenot confined to a single plane. For example, in fig. 1 stacking fault A wandersthrough a wide range of planes giving itself a wavy appearance. This is due 5 ~ 6to cross-slip of the leading partial dislocation on to any plane containing the stack-ing fault displacement vector.The present work reports the effects of heating quenched U02 during observationin an electron microscope. Two new features have been observed, namely, theformation of hexagonal networks of extended dislocations and the breaking-up ofdefect tetrahedra by the creation and glide of dislocations.EXPERIMENTALSintered pellets of U02 were purchased from the U.K.A.E.A.(Risley) and were meltedunder one-third of an atmosphere of argon. The melts so produced were rapidly cooledby switching off the current. This gave an estimated initial cooling rate of about 45OoC/secdown to red heat after which several minutes were required to cool to room temperature.Slices about mm thick were cut from the resulting buttons of U02 and polished down tothin foils in a jet of hot nitric acid. These were examined in a Philips E.M. 200 and annealedduring observation by decreasing the strength of the first condenser lens for various com-binations of condenser aperture sizes. Selected area diffraction patterns were taken fromareas 1 micron in diameter.* present address: H.H. Wills, Physics Laboratory, University of Bristol.N 30310 ANNEALING OF QUENCHED URANIUM DIOXIDEThe original pellets contained about 0.025 % by weight of impurities (mainly iron andsilicon) and had an oxygen/uranium ratio of 2.01.RESULTS AND DISCUSSIONHEXAGONAL DISLOCATION NETWORKSAnnealed foils frequently contained dislocation networks in which perfectlystraight dislocations join each other exclusively at 3-fold nodes giving the appearanceof an arrangement of regular hexagons. An example is given in fig. 2. The com-monly occurring precipitates also produced by this treatment give diffraction spotson selected area diffraction patterns at points 1/3 the spacing of UO2 1 I 1 spots.Allthe networks seen so far have been discontinuous ; in fig. 2, for example, no disloca-tion line extends through more than about three hexagons. This means that thedislocations are arranged in surfaces which divide the crystal into cells about *,Yapart. Similar observations have been made in other materials, e.g., by Hedgesand Mitchell,7 who used dislocation decoration with photolytic silver in AgBr, andby Whelan, Hirsch, Horne and Bollman 8 and later Whelan 9 in their transmissionelectron microscope studies of stainless steel.CRYSTALLOGRAPHIC ANALYSISAn analysis has been made using stereographic projection of the orientation ofthe dislocation lines in fig. 2. The foil plane is (101) and the projections in this planeof dislocation lines X, Y and 2 have been transferred, in fig.3, to a cubic standardFrG. 3 . 4 u b i c standard stereographic projection showing the projected directions of dislocationlines X, Y and 2 in fig. 2. F.N. denotes the foil normal.stereographic projection. The actual dimtiads of X, Y and 2 lie ia those planesnormal to the foil plane which htersect the latter in these three directions, i.e.,X, Y and 2 lie in the planes whose traces afe drawn as great circles in fig. 3 throughF.N. and X, Y and 2 respectively. It is evident from fig. 3 that, other than the foilnormal which is an impossible solution, there are no { 1 10> directions in these planesFIG. 5.-Thompson 11 tetrahedron notationfor Burgers vectors of dislocations in f.c.c.crystals.FIG. 6.-Defect tetrahedra in quenchedu02.FIG.8.-(a) A vacancy located near the core of a Shockley partial dislocation AB with Burgers vectorbp, and (b) the annhilation of this vacancy by the Shockley partial shears bp ' and bp ' I .FIG. 9.-Growth of a stacking fault tetrahedronby vacancy absorption.FIG. 10.-Defect tetrahedra annealingby emitting dislocations X FIG. 1 .-Asquenched U02 containing FIG. 2.-Hexagonal arrangements ofstacking faults ; g = 220. dislocations in quenched and an-nealed uranium dioxide. The foilnormal is 101 and g = 31 1.FIG. 4.-The same area (located by precipitates A) showing hexagonal dislocation networks underdifferent diffraction conditions. The g values for (a), (b) and (c) are 202, 313 and 020 respectivelyK .H . G . ASHBEE 311This means that X, Y and 2 are not pure screw Q [ 1101 dislocations. The networksare not, therefore, simple twist boundaries of the kind discussed by Frank.10From fig. 3 it is seen that the following possibilities exist for the true directionsof X, Y and 2. (Directions normal or parallel to the foil plane are not possible.)x: [i3i], [121], [111], [212], [313], p i q , [212], [iii], [121], [iSi],Y : 13211, [ i ~ ] , [012], [ m ] , pii], [zio],z: [321], [i23], [oi2], [ii3], [3ii], [210].If X , Y and 2 all belong to the same family of directions, then the only possiblesolution is that they are parallel to (31 1) directions. This solution has two importantfeatures. (i) Since the angles between (311) directions are not multiples of n/3,each hexagonal net must be corrugated, i.e., not confined to a single plane. (ii) Am-lysis of the possible (loo), { 110) and (1 11) slip planes and 8 { 110) Burgers vectorsfor a three-fold node of whole dislocations lying parallel to (311) directions showsthat the sum of the Burgers vectors is not zero.It may well be that X , Y and 2 are not members of the same family of directions.However, since Y and 2 in fig.2 have equal projected lengths which are parallel to(1 11) directions, they must belong to the same family of directions. With the pro-viso that all three dislocation lines are co-planar, there are two solutions, namelyalso possible but unlikely since [212] is only - 19" from F.N.and would not givethe observed projected length of X.) Analysis of the slip planes and directionsinvolved confirms that the sum of the three whole Burgers vectors at the node is zero.x [ i i i ] , ypio], z[oi2], and ~ [ l l f ] , ~[012], 2[210]. (x[212], y[ii3], z [ m ] isSTACKING FAULT CONTRASTFig. 4(a), (b) and (c) show the same area, located by precipitates A, under differentdiffraction conditions. In fig. 4(a) the dislocation nodes are eharacterized by theirassociated stacking fault contrast, e.g., at areas B. The reciprocal lattice vector tothe plane producing this contrast is g 202 which is consistent with the 1/6 [211]stacking fault displacement vector previously reported 59 6 for UOz. In fig. 4(b) and(c), where the g values are respectively 313 and 020, the stacking fault contrast is notso well resolved. Instead, many dislocations show double contrast in 4(b) anddisappear completely in 4(c) (compare, for example, dislocations C in fig.4(a), (b)and (c)).The fringe contrast at B in fig. 4(a) is sufficiently well defined to establish that,at a given node, at least two different stacking fault planes are occupied. There is,however, no evidence for contracted nodes in any of the foils so far examined.DISCUSSIONAs pointed out by Frank,lo only two sorts of dislocations are required to form ahexagonal network in a f.c.c. crystal, the third member being the resultant of inter-action between the other two. The reaction is &[liO]+~[Olf] = +[lOi] and hasbeen re-interpreted by Whelan 9 in terms of the short-range interaction between theleading partials of the intersecting and intersected dislocations.This showed thatthe resulting network consists of extended and contracted nodes on a single (111)plane. However, it is difficult to see how an interaction between the leading partialscould degenerate into a node configuration involving two or more different stackingfault planes. Consequently, the present interpretation will be based on degenerationinto extended dislocations after interaction between whole dislocations, i.e., if theinteracting dislocations a e extended, they constrict before interaction312 ANNEALING OF QUENCHED URANIUM DIOXIDEIf' the component dislocation lines are parallel to (31 1) directions, the networksmust represent corrugated surfaces.Such a configuration is unlikely since the dis-location line tensions will be unbalanced. Further, it has been mentioned abovethat the sum of the whole Burgers vectors at a node of such dislocations is not zero.It is evident, therefore, that the dislocations do not lie parallel to (31 1) directions.Consequently, the true directions of X , Y and 2 must be [lil], [210] and 10121 re-spectively lying in the (121) plane (or [lll], [012] and [210] respectively lying in theplane (I Z 1)).In table 1, possible whole Burgers vectors for X, Y and 2 are deduced from their(loo), (1101 and/or (11 1) slip planes. Previous work 5, 6 has shown that in U02 thestacking fault plane is undefined as long as it contains the stacking fault displace-ment vector.Consequently, in fig. 2, each whole dislocation is able to dissociatein a plane containing its length and its 1/6 (21 1> stacking fault displacement vector.The trailing partial is, of course, fixed along the length of the original whole dis-location. The loci of the poles of planes containing dislocation lines X, Y and 2are drawn in fig. 3 as great circles 90" from 111, 210 and 012 respectively. Possiblestacking fault planes (i.e., planes containing (21 1> directions) are (XOl), (01 1) and(110) for X, (120) for Y, and (021) for 2.TABLE POSSIBLE SLIP PLANES AND BURGERS VECTORS FOR DISLOCATIONS X, Y AND ZIN FIG. 2x[iiij ~[2i01 z[oi21slip plane (iio), (oil), (101) (ow (1Wb M i 0 1 , ~ 0 1 1 1 , 1311011 ~ ~ 1 1 0 1 , + ~ i i o i f ~ o i 11, h*[oi 1 IOne self-consistent solution for the Burgers vectors, R-vectors and stacking faultplanes involved in fig.2 is :X, ~[101]+1/6[2~1]+1/6[112], R = 1/6[2il] on (Oll),2, Q [Oll] not dissociated.Y, Q [iio]-,i/6[~ii]+i/6[i2i], R = 116 pi11 on (120),For an extended node involving two or more different stacking faults, the linesof mutual intersection of the stacking fault planes must represent zero stacking faultdisplacement, i.e., a stair-rod dislocation is required at the line of intersection of eachpair of stacking fault planes. Such a stair-rod can be seen at D in fig. 4(a).It would evidently be very difficult to deform U02 after a well-developed hexa-gonal network has been built up. This is true whether subsequent deformation in-volves the glide of new dislocations through the hexagons, or the breakdown of thenetworks themselves.In fact, it has been found impossible to move with thermalstressing any dislocations in or near networks by heating, even to such temperaturesthat the specimen evaporates. Consequently, it is likely that in a crystal containingnetworks of the kind observed in the present work, strains resulting from large ap-plied stresses will be relieved by the propagation of cracks rather than the glide ofdislocations. Such stressing conditions result from the large temperature gradientswhich occur during reactor start-up and shut-down.DEFECT TETRAHEDRAPRELIMINARY OBSERVATIONSEarlier work 5 9 6 has shown that arc-melted UOz cooled quickly from its meltingpoint so as to produce a quenching effect contains areas of contrast having crystalloK .H . G . ASHBEE 313graphically well-defined triangular outlines. Each triangle has one side where thecontrast is darker than over the remainder of its area suggesting they are in fact threedimensional defects, probably tetrahedra. Three possible explanations for the tetra1hedra are that they are cavities, e.g., etch pits or pores, precipitates or tetrahedra-arrangements of stacking faults. The cavity explanation is ruled out for the follow-ing reasons : (i) in bright field, a cavity would show up as an area of lighter contrastthan the background; (ii) for an open cavity such as an etch pit, fringes would beexpected lying parallel to the edges defined by the foil surface; fig.6 and 10 showthis not to be the case ; (iii) foils prepared in exactly the same way from slowly cooledUOz do not contain tetrahedra; (iv) since the contrast at the tetrahedra varies withs, the deviation from the exact Bragg condition, it must be due to diffraction andnot absorption.The possibility that the tetrahedra are particles of, say, uranium as a result ofnon-stoichiometry or a higher oxide precipitated during quenching, has not beenconfirmed by electron diffraction. The sidc of the average tetrahedron is - 600Aand, in regions of crystal containing a large density of tetrahedra, there would besufficient precipitate to give fairly strong electron diffraction. Consequently themost likely explanation seems to be that the tetrahedra are three-dimensional stackingfault configurations similar to those observed by Silcox,l and Silcox and Hirsch 2 inquenched gold, and by Mader, Seeger and Simsch 3 in quenched NiCo.The present experiments on quenched UOa have been made to study the natureand crystallography of the tetrahedra, and also their behaviour during annealing inthe electron microscop: beam.The triangular outlines vary in size, up to a maximum side of N 700A.Someof this size variation is due, no doubt, to intersections of tetrahedra with the foilsurface. An estimate of the foil thickness, and hence of the likelihood of such inter-sections, can be obtained from the projected widths of stacking faults and a know-ledge of the extinction distance to, values for which have been calculated for somecommonly occurring reflections, table 2.Thus, in fig. 1 for which g = 220, thefoil thickness is about 1500 A. This means that large numbers of tetrahedra willbe intersected by the foil surface. At the present time, it has not been possible toprepare foils with orientations sufficiently far from (1 1 1) to see sections of tetrahedrahaving shapes other than distorted triangles.TABLE 2.-u02 EXTINCTION DISTANCES. RELATIVISTIC CORRECTIONS HAVE BEEN MADE TOTHE ELECTRON WAVELENGTH AND TO THE ATOMIC SCATTERING FACTORS FOR ELECTRONShkl 220 31 1 331to (A) 290 445 590CRYSTALLOGRAPHYIt is convenient when discussing stacking fault tetrahedra in f.c.c. crystals to usethe Thompson 11 tetrahedron notation for Burgers vectors.Thompson’s tetrahedronshown in fig. 5 is comprised of the four (I 11) planes, and has its edges (AB, BC, CDand AD) corresponding to the Burgers vectors -& (1 lo). The mid-points of the facesare labelled a, /I, y and 6 and A6 etc. represent Shockley partial Burgers vectors,a6 etc. stair-rod Burgers vectors of the type 1/6 (110), and ccA etc. Frank partialBurgers vectors.Fig. 6 shows a number of well-defined tetrahedra at high magnification. Weshall consider the crystallography of the defect whose edges are labelled a, b, c, d, eand f. To find the crystallographic directions to which these edges correspond, itmust be remembered that, on the micrograph, each edge is projected onto the foi314 ANNEALING OF QUENCHED URANIUM DIOXIDEplane * and that the observed projection merely defines the plane perpendicular tothe foil plane in which the edge lies. The foil normal (F.N.) in fig.6 is 321 and, infig. 7, the projections of edges a, b, c, d, e andf(d, e andfare almost co-linear) havebeen transferred to a cubic standard stereographic projection. Great circles havebeen drawn corresponding to the planes defined by these projections and whichcontain the edges of the tetrahedron. The only rational indices evident in fig. 7 forthe edges are listed in table 3.TABLE 3.-THE EDGES AND FACES OF THE TETRAHEDRON ANALYSED IN FIG. 6edge planes intersectedSince the tetrahedron has its edges parallel to (110), it must be regular with(111) faces. The indices of the individual faces deduced from fig.7 are given intable 3.FIG. 7.-Standard stereographic projection of U02 showing the projections of edges a, b, c, d, eand f of the tetrahedron labelled in fig. 6.THE ORIGIN OF TETRAHEDRADefect tetrahedra were first observed, by Silcox 1 and SiIcox and Hirsch,Z in goldquenched into iced brine from 950°C and subsequently aged at 100°C. These authorssuggested the following model to explain the formation of tetrahedra. Referring to the* This is true ody if the foil is normal to the electron beam as for fig. 6. For a foil inclined tothe beam, the micrograph shows the projection in the plane defmed in reciprocal space by the cor-responding selected area diffraction patternK . H . G. ASHBBE 315Thompson tetrahedron, fig.5, suppose a triangular monolayer of vacancies con-denses on plane a with sides parallel to BC, CD, and DB. If this monolayer collapses,a triangular-shaped prismatic Frank partial dislocation loop is formed with Burgersvector aA, and each side of the triangle dissociates into a stair-rod dislocation anda Shockley partial dislocation thus :MA = @+PA,aA = ay+yA,aA = a6+6A.The three Shockfey partials PA, yA and 6A bow out on their respective slip planessince they are repelled by the stair-rods, and attract each other in pairs to form stair-rods along DA, BA and CA according to the reactions :PA+& = BY,dA+AP = sp.yA+A6 = yS,The final defect is thus a tetrahedron of intrinsic stacking faults, the edges of whichare low energy stair-rod dislocations.The stacking fault tetrahedra in quenched UO2 can be explained 5 with an identicalmodel except that the original disc of vacancies must contain oxygen and uraniumvacancies in stoichiometric numbers and occupies three adjacent (1 11) planes.As a result of annealing studies on quenched gold, de Jong and Koehler 12 doubtthe validity of the Silcox and Hirsch 1, 2 model for the formation of stacking faulttetrahedra, and suggest the following as a more satisfactory interpretation.Thedefect tetrahedron of intrinsic stacking faults with stair-rod dislocations along itsedges is nucleated by the collapse of six vacancies, and grows by the addition ofvacancies to its edges. To see how this growth occurs consider, in fi.8. 8, a vacancysat on the row of atoms just outside the core of a Shockley partial dislocation, AB.The vacancy is removed by the glide of the complete row of atoms towards the stack-ing fault in the directions indicated by ()p’ and b;.The net result is that, by absorb-ing one vacancy, the Shockley partial has glided backwards by one Burgers vector.As far as an approaching vacancy is concerned, the stair-rod dislocation along theedge of a tetrahedron is two Shockley partials, the two into which it could be re-solved. Consequently, both the tetrahedron faces defined by the stair-rod can growby the vacancy absorption mechanism. If this process occurs at all six edges of thetetrahedron, the resultant configuration will have the appearance of four tetrahedraas shown in fig.9. Further growth can then occur at the re-entrant corners, e.g.,AB, of fig. 9 to form a large tetrahedral defect. For UO2, the de Jong and Koehler 12model requires the absorption of stoichiometric numbers of oxygen and uraniumvacancies.Occasionally, defect prisms have been observed6 in UOz which apparently havesquare bases, an example of which is inset in fig. 6. It has not been possible to inter-pret these as arising from intersections with the foil surface. Consequently, it isconcluded that these are not artefacts and are, in fact, defect pyramids having squarebases. Another variation sometimes observed is pairs of tetrahedra as in fig. 6.ANNEALING BEHAVIOURIn gold, stacking fault tetrahedra disappear 1 after annealing at ~700°C.Inthe present work, UO2 foils containing defect tetrahedra were annealed during ob-servation in the electron microscope. This caused the tetrahedra to break up byemitting dislocations as shown in fig. 10. An outline of one of the tetrahedra i316 ANNEALING OF QUENCHED URANIUM DIOXIDEfig. 