GROWTH OF SINGLE CRYSTALS Fi{OM MOLTEN METAL FLUXES
Transcription
GROWTH OF SINGLE CRYSTALS Fi{OM MOLTEN METAL FLUXES
11."u.I/1oul.. (I" 'lrc# III" He , ",,,I ( h,.",,,,,, II/ U." .. f&lT,/B . n,l. I! ""ull I:." ;':.A (,\.}",.·"[,,.'t." ,utrl I. 1.''''111':': V I. /)n le'I S, " ',,,,' I'"bfl\ ,'''·' \ /l , ' . /'.IS'J Chapler HI GROWTH OF SINGLE METAL FLUXES CRYSