AMERICAN INSTITUTE OF JIINING AND METALLURGICAL ESGISEERS

Technical Publication No. 868 C L I ~ S C. IRON AND STEEL DNIRION, SO. 146; CLAUU E, INRTITTTE O F >IET.ALR DWIRION, SO. 205)

DISCUSSIOS OF THIS PAPER IS ISVITED. It should referably be prenented by the ron- rributor in person at the S e w York Meeting. Februar 1936 wgen a n abstract of t h e paper will be read. If t h i s is impossible. diqrunuion in writing may ge sent'to the Serretary, Ameriran In~ t i t u t e of AIinina and AIetallurgirnl E:n$~neeru. 29 West 39th Street. S e w York. S. Y. Unless mperial arranpe- ment is made, di~rusaion of t h ~ a paper will close April 1. 1936. Anydiscuaaion offered thereafter should preferably be in the form of a new paper.

Surfacc Magnctixation a r~d Block Structure of E'crrite*

( S e w Yurk SIeeting. February. 1936)

THI.: ~ l l i t g ~ i ~ t i ~ 1)oivdrr n~rtliod, long used for rouglily n~appi~ig ]nag- netic ficlds, lins rcccntly l)ccn refinedl.Vor invrstigating tlic n~icroscopic variations in tlic surfarc ~nag~irtization of ferromagnrtic crystals. Bitter2 first discovcrcd rrgular powder patterns coiliicctcd witli tlic crystal structure of snrli sprrimrns. Srvcral otlier investigators3 Iiavc used Rittrr's origin:il trrliniqnr, with slight nlodifications, t o study tlic varia- t ion of tlic pnt t cbr~is wit11 variations in the magnetizing firld applied parallrl to tlir sl~rf:ic.cl, and tlicir drprndc~ico upon tlir stntc of strain, of polisli nlici of crystal rut. All of tllr rxplanatio~is of tlirsc pattenis on ~iirtallir frrro~ii,zg~ictirs li:tvc bcrn couched ixi gcncral terms. In a prr l in~i~i:~ry rrport wr liavr dcscrihed an improvement in the powder tecliniclnr ivliicli lrd us to tlie discovery of a nriv rffrct throivi~ig ron- sidrrable liglit on tlie magnetic secondary strlictl~re of alpha iron4 and of nirkclh si~iglr rrystals. \Ye now report additional cvidcnce confirming the cxistr~irc a~ i ( l st:tl)ility of a magnctir block srco~idary structl~rr.

Tlie follo\virig list dcscribes the specimens used in t,he present, invrstigatio~i. DESI~NATION DESCRIPTION

A, Silicon ferrite furnished to us hy Dr. W. E. Ruder, of the Gen~ral Electric Co., for which he gave the following typical analysis: C, 0.05; P, 0.038; 8,0.026; JIn, 0.15; Si, 3.24. Disk, 0.60 cm. diameter, about 0.05 cm. thick.

* Condcnsrti from n tiisscrtation suhn~ittcd hy W. C. Elmore in candidacy for t h ~ degree of Doctor of Philosophy. JIanuscript received a t thc office of the Institute Aug. .5, 1935.

t Yalr I-r~iversity. 3 Professor of Physics and Director of the Sloane Physics I,ahoratory, Yale

rnivcrsity, S r w IIavm, Conn. 1 Rrf r r~ncrs nrr s t the rnti of the paper.

1

Copyright, 1035. by the Ameriran Insti tute of Jfinina and hfetallurgical Engineers. Inc. Printed in U. 8. A.

JIETALR TECBNOLOQY December, 1936

2 BUHFACE MAGNETIZATION AND BLOCK STRUCTURE: O F FERRITE

DEIIIUNATION DESCRIPTION

Hy X-ray analysis angles between disk normal and [loo] 75", [010] 76", IOOl] 21".