10 has been drawn alongside a tetrahedron to aid its resolution, and each dis-location emitted is labelled X. It is evident that a trail possibly of stacking fault isleft behind each dislocation and, if the tetrahedra are stacking fault configurations,this suggests the following annealing mechanism. Since the fault on each face of atetrahedron is intrinsic in nature it may be corrected by the nucleation and growthin its plane of an interstitial prismatic dislocation loop.Direct evidence for theexistence of prismatic loops on stacking faults has been previously reported inTi02 13 and UOz. 5 In UOz, oxygen interstitials introduced by non-stoichiometrysuggest an ideal nucleus for such a loop, the uranium interstitials needed to com-pletely remove the fault being provided by the emission of vacancies. Under thethermal stressing conditions during which this process occurs, the slip traces infig. 10 show that some of the dislocations X are able to move away from the tetra-hedra by, for example, the mechanism illustrated in fig. 8.DEFECT TETRAHEDRA AS TRAPS FOR FISSION GASESThere is little doubt that the oxygen and uranium vacancies required to form thedefect tetrahedra observed in UO2, by either the Silcox and Hirsch 1 9 2 or de Jong andKoehler 12 methods, result from the in-quenching of a supersaturation of vacancies.To nucleate a tetrahedron, however, stoichiometric numbers of these vacanciesmust diffuse together, and the time needed to do this is thought to be that takenfor the U02 to cool through the last stage of its arc-melting, is., from -1000°C toroom temperature.Now, in-pile U02 has an alternative source of vacancies re-sulting from irradiation damage, and in-pile temperatures and times will be sufficientto allow the diffusion required to nucleate tetrahedra. Consequently, UOz can beexpected to contain tetrahedra under reactor conditions.There is a strong possibility that tetrahedra could constitute traps for fission gasatoms. An apex of a tetrahedron is a point of dilation and a gas atom here wouldbe very strongly bound.The stair-rod dislocations along the edges of a tetrahedronare also centres of dilation and could capture gas atoms. In this context, MacEwanand Stevens14 have observed decreases in the diffusion coefficients of xenon inU02 with increasing radiation exposures. These authors find that their results canbe interpreted only by assuming the presence of both existing and irradiation in-duced trapping sites on a submicron scale.The author would like to thank Dr. G. K. Williamson for his criticisms of themanuscript, and the Central Electricity Generating Board for permission to publishthis work.1 Silcox, Ph.D. thesis (Cambridge, 1960).2 Silcox and Hirsch, Phil.Mag., 1959, 4, 72.3 Mader, Seeger and Simsch, 2. Metallk., 1961,52,785.4 Rapperport and Huntress, U.S. Atomic Energy Comm. Rep., 1960, NMI-1242.5 Ashbee, Pruc. Roy. SOC. A, 1964,280,37.6 Ashbee, 3rd Eur. Reg. Con$ Electron Microscopy (Prague, 1964), in press.7 Hedges and Mitchell, Phil. Mag., 1953,44,223.8 Whelan, Hirsch, Home and Bollman, Proc. Roy. SOC. A, 1957,240,524.9 Whelan, Proc. Roy. SOC. A, 1959, 249, 114.10 Frank, Report Cunf. Defects in CrystaZIine Soiids (London), Physic. SOC., 1955, 159.11 Thompson, Proc. Physic. SOC. B, 1953, 66,481.12 de Jong and Koehler, Physic. Rev., 1963, 129,49.13 Ashbee, Smallman and Williamson, Proc. Roy. SOC. A, 1963,276,542.14 MacEwan and Stevens, J. Nucl. Mat., 1964,11, 77100 pIFIG. 1.-Dislocation network inside a NaCl single crystal0.8 % deformed.FIG. 2.-Dislocation network.20 p below200 pI IFIG. 3.-The as-grown dislocation structure inside a NaCl FIG. 4.Dislocation networksingle crystal. subgrains(c) (4FIG. 1 .-Reflections of a CdS crystal showing different diffraction contrasts at dislocations relatedto the sense of b (Lang’s technique, Mo Kzl radiation, crystal thickness -0.1 mm, p . t-1, sign of carbitrary, scale marks 0.2 mm).-+ -+3(a) (0002) reflection. Dislocations with b = <11.1) appearing diffuse light at P ; at N dislocationswith b = (00.1).(b) (0002) reflection. Dislocations at P appearing diffuse dark whereas those at N appear nearlyunchanged.(c) (0002) reflection. Edge dislocations at El and E2 showing different diffraction effects ; b = (00.1),( d ) (0002) reflection. Diffraction contrast is reversed with respect to fig. Sc.i3[See Mohling, page 317FIG. 1 .-Silicon (100) slice after phosphorus diffusion. (010) reflection. 4 x enlargement.FIG. 2.-Silicon (1 11) slice after deposition of lop of epitaxial silicon. (220) reflection.4 x enlargement.[See Henderson, page 318

