1

Defects in AgCl and AgBr(100) Tabular Crystals studiedby TEM

W. Van Renterghema,b, C. Goessensa,*, D. Schryversa, J. Van Landuyta, P. Verreptb, D.Bollenb,

C. Van Roostb and R. De Keyzerb

aUniversity of Antwerp, RUCA, ANTWERPEN, BelgiumbAgfa-Gevaert N.V., MORTSEL, Belgium*Now at Union Minière, OLEN, Belgium

Abstract

AgCl(I) crystals with (100) surfaces are studied withconventional transmission electron microscopy (TEM) fortheir microstructure in relation with their growth habits.The crystals are prepared according to two different growthmethods which give slightly different results. It isobserved that the presence of dislocations of mixedcharacter that end at only two of the six cube faces are inboth cases responsible for the tabular growth, but theformation of these dislocations is different. In the firstcase a stacking fault is formed which is eliminated bydislocations. In the second case the incorporation ofiodide impurities is at the origin of the formation ofdislocations. For comparison the defects in AgBr (100)tabular crystals are studied. Again dislocations of mostlymixed character are responsible of the anisotropic growth,but they are much more mobile compared to dislocationsin AgCl.

1. Introduction

If AgBr crystals are grown at high Br- concentration with

respect to the concentration of silver, tabular crystals with(111) faces are formed. The anisotropic growth habit isdue to the presence of usually two or three twin planes[1,2] parallel to a (111) plane, which are easily formedunder this condition.

Applying the same procedure to make AgCl tabularcrystals, gives rise to a few problems. Due to the highionic charges, the (111) planes of AgCl are not stable inan aqueous environment, even at high concentrations ofCl-. Consequently no (111) planes can be formed. Asolution to this problem is the addition of stabilizers.This, however, reduces the sensitivity of the emulsion andinterferes with chemical and spectral sensitization [3].

To avoid the use of stabilizers a method was developed togrow tabular crystals with (100) surfaces [4]. Large (»30 mm) flat AgBr crystals with (100) faces have beenprepared for the first time by Mignot et al. in 1974 [5].

Brust and House [6] found a method to produce AgCl(100) tabular microcrystals. Both authors argue that notwin planes are formed and Mignot et al. suggest thatscrew dislocations are responsible for tabular growth, butonly little experimental evidence is provided. RecentlyOyamada et al. [7] have presented a growth mechanism for(100) tabular crystals. These authors state that the tabularcrystals contain two dislocations which intersect in onecorner of the crystal and that their direction is [310] or[130]. Although the Burgers vector is not explicitlydetermined, they conclude that the dislocations are screwdislocations, which induce the tabular growth. This,however, would mean that the Burgers vector is of theá310ñ-type, which will be contradicted by ourmeasurements and is also unlikely according to theliterature [8].

In this presentation the nature of the defects present inAgCl and AgBr (100) tabular crystals is discussed basedon a conventional transmission electron microscopy(CTEM) study.

2. Experimental procedures

The AgCl crystals are prepared at Agfa-Gevaert N.V. andtwo different growth methods are used to produce them.In both cases a double jet method of an AgNO3 solutionand a KCl solution is used. The first method starts with anucleation phase, followed by a physical ripening andfinally a growth phase. The crystals grown in this wayare in this paper referred to as type 1. The second method,referred to as type 2, also starts with a nucleation phase,after which a small amount of iodide is added. Afterwardsthere are three physical ripening phases each of themfollowed by a growth phase [9].

The AgBr crystals are prepared at Agfa-Leverkusen. Adouble jet method of an AgNO3 solution and a KBrsolution is used. The nucleation phase, where smallnuclei of AgCl are formed, is followed by a first growthphase with AgBr. Afterwards there are two physicalripenings at different pAg values and finally there is asecond growth phase [10].

2

For the preparation of the specimens for TEM microscopy50 ml of each emulsion is thinned in 50 ml of doublydistilled water. A drop of this solution is placed on acopper grid covered with a carbon foil and the water isevaporated. To avoid photolytic reaction the preparationof the specimen and the mounting in the microscope aredone in a dark room under red light conditions.

The samples are studied with Philips CM20 and CM200microscopes using a Gatan double tilt liquid nitrogencooling holder. The results are obtained mainly bydiffraction contrast observations. The temperature of theenvironment around the specimen during the study is keptat approximately 100K to minimize radiation damage.

3. Results for AgCl.3.1 Characterization of the defects.

The two types of growth methods yield tabular crystalswith a different defect structure and it appears that adifferent mechanism must be responsible for the defects.