A: S:ilnp a3 Al exrept: diameter 0.75 cni. A 3 S:inir 3s Al exrept : tlianieter 0.68 cni. 1% Sanle as A , exrept: polygonal piece a few n~illinlet~ra on each edge with

nl~glrzr 1)etwrrn nornial and [I001 76", (0101 73", [001] 22". C Ssnir ns 1% exrept: nngles between norn~al and [I001 87", [010] 56", [001] 35". D S:IIIIC US A , ex(.ept: (liall~eter 1.33 cnl, thickness 0.022 cnl. This specimen

is So. 1 of R. F. Clash ant1 F, J. Beck6, who usrd it in studinl on the direc- t ional dist ri1)ution of Harkhausen changes in magnetization.

E Ferrite furnished to us by Dr. D. D. Foster, of the Bell Telephone Lahora- torirs, for which he ga\-e the following analysis (hefore wire-drawing ojjrrutions): C, 0.02; P, 0.012; S, 0.005; Jig, 0.01; Al, 0.36; Si, 0.008; CII, 0.04; PI,, 0.02. Wire 1 mnl. in tlianicter. Angles t)etween axis of wirr and [loo] 82", [010] 68", [001] 24".

F Silicaon ferrite furnished to us by N. P. G o s ~ , Cold bletal Process Co. Severnl pieres csut fro111 a strip about 0.033 rnl. thick, including a disk usrd l ~ y Clnvh and Beck6.

Tlic polishing proccd~ire finally adopted was olic that niininiized surfacc strains. Kncli of the rlsual polishing stcps was continued about twicc as long as it took for the scratches from thc previous stcp to dis- appear as judgrd 1)y tlir unaidcd eye. The final polish was done on a liorizo~it a1 rotating disk covered with wool broadcloth impregnated with a pastr of c.1). nlag~icsiun~ oxide (3Ierck) in water.

Thc colloid was prrpared by grinding ganlnla (frrromagnrtic) F r203 to sut)~iiicroscopic sizr in a porcelain ball mill containing watrr and a little gum arahic. It was furnisllrd to us by Dr. L. A. Wrlo, of Tottrnville, Statrn Islalid. Tlir individual particles, in violr~it Brownian movrment, could just I)(* scrn at high nlagnefication with dark firld illumination. A fr\v drops of colloid wrrr placed on the polishrd surfacr of tlir mag- nrtizrd sprrinlr~i. Aftrr a few n~inut,es, a covcr glass was plscrd ovrr the sprrimrri, squrezing tlir liquid to a thin layrr. Thr covrr glass was hrld from to~ic.lling tllr sprci~nrn by mounting thr sprcimrn in s sliallo~v well, or by putting a fo\v \)its of wax near the rdgrs of tlir covrr glass.

Tlir sprrinlrris wrre always magnetizrd by mounting on tlir polr of a srllall vrrtieal rlrctromagnrt, prcviously drscribcd! In onr rxperinlcnt the sprcimrn, firmly attached to the pole of this magnrt, was mounted in tllr liorizorital firld of a larger (du Bois) rlectromagnrt. This arrange- mrnt prrnmittrd indrpcwdrnt control of the nornlal and transvrrsc mag- netizing fictlds. Tlir patterns were observed witli a nlicroscopr rnlploying vertical illunli~latio~i; an 8-mm. objective proved to be most useful.

W. C. ELMORE AND L. W. hIrKEEHAN 3

To prevmt thr colloid from att:icliing itscllf to t l i c b s~ )e t . i~ i~c~r~ , it \\-:is fo~ind xierrssary to lrave all i11visii)lr 61111 of grcb:ise or1 t lit. polislic~ri surfac.cb. If thr surface was washed wit11 C('I4, tlir colloid t t~~~cit~ci to stivk fast, \vl~ilo if a prott.c.ti\.e coat of grcwe \\-as 111c~cly wiprd off \vitli caotto~i, this st icaki~ig did not occur. The* trndr11c.y of t l i t ) c.olloit1 to for111 rust o11 soriit* iron crystals was rcbducrci by adclir~g a s~iiall a n ~ o u ~ ~ t of IiOH to it. O \v i~~g to t h r delrtilrious roagulating c.ffect of IiOH, tliis ~)rocrclurr \\-as lntctr discardc.d in favor of ropprr-plating thcb carrfully clt~ar~t~cl surfac.tl. A cayanidt. 1)ath \\-as usrd. Tlir t 11i1i laytar of copptlr (lid ~ io t i~~trrftlrtl wit 11 t 111. for~riat io11 of pat t rrrls7, hut \v:ts usc*ci o111y \ \ . l ~ t l ~ i ~ie(*~~ss:try.

KE;SITLTS

We prtlviously rthportcld4 :t c*l~amc.tc~ristic sliifting of the colloid 1i11es ~ I I

silicon frrritc crystals u p o ~ ~ rr\.ersal of tl111 11orn1n1 ~i lag~~t*t izir~g firld. \Yitl~ no firld a pnttrrn npprarrd comprisir~g I)otl~ ststs of li~it.s, spac.t~i ahout 2p. Fig. 1 sl~o\vs higl~ly ~nagr~ified tlio t l~rec pattclrns o11 n ~)olishrd fact. not 1nuc11 i~~(ali~~tad to :i [100] p1a11e. To (1xpl:ii11 tl~rst> patt(tr11s \vtb

irl~agined thtl surf:ic.t. to 1 ) t b con~posctd of sri~all 1)loc-ks, 2p or1 all tlcigtb, 111agnetizt.d parall(*l to tlica surfac.c~ along axrs of form < 100 > a ~ ~ d < 110 >. 'I'llr n~ag~~ t> t i za t i o ~ ~ of tlirstl hloc'ks oppostbs at half t lie ho~uid- nritu, so that t he 11t.t magnt~tizatiori is as sr11al1 as possil)lt~ o\.tlr ally region. Wr discussed the rnariner in which this nlodrl acrour~ted for t11e three ot)sc~rved patterns arid t h r shift from one pattcbrn to ariot lirr. Tlw blocks (possibly cui)es), or pcbrhaps groups of then], rllay I)(> tt~~~tativculy identifitd wit 11 the domains of the Ht~isc~iibcrg-f riss tlic.ory of fcrron1ag1ic1tisril~.

We now report, and discuss furtlier rosults obtuined wit11 silicori fvrrite crystals. Fig. 2 slio\vs three patterns obtail~cd on A2. Ir~strad of the usual non-preferential polisli with JIgO paste, this sprcinlc~n \\-as given a preferential polish wit11 alundum, the dirrcation of polisliir~g strokes being evident from the finr scratches, which show Iwst ill tlie no-field pattern. Tllc majority of the linrs wit11 applied firld are per- ~~endicular to the polishing directio~l and are spaccd about 12p. Tlle other set of lines, many of which apprar split, occur only wl~ere parallel statv of the first lines fail to match up. A fine structure is apparent 1)etween thr pronli~icnt lincv, as is shown niorc clearly in Fig. 3, taken nt higher magnification.

Additior~sl nonpreferentiul polish with JIgO yiclded t l ~ r usual fine scale pattrrn. Fig. 4 sho\vs a pattern takv11 beforca return to thr fi11t.r pattern was complete. Altl~ough orir set of lilies is still somewhat more conspicuouv than the set at right angles, tlie average spacing has beconlr 4p again. The occasional enhancing of linrs of thc fine pattern dis- appeared after further polish.

4 SURFACE MAGNETIZATION AND BLOCK STRUCTURE OF FERRITE

FIG. 1.-SPECIMEX A3. GRILLE IS 14p X 82p. a, normal applied field positive; b, zero; c, negativc.

W. G. ELMORE AND L. W. McKEEHAN' 5

Tlze block model seems adequate to account for the patterns with larger spacing. We suppose that the direction of magnetization of the 2.u blocks are arranged in sequences of threes opposing, to the depth of several layers, with occasional surface blocks paired off with opposing magnetizations, as indicated in Fig. 5. This scheme gives with normal field dense lines spaced a t six block n~idths, and occasional fainter Iines spaced at two block widths, all of which move to unoccupied spaces upon

reversing the field. Stray field a t the corners of the blocks accounts for the fine structure of dots existing between the lines. With no field the lines of both patterns appear, since there is now stray field a t all of the opposing boundaries.

The fine spaced pattern, Fig. 4, still exhibits characteristics of the former pattern. Blocks with magnetizations opposing in sequences of twos are occasionally evident. The enhancement of some of the lines indicates that the arrangements of block magnetizations beneath the surface layer must be contributing excess stray field along particular block boundaries.

That tension will .produce magnetic nonuniformities along the direc- tion of easy magnetization most nearly perpendicular to i t has been

W. C. ELMORE AND L. W. McKEEHAN 9

sequences. For the other surfaces, each value in Table 1 represents an average of about 400 line spacings. Individual line spacings varied from the average by as much as 10 per cent.

Cut

A similar experinlent with the average block size, with the following average dimensions :

One of the three vertical cuts approximated a ( 111) plane, the pattern on which consisted of dots. The tendency of the dots to form lines was not pronounced.

That these results are not peculiar to silicon ferrite crystals was verified by obtaining patterns on two perpendicular surfaces of

a Estimated. a = angle between plane of cut and axis indicated at t,op of column D = average separation of the closely spaced lincs on each cut.

D cos a d = average length of block edge A - -

2

Rltdcr crystal C yielded a slnaller

- -

the more nearly pure ferrite crys- FIG. 7.-STEREOQRAPHIC PROJECTION OF

tal E. The crystal was anilealed CUTS OF SPECIbfEN B.

in Hz and then soldered into a hole in a soft iron block, so that one side and one end could be polished. Patterns on these two surfaces gave the following average block dimensions:

These results point to the existence of a magnetic block secondary structure in ferrite crystals. Although the same general polishing

10 SURFACE MAGNETIZATION AND BLOCK STRUCTURE O F FERRITE

procedure was consistently adhered to, the polishing history of the individual surfaces must have varied considerably. The agreement of the block dimensions found on several cuts of the same crystal strongly indicates that polishing does not influence the observed block size, and hence that the pattern appearing on a given surface is the result of a structure already existing in the iron crystal. That this structure is common to most iron crystals is suggested by the mazelilte patterns found on iron by various other investigators.

a b FIG. 8.-SPECIMEN F. DIRECTION O F ROLLING VERTICAL. REGION SHOWN IS ABOUT

4 6 0 ~ X 3 5 0 ~ . a, normal applied field positive; b, negative.

The size of the block varies from specimen to specimen. Pattern lines have been observed as close as lp, which would require a block 0.5p on an edge, while the maze patterns previously shown by others would require much larger blocks. Of course i t is not clear whether these patterns with larger spacings are, or are not, the result of an arrangement of blocks in sequence opposing. Patterns on polycrystalline specimens (Fig. 8) show a considerable variety of spacings in adjacent crystallites, which presumably differ very little except in orientation. Any con- siderable degree of preferred orientation is easily det,ected. There are about 1012 a ton~s in the blocks here proposed to explain the colloid pat- terns. This figure is in good agreement with estimates of domain size made from Barkhausen noise, which range from 101° to 1015 atoms8.

R'. C. E:LMORE A S D L. W. M c K E E H A S

E:FF~::(-T OF TRASS~ERSE FIELD os FISE PATTERS

If a small firld was applied parallel to a (100) surfacc in addition to tlie usnal normal ficld, simultanrons 01)scrvation with the microscope revcalrd surprising chaligrs taking place in the mazrlike pattern. The li~ics of rolloid ivollld movcb locally by jumps equal to the width of onr 1)loc.k. 1,irics mrrtirig a t corners wrre particularly sensitive to ctiangc~s in transvrrsr firld. If the tralisvcrsr ficld was turned on and off, th r normal firld rrriiaining constant, rach thro\v of the switch would produrr a new mazr having thr same grncral appcararicc as previous nlazrs, yet diffrring in drtails. All of the many pattcnis to br sccn, holvc.vc~r, wrrr c.onsistrnt wit11 thr view that hlocks retain their identity during the changes. Ttic. changrs, apparmtly, arc brought about hy rotation of tlir direction of mag~irtization in the individual blocks to one of the 0tht.r stkvell possiblc dirrctiolis (of forms < 100 > and < 110 >) lying ill th r slirfacr. Upon iricrcasi~ig tlir transvrrsr firld to values produc:ing approximatt* magnetic saturation, tlir fine pattern disappeared, lcavirig only :i fr\v 1)rokt.n strclaks of colloid ~)crpmdicular to the tralisvorsc. fitbld.