ISSN:0366-9033    

DOI:10.1039/DF9643800309

出版商:RSC    

年代:1964

数据来源: RSC

摘要:

GENERAL DISCUSSIONDr. G. Pampus (Physik. Inst. der Univ., Miinster) said: Dr. Lang has shownsome results that could not readily have been obtained by other means. For trans-parent crystals such as the alkali halides we obtain nearly the same pictures as Dr.Lang shows in fig. 7 and 8, by means of ultramicroscopical observations. I wouldlike to give some examples.Fig. 1 and 2* show a dislocation network in NaCl single crystals which is not aplane boundary situated in (100) but has a volume extension. The observation planesin all cases are (100). The plane of fig. 2 lies 20 p below plane 1. We can observealterations of structure in accordance with varying depth of observation. Fig. 3"shows a dislocation arrangement inside an as-grown NaCl crystal. Greater mag-nification would allow more detailed studies.The crystals are undecorated andtherefore the observations are different from those of Amelinckx.Concerning the dislocation network I am not in entire agreement with Dr. Lang.First, in NaCl crystals the network seems to have an extension in depth and secondly,nearly all the crystals examined contained such a network. On the other hand,I confirm the observation that dislocation lines and networks abruptly terminatewithin the subgrains (see fig. 4*). A detailed paper will be published later. Onthe whole there is good agreement between the X-ray studies of Dr. Lang and myown ultramicroscopical observations.Dr. W. MohIing (Physikalisch-Technisches Institut, Berlin) said : Referring toDr. Lang's remark that there are possibilities to determine the sense of the Burgersvector of a dislocation from diffraction contrast (ref.(ll), (15) of Lang's paper)two contrast effects at dislocations are represented which seem to be related to thesense of b. In fig. 1, (0002) and (OOOg) reflections of a CdS crystal taken with Lang'stechnique are to be seen. Whereas dislocations having Burgers vectors perpendicularto the reflecting lattice plane appear predominantly by extinction contrast (at Nin fig. la, b), those, the Burgers vectors of which make an angle of about 60" withthe reflecting lattice plane (b = <ll*l)) appear with diffuse light or dark contrastrespectively (at P in @.la, b*). In other regions of this same crystal, not shown here,dislocations having the same type of b behave inversely, being dark in the (0002)reflection and light in the (0003) reflection.From this result it is concluded thatthis contrast effect is related to the sense of b just as that effect shown in fig. lc, d.*All straight dislocations are purely edge type, having Burgers vectors of the type(00.1). The glide plane of their visible parts is of the type {lolo) which makesan angle of about 30" with the crystal surface. Whereas in fig. lc dislocationsdenoted E:! appear smooth others (El) show an oscillating contrast similar to the zig-zag contrast at dislocations in an electron microscope. (However, with this crystal,there is no agreement between the depth periodicity of contrast and the extinctiondistance.) In the (0002) reflection (fig.Id) this diffraction effect is reversed. Sinceall the Burgers vectors are normal to the reflecting lattice plane the only differencebetween these two kinds of dislocations is the sign of b.Mr. J. C. Henderson (Research Station, Dollis Hill, N. W.2) said: I should liketo present some results of a series of experiments, using X-ray projection topographywhich confirm Dr. Queisser's findings.-3-+-++--t* Between pp. 316-17.32318 GENERAL DISCUSSIONA (100) slice of 2Q cm n-type silicon about 1 mm thick was cut from a 2 cni diam.(111) ingot. After etching to remove the cutting damage the slice was phosphorusdiffused for 2 h at about 1000°C. A control experiment on a 2 Q cmp-type sliceshowed that a junction depth of about 3 ,u was obtained from a surface concentrationof phosphorus of about 1020 atoms/cm3.X-ray projection topographs were takenbefore diffusion, after diffusion and after a 1-min etch to remove the diffused layer.No attempt was made to remove the thin phosphorus-doped oxide layer producedprior to taking the second X-ray topograph.The experiment was repeated a second time on the same slice with essentiallythe same result. Fig. 1" shows the X-ray topograph obtained after the second dif-fusion. The topographs taken before diffusion, and after removal of the diffusedlayer, were essentially the same showing only the grown in dislocations. Thediffusion induced slip is restricted to a region within a few tens of microns from thesurfaces of the slice; the Burgers vectors were found to be those suggested by Dr.Queisser, namely, a/2 [110].An extensive examination of the depth of the diffusioninduced slip was not attempted due to the limited resolution of the X-ray method.What is evident, however, is the patchy nature of the slipped region, probably dueto the difficulty of obtaining uniform doping over the whole of the slice. It wouldthus seem that X-ray topographs can act as a useful precursor to a more detailedelectron microscope examination, in that they can pinpoint areas most likely toprove of interest.In connection with Dr. Booker's paper, I should like to add some results thatwe have obtained, again using the X-ray method, during a series of experimentson epitaxial growth of silicon.A 1-mm thick slice of 1 0 cm ( 1 11) dislocation free silicon (as shown by X-raytopography) was used as a substrate for the deposition of a lop layer of epitaxial1 R cm n-type silicon.The object of the experiment was to see if we could detect stacking faults in theepitaxial layer by the X-ray method.The topograph obtained after deposition ofthe layer is shown in fig. 2.* To prove whether the slip pattern shown was in thelayer or in the substrate, the layer was removed by chemical etching and furthertopographs were taken. It was shown that the slip pattern extended throughoutthe slice and was not confined to the epitaxial layer. Further experiments on similardislocation free slices, subjected to the same thermal treatments (i.e., heating to-1250'C) but without epitaxial deposition, failed to show this pronounced slip.Further experiments are in hand to res.olve the problem.It seems almost certainthat slip is due to thermal stresses set up during the epitaxial deposition, possiblyin conjunction with the constraint imposed by the epitaxial film bridging the gapbetween the silicon substrate and the silicon susceptor upon which the slice restsin the r.f. furnace. These results seem to be at variance with those of Dr. Schwuttke,lwho suggesced that an essentially similar pattern found by him was due to stackingfaults within the epitaxia layer.Dr. G. R. Booker (Cavendish Laboratory, Cambridge) and Dr. R. Stickler (Westing-house Electric Corporation, Pittsburgh) (contributed) : Dr.Queisser has reported thatwhen Si specimens