When the crystals are fully grown to the desired size (»1mm) the only defects that occur in both cases aredislocations. The number of dislocations in tabularcrystals is limited and usually varies from 1 to 4. Theymostly lie in a (001) plane parallel to the top surface ofthe crystal, although a small component along the [001]axis can not be excluded. During the observations thedislocations appear to be immobile and do not move underelectron irradiation. However, it can not be excluded thatthe dislocations move during the growth process.Exceptionally a tabular crystal occurs with no defects atall. It is, however, impossible to decide whether therewere defects present during growth which have disappearedor that the crystal was defect-free all through the process.

a) type 1 crystalsThe Burgers vector of the dislocations is determined fromdiffraction contrast extinction conditions and two differenttypes are found. If the top surface is referred to as a (001)-surface then most of the dislocations are extinct for a 111and a 220 type of reflection, as shown in figure 1. Thecombination of the two extinction conditions always leadsto a Burgers vector of the á110ñ type, parallel with thecrystal surface. This result is not surprising since incrystals with a face centred cubic structure and in particularwith a rock salt structure dislocations with an a/2[110]Burgers vector have the lowest energy [8] and are thusmost likely to occur. The direction of the dislocation ismostly along [100] or [010] and it switches often betweenthese two, as observed in figure 1. Also other directionsare possible, but the dislocation line practically never runsalong a [110] or [110 ] direction, which means that thedislocations are never pure screw nor pure edgedislocations but have a mixed character.

Figure 1 : Dark field images of a) a tabular crystal oftype 1 with one dislocation with an a/2[110 ] Burgersvector; b) extinction for a [111] diffraction vector; c )extinction for a [220] vector.

A smaller number of dislocations are not extinct for theabove conditions, but for a 200 reflection and a 202reflection, leaving only the a[010]-vector as a possibleBurgers vector. The direction of the dislocation is alwaysparallel to the Burgers vector, indicating that thesedislocations are screw dislocations

b) type 2 crystals

Figure 2 : Typical crystal of type 2, with two a/2[110]dislocations, which are marked by the white arrows.

In this type of crystals only one type of dislocationsoccurs, which all lie in a (001) plane. From extinctionconditions it follows that the Burgers vector is always ofthe a/2[110] type (the dislocations marked by arrows infigure 2). The main dislocation directions are [410] and[100], but practically never [110]. So all dislocationsagain have a mixed character. The shape of these crystalsas seen in top view is rectangular. In one of the cornersone often observes a small dislocation loop, from which

3

two dis ocations o iginate : one goes to the opposite (100)side plane along a [140] direction and the other goes to theopposite (010) side plane along a [410] direction.

3.2 Origin of the dislocations.

In order to investigate when and how these dislocations areformed, samples are studied that have been taken from theemulsion at different stages during the growth process.This also gives different results for the two types ofcrystals.

a) type 1 crystalsAfter the nucleation phase conventional TEM can notdistinguish between the different nuclei. They are verysmall (less than 50 nm) and no defects are observed inthese crystals. The shape of the crystals is cubic.

After the physical ripening most of the nuclei did notgrow and still do not show any defect contrast. Thecrystals that did grow are also cubic and they all containone stacking fault, but contrary to the crystals at the endof the growth process, no dislocations are found (figure3a). The displacement vector for stacking faults in facecentred cubic materials is a/6[112 ] and this is confirmedin the present crystals by diffraction contrast analysis.

In samples, taken from the emulsion after 10 to 20minutes in the growth phase, the number of crystals witha stacking fault decreases. Moreover the stacking fault isnot the only defect in those crystals anymore, but one ortwo dislocations are observed always related with thestacking fault (figure 3b). The directions of thesedislocations are [100] or [010], but they can changedirection when the crystal further grows. Crystals with astacking fault and one dislocation have the shape of aneedle (figure 3c), with the long side parallel to thedislocation.

The next step is the elimination of the stacking fault. Itis known that a stacking fault can only end at the surfaceor at a partial dislocation. Moreover, the movement ofsuch a partial dislocation causes the faulted area to enlargeor to reduce or even to disappear. So the formation of apartial dislocation and the subsequent movement to a sideplane, can eliminate the stacking fault. Figure 3d showsthat such a process is very likely to be the cause for thedisappearance of the stacking faults in these samples. Thestacking fault (SF) no longer runs from one side of thecrystal to the opposite one, as in figure 3a-c, but seems tobe partially eliminated at two places by the formation andsubsequent movement of the indicated partial dislocations(PD). At the end of the process the partial dislocation alsoreaches the surface and disappears together with thestacking fault. This process will probably be very fastand a intermediate situation like in figure 3d is only rarelyobserved. However, we assume that most crystalsundergo this process because when they are fully grownall tabular crystals contain exclusively dislocations asresidual defects.