A ~lovcl sc.ht.riir was hit upon to study changes in thc fine pattern undrr tlle coliclitions drscribrd shove. A colloid pattern was drpositcd with ~itrrnlal firld, tlir chxcrss rolloid was rinsrd off with watrr and the surfacth quickly dried with comprrssed air. Sncli a procedure left the colloid pattrrri illtact, and firmly attachcd to the surface. Then anothrr colloid pattcrll was fornird arid the sptbc.initri su1)jtlcted to a transvc~rse field. Thc tivo patterns, one fix(~1 and onr clianging with tlie transverse ficld, colild t ) t l o1)srrvrd simultn~icously. I t was p1:tinly cvidrnt that the 1)locks that could br loratrd fronl thr old pattchni ~v r r c rrsponsi1)lr for the 11t.w pattcbrns. That sucli is the casr, rvrn though the specin~rn in tlir mrantimc suffcrs complctc magnetic saturation in arbitrary dirce- tions, will br ohvious fro111 a study of Fig. 9.

Thrsr pattrrns, takt.11 on ABl show thr comhincd nciv rtlicl old patterns wit11 n o ficld and wit11 nornmal ficld in the original and revcrscd dircctions. Before the deposition of tlic nciv pattern the sprcimcri was rrmovrd from ttir vrrtical niagllrt slid arbitrarily movrd around ill a ficld sufficirnt to saturate it parallrl to its surfacc. The fixed pattcrn may be identified in thc 110-field pattern.

Anot hcr rxpcrimcnt \\.it h tralisvcrse field dcnmonstratcd the relative importance of magnetic history as compared with the polishing history. The disk D was polislird with the two dircctions of coarser polish coinci- dent wit11 tlir < l o o > dircctions most licarly in the facr of the disk, folloivrd by thtl tislial nonprcfrrential polish with JIgO paste. Photo- graphs were made of the same rrgion with riormal firld alone, follo\ving, respectively, several saturation reversals in the [100], [110], [010] and [i10]

12 SURFACE MAGNETIZATION AND BLOCK STRUCTURE OF FERRITE

directions lying in the plane of the disk. The regions selected, Fig. 10, show the same scratch near the left (west) side of each photograph. This scratch was made by some contamination of coarser emery on the 0000 polishing paper.

The part played by magnetic history is a t once evident. Lines tend to form perpendicular to the previous direction of saturation.. Therelore they appear as two sets of parallel lines or as two sets of steplike lines depending upon the direction of saturation, the latter corresponding to saturation along face diagonals of the blocks. In the region of the

W. C. ELMORE AND L. W. MCKEEHAN 13

- s*: - p " P

scratch, howevrr, thr situation is not so simple. The lines have a greater spacing with a fine stnicturr showing hetwrrn thrm. Lines perpendicu- lar to thr scratch orrur most frrq~~eri t ly wlirri thrre lias been no compo- nent of tlir prrvious nmagnrtic firld along tlir scratch.

Thr rx~~rrinmrnts with a trarisverse firld emphasize the stability of tlmr block boundaries in tlie crystal latticr. Tlir individual block has bcrri found to krrp its identity throughout any arbitrary magnetic treat- mrnt. Tliis conrlusion accords with tlie rxplanations proposed for the fine structure of linrs and for thr dot strurture found in patterns with widrr linr spacings.

\Ye l i o ~ ~ r d tliat more infornmation could be o1)tairird by heating a crystal a1)ove its Curie point. In particular, do th r blocks rrtain their identity aftrr losing and suhscqucntly rrgaining tlicir spontaneous magnrtization? Tliis qurstion rould not b r ans~verrd with certainty, siricc lieatirig thc polislird sprcinmmi A3 iri vacuum hy an iriduction furnacr rrcrl for as short a timr as onr minutc a t 850° C. rcsulted in random recrystallization of tlic polished surfare and consrqucnt disapprarancc of tlic mltgnrtir structurr pattern. Complctc rcpolishing, however, rrnloved all thr ncw crystals arid yieldrd tlir usual mazr pattern having tllr same average sparirig of lines as in the original state.