are diffused with P or B, the outer portions of the specimensbecome strained, and the strains are relieved by the formation of dislocation networks.A large number of such specimens have been prepared by Dr. H. F. John of theWestinghouse Electric Corporation covering a wide range of diffusion conditions,including those used by Dr. Queisser. Examination of such specimens by opticalmicroscopy after etching (John) and by transmission electron microscopy after1 Schwuttke, J. Appl. Physics, 1962, 33, 1538. * Between pp. 316-1GENERAL DISCUSSION 319chemical thinning (Booker and Stickler) did not in general reveal any dislocationnetworks. Hence, it can be concluded that the networks which have been observeddo not arise solely as a result of the diffusion process.Other factors appear to playa role, and it is suggested that these may be either surface irregularities or oxide filmsox1 the surface. Similar observations are apparently currently being made by otherinvestigators (Czaja and Wheatley, J. Appl. Physics, 1964, 35, 2782, and Czaja andPatel, J. Appl. Physics, to be published). It is perhaps significant that slip tracessimilar to those observed by Dr. Queisser can be produced without performing adiffusion treatment, e.g., during the deposition of expitaxial Si layers (Henderson,this discussion).Dr. Queisser has also shown that when Si slices with mechanically damaged surfacesare annealed in either wet oxygen or steam, two-dimensional defects grow from thesurface into the slice.However, the nature of the defects was not established. Wehave used an extraction technique to investigate such defects formed in a Si slice byannealing at 1200°C for approximately 2 h in wet oxygen. The defects proved to bethin plates of amorphous material, probably oxide. It is suggested that these defectsform by a process analogous to that described by Silcock and Tunstall 2 (Phil. Mag.,1964, 10, 361) for the precipitation of NbC in steel.CONCLUDING REMARKSProf. F. R. N. Nabarro (Witwatersrand University, S. Africa) said: After sucha feast of new results one is inclined to wonder if there is anything left to do in thefuture. However, discussing the position with Dr. Hirsch, I was able to list a numberof directions in which progress may soon be made.First, in estimating stacking-fault energies from the size of dislocation nodes,we must learn to allow for elastic anisotropy and for the anisotropy of the energyof the dislocation core.As the method is applied to higher stacking-fault energiesand so to smaller nodes, we must pay attention to diffraction theory and make surethat what we see is what is really there.The next problem concerns the thermal resistance of dislocations in insulators.We have got used to living with a discrepancy of a factor of ten between the ob-served and the theoretical values of the electrical resistance, but when Prof. Seegertells us that the discrepancy in the thermal resistance is a factor of ten thousand itis clear that more work needs to be done.Next we come to Prof.Gilman’s theory that the Bordoni damping peak is pro-duced by the motion of orientation junctions in dislocation dipoles. The rivaltheory is due to Prof. Seeger. We know its complications and its difficulties. TheGilman theory has not been developed in any detail, and, for all we know, it niaybe right.As Prof. Haasen has shown us, the theory developed by Johnston and Gilinanof rounded yield points controlled by the velocity-stress relation for a dislocationcan be used in many problems, and we must expect more developments along thesame lines. However, I believe that in the soft face-centred cubic metals such ascopper and aluminium, in which the dislocation velocity increases very rapidly withthe excess of the stress above some critical value, this approach will not be veryhelpful, and we shall have to perfect work-hardening theories of the usual kind.Turning to work-hardening, which we all agree is the most important branchof physics, we shall have to understand Dr.Hirsch’s theory, and find out howsensitive its predictions are to the details of the assumptions. In particular, w320 GENERAL DISCUSSIONmust examine the importance of the pile-ups, both bare and clothed or relaxed.We must see what pile-up phenomena are essential to the theory, and whether ornot these essential phenomena are consistent with the experimental observations.We must discover what proportion of the dipoles in metals, in ionic and in covalentcrystals, are formed by dragging points in regions where screw dislocations aremoving, and what proportion are formed by collisions between edge dislocations.We must find the total contribution to the strain of slip on secondary systems, andcompare it directly with work-hardening studies of the same specimens.I should like to say of work-hardening theory what Mott said twelve years ago,that we are on the verge of a working understanding of the hardening of puremetals, and the time has come to start thinking about alloys.Here the difficultiesare multiplied enormously. Not only are stacking-fault energies altered by alloying,but they are also altered, according to the Suzuki effect, differently in freshly-formedfaults and in faults which have been annealed. If alloying produces a resistanceto dislocation motion, the measured stacking-fault energies may not be the true onesunder the conditions of measurement.Then we have the new mechanisms ofhardening discussed by Fleischer. Tetragonal strain hardening was thought oflong ago, but almost forgotten, and the short-range interactions which Fleischerhas shown to give a concentration dependence as ct instead of the usual roughlylinear proportionality to c were neglected, probably just because they were of shortrange.Finally, we should remind ourselves that this is a meeting of the Faraday Society,and that we are interested in the interactions of dislocation theory and physicalchemistry. Dr. Westwood's paper has emphasized the important physico-chemicaleffects of the condition of the surface on plasticity and fracture, which are no lessinteresting because they are important in technology. Once we hoped to under-stand them in terms of a reduction of the surface free energy in Griffith's theory offracture. Now we have to consider the complexities introduced by complexes.The physicist is out of his depth, and must call in the chemist to help him.It is perhaps appropriate that one of the papers in this very last session, that ofDr. Queisser, should have emphasized another interaction between dislocationtheory and physical chemistry. The presence of dislocations accelerates diffusion,and Dr. Queisser has shown that diffusion leads to the formation of dislocations.This strong interaction between dislocations and diffusion may well provide thelink which will allow us to welcome the President of the Bunsengesellschaft to theranks of the dislocation theorists