Figure 3 : Different stages of the defects in crystals oftype 1. a) taken from the emulsion after the physicalripening and containing only a stacking fault (SF), b)after 15 minutes during the growth phase and containing astacking fault (SF) as well as two dislocations (D1 andD2), c) a needle which has only one dislocation (D) andd) shows the removal of the stacking fault by partialdislocations (PD) leaving a weak residual contrast (RC)in the figure.

b) type 2 crystalsA totally different kind of process seems to be responsiblefor the formation of the dislocations in type 2 crystals.Crystals that are taken from the emulsion after thephysical ripening or during the first growth phase have nostacking fault, but all crystals are very small and onlyduring the first growth phase dislocations are observed.Once the dislocations are formed the tabular growth of thecrystals proceeds. The direction of the dislocations doesnot change very much, except for a few short segments.The origin of these dislocations is related to the additionof iodide after the nucleation phase, which induces latticestrains because of their larger radius. To verify this, anemulsion is grown under the same conditions as for thesecond growth method but without the addition of iodideafter the nucleation phase. The result is a serious decreaseof the number of tabular crystals and the defect structure issimilar to that of the crystals of type 1.

In order to correlate the presence of the dislocations morefirmly with the tabular growth, the position of thenucleus in the fully grown crystal is determined. Crystalsare grown according to growth method 2 but at three

4

diffe ent moments a mixtu e of AgC B (10%) is addedinstead of pure AgCl, while all the other parameters arekept the same as before. The disadvantage is that thebromide influences the defect structure. The number oftabular crystals decreases to approximately 2 %, the othercrystals are subjected to thickness growth. Fortunatelythe defect structure in the remaining tabular crystalsappears to be the same as before and no extra dislocationsare created. The only difference is the occurrence of aninterface contrast at the transition regions from AgCl toAgClBr, just like in (111) tabular crystals with an AgBrcore and an AgBrI shell [11]. The resulting strain contrastenables us to determine the position of the nucleus withrespect to the fully grown crystal.

Figure 4 : AgCl(100) type 2 tabular crystal with threerings of AgClBr(10%). a) Two of the rings are visiblebecause of the interface contrast and they lieconcentrically around the point where the dislocationscome together. b) Because more dislocations arepresent, the complete ring is visible.

Examples are shown in figure 4. In figure 4a, only twoedges of a strain contrast ring are sufficiently spaced to bevisible and they extend parallel with the faster growingedges. This means that the position of the nucleus forthis crystal is in one of the corners where the dislocationsdo not cut the side edges. In cases where moredislocations are present, like in figure 4b, the rings ofinterface contrast lie more concentrically around the pointwhere all dislocations intersect, which means that the twodislocations cut the crystal edges and promote concentriclateral growth as visualised by the strain pattern.

4. Results for AgBr .

At the end of the growth process dislocations are the defectsthat occur in most crystals. From extinction conditions itwas determined that the Burgers vector is again of thea/2á110ñ type. The dislocation lines do not run as straightas in the AgCl-crystals, but, as shown in figure 5, they arestepped. However, they still lie in a plane parallel to the(001) top surface and end only at the side planes.

Another difference with AgCl is that the dislocations aremuch more mobile in AgBr. Dislocations in an AgClmicrocrystal were not observed to move under electronirradiation, but in AgBr crystals they were. This could alsoexplain why the dislocations are stepped.

Figure 5 : Example of an AgBr (100) tabular crystal withstepped dislocations.

In approximately a third of the tabular crystals, no defectswere observed, but for a small dislocation loop in one of thecorners of the crystal. This number is much higher than inAgCl emulsions. The most logical explanation for this isthat the crystal during the growth process did containdislocations that induced the anisotropical growth, but that,due to the higher mobility, the dislocations have moved outof the crystal at a later stage during the growth process.

The origin of the dislocations in AgBr crystals was notstudied since no samples were taken from the emulsionduring the growth process. It is therefore not possible tocorrelate the AgBr crystals with AgCl crystals of type 1 ortype 2.

5. Discussion of the results related to tabulargrowth.

In the crystals studied, many indications are found thataccelerated growth occurs at surface steps caused bydefects. In crystals of type 1 after the physical ripening,the largest crystals are those that contain a stacking faultand their shape is cubic, which confirms that there is astep of equal height on all three faces. In the needle-likecrystals, containing a stacking fault as well as one

5

dis ocation afte the physica ipening, the dis ocationemergence point has promoted accelerated growth and as aconsequence the dislocation is parallel to the longest side.At the end of the growth process these needle shapedcrystals no longer contain a stacking fault, but have onedislocation parallel to the longest side plane, since growthhas occurred along this direction. In the tabular crystals,mostly two dislocations are present which surface at twomutual perpendicular side planes. These dislocationscome together in one corner which indicates that growthonly occurs along side planes where a dislocation emergesa surface.

The firmest proof is found in the crystals with the AgClBr“growth” rings. In crystals with two dislocations, theposition of the nucleus as well as the fact that only twobranches of the ring are visible show that only two edgeplanes, those where the dislocations emerge the surface,have grown, while the other two have hardly grown at all.

Based on our observations we propose the followinggrowth models for the different kinds of crystals.

a) type 1 crystalsDuring the nucleation phase cubic nuclei are formed.Some of these contain a stacking fault which causesaccelerated growth in the three dimensions; these are thelargest crystals, already after the nucleation phase. As thecrystals containing stacking faults grow, one or twomixed dislocations with a Burgers vector of a/2á110ñ orexceptionally aá100ñ and a direction of [100] are formed.The planes where these dislocations emerge the surfacegrow faster and in this way one- or two-dimensionalgrowth occurs. Finally the stacking fault is eliminated bythe creation and subsequent movement of partialdislocations with a/6[112 ] Burgers vector. After thestacking fault has disappeared and because all dislocationslie in the (001) plane, the growth continues in only twodimensions and tabular crystals are formed.

b) type 2 crystalsThe growth process for this type of crystals differsslightly from that of the first type. Defects arise after theaddition of iodide after the nucleation phase. Theincorporation of iodide causes a lattice mismatch andelastic strain due to its larger atom radius. This strain isusually relaxed by (mostly two) mixed dislocations withBurgers vector a/2[110] and direction [140] of [410]. Atthis point it is still unclear why these directions arepreferred. The edge planes where these dislocations crossthe surface grow faster than the other planes. Because alldislocations lie in a (001) plane growth occurs in twodimensions and tabular crystals are formed.

c) AgBr crystalsSince the origin of the dislocations has not beeninvestigated, it is not possible to propose a growth modelfor these crystals, although the observations show strongindications that again dislocations of mixed character areresponsible for the anisotropic growth.

6. Conclusion.

Two different growth methods are developed to growAgCl tabular crystals with (100) surfaces. For crystalsresulting from growth by these two methods a limitednumber of mixed dislocations with mostly an a/2[110]Burgers vector are responsible for the tabular growth. Theorigin and formation of the defects is different for the twotypes of crystals. In crystals of type 1 first a stackingfault with a displacement vector of a/3[111] is formed,while later during the growth process dislocations areformed always in correlation with the stacking fault. In afinal step the stacking fault is erased by the formation andsubsequent movement to the surface of a/6[112] partialdislocations. For type 2 the dislocations are formed dueto the incorporation of a small amount of iodide after thenucleation phase. They already originate in the earlystages of the growth process and are not subjected tomajor changes anymore. For AgBr (100) tabular crystalsalso mixed dislocations with an a/2[110] Burgers vectorare responsible for the anisotropic growth, but the originof the dislocations was not studied.

AcknowledgementsThe authors wish to thank Dr. H.-H.. Teitscheid for thepreparation the AgBr samples. Work supported by theFlemish Institute for the Encouragement of Scientific andTechnological Research in the Industry (IWT).

References1. R.W. Berriman and R.H. Herz, Nature 180, p.293-294,

(1957).2. J.F. Hamilton and L.E. Brady, J. Appl. Phys. 35

p.414-421 (1964).3. P. Verrept, internal communication Agfa-Gevaert.4. G. House et al., US Patent 5,320,938 (1994).5. A. Mignot, E. François and M. Catinat, J. Cryst.

Growth 23, p.207-213 (1974).6. T.B. Brust and G.L. House, Proc. of the IS&T 49th

Annual Conference p.35-37 (1996); T.B. Brust andG.L. House, J. Im. Sci. Tech. 41, p.413-415(1997).

7. T. Oyamada, T. Shiozawa and S. Yamashita, 1997 Int.Symp. on Silver Halide Imaging, p.2-7 (1997).

8. D. Hull, Introduction to dislocations (Pergamon,Oxford, 1965).

9. P. Verrept, 1997 Int. Symp. on Silver Halide Imagingp. 8-10 (1997).

10. H.-H. Teitscheid, internal communication Agfa.11. C. Goessens, D. Schryvers, J. Van Landuyt, S.

Amelinckx, A. Verbeeck and R. De Keyzer, J. Cryst.Growth 110, p.930-941 (1991).