Thc inlportarlt part tliat magnrtic history plays as compared with ~~olisliing procrdure has l)rrxi shown. Tliis nc1v result suhstantiatrs the ronrlusion already rrarhed that the rxistrnce of a thin polished layrr suitahly producrd is mrrely incidrntal to thc production of patterns and riot an cssrrltial condition for block formation.

An csplanation of the rcmarkablc stability of tlie hlock boundarirs is rallcd for. \Vliat dctcrmines tliat tlic clcctron spiris rrsponsi1)le for frrronlagrirtism (according to the Hrisenbrrg-\Vciss t h ~ o r y ) ~ should srlcrt only rrrtain pla~ics in the crystal for rrvcrsal? How dors it happen that somr stray firld is prrsrnt a t the houndarics of the blocks rven whrn they are arrangrd in scqucnccs? Evidently tlic boundary rrgioris arr in sonir way ~~tlruliar. \Vc suggrst tliat forcign atoms scgrcgated along thrsr plants may be rrsponsiblc for th r stability of tlic mngnrtir block str~irture. Kvidcncr for tliis sort of srgrcgatiorl has bccn found in non- fcrromagnrtic crystals.l0 A possil)le mcchariism for such a srgrcgation may hc found in the systcriiatir imprrfcction of crystals that Zwirkyll ad\.ocatrs, or in a linrage tlicory dcprriding on conditions during crystal growth, such as Burrgcr1? lias advancrd. Since demagnetizatiori rrquires minimu111 stray firld to hr prrsrnt in the crystal, the boundaries, if tlicy brcome fixcd brlow tlie ('lirie point trmprraturc, would tend to assunic thc synimctriral arrangcmcnt (ronsistcnt with the crystal lattirc) that would bring this shout. If, as srrms more probahlr, segrcga- tion occurs a t higher trnlprraturcs, some nonmagnetic mechanism must he invoked.

W. C. ELMORE AND L. W. McKEEHAN

REFERENCES 1. L. v. Ham6s and P. A. Thiessen: Ztsch. j. Physik (1932) 71,442. 2. F. Bitter: Phys. Rev. (1931) 38, 1903; (1932) 41, 507. 3. N. Akulov and M. Degtiar: Ann. Physik (1932) 16, 750.

R. Becker and H. W. F. Freundlich: Ztsch. j. Physik (1933) 80, 292. K. J. Sirtus: Phys. Rev. (1933) 44, 46; (1934) 46, 565. N. Muller and D. Steinberg: Zhurnul Eksper. Teoret. ~ i n ' k : (1934) 4, 717, (In

Russian); Tech. Phys. (1934) 1, 1. (In German.) N. Akulov and S. Raewsky: Ann. Physik (1934) 20, 113.

'

S. Kaya: Zlsch. j. Physik (1934) 89, 796; (1934) 90, 551. 4. L. W. McKeehan and W. C. Elmore: Phys. Rev. (1934) 46,226. The photographs

are of the specimen here designated A,. 5. L. W. McKeehan and W. C. Elmore: Phys. Rev. (1934) 46,529. 6. R. F. Clash, Jr., and F. J. Beck: Phys. Rev. (1935) 47, 158. 7. K. J. Sixtus: Reference 3 above. 8. E. C. Stoner: Magnetism and Matter, esp. Chap. XI. London, 1934. Methuen

and Co. 9. E. C. Stoner: Reference 8, 413.

10. A. B. Focke: Phys. Rev. (1934) 46, 623. 11. F. Zwicky: Helveticu Physicu Ada (1930) 3, 269; (1931) 4, 49; (1933) 6, 210;

(1934) 7, 294. 12. M. J. Buerger: Zlsch. j. Kristallographie (1934) [A] 89, 195. .