ISSN:0366-9033    

DOI:10.1039/DF9643800317

出版商:RSC    

年代:1964

数据来源: RSC

摘要:

AUTHOR lNDEX *Ahlers, M., 157.Alexander, R., 262.Amelinckx, S., 7.Ashbee, K. H. G., 78, 90, 309.Ashby, M. F., 88.Basinski, 2. S., 93, 171.Bethge, H., 79, 281.Bilby, B. A., 61.Bollma~, W., 26, 81, 82.Booker, G. R., 78, 275, 298, 318.Brandon, D. C., 262.Bross, H., 69.Brown, E. M., 35.Bullough, R., 61, 8 I, 91.Cadoff, I., 188.Cotner, J., 225, 272.Davidge, R. W., 92, 181.Diehl, J., 168.Differt, K., 85.Forty, A. J., 58.Gallagher, P. C. J., 157.Gilman, J. J., 123.Goldheim, D. L., 147, 189.Grinberg, D. K. de, 61.Griffiths, L. B., 92.Gruner, P., 69.Guiu, F., 287.Haasen, P., 78, 91, 178, 183, 191, 275, 283.Harrison, R. P., 21 1.Hazzledine, P. M., 103, 184.Henderson, J. C., 317.Hesse, J., 283.Hirsch, P. B., 49, 111, 157, 166, 171, 175, 179.Hornstra, J., 81, 91.Hull, D., 251, 288.Ilschner, B., 187, 243, 278, 287.Kastner, G., 281.Kronmuller, H., 173.Kroupa, F., 49, 289.Labusch, R., 273.Lang, A.R., 292.Levine, E., 188.Li, J. C. M., 138,233.Miles, G. D., 285.Mohling, W., 317.Mordike, B. L., 280.Nabarro, F. R. N., 173, 177, 187, 319.Newly, C. W. A., 211.Noble, F., 251, 288.Pampus, G., 282, 317.Patel, J. R., 201, 275, 277.Peissker, E., 178.Pesse, J., 183.Pratt, P. L., 92, 181, 211,277, 287.Pugh, E. N., 147, 189.Queisser, H. J., 305.Reppich, B., 243.Riecke, E., 243.Royce, B. S. H., 218.Seeger, A., 69, 85, 157, 158, 163.Schmidt, V., 79.Schoeck, G., 274.Scholz, R., 79.Siems, R., 42.Srnoluchowski, R., 218.Spreadborough, J., 262.Steeds, J. W., 103, 161.Stickier, R., 318.Stokes, R. J., 182, 233.Swann, P. R., 84.Tholen, A. R., 35.Weertman, J., 225.Welsh, H. K., 278.Westwood, A. R. C., 347, 189.Wilkens, M., 85, 159, 172.Young, D. A., 90.* The references in heavy type indicate papers submitted for discussion.32

ISSN:0366-9033    

DOI:10.1039/DF9643800321

出版商:RSC    

年代:1964

数据来源: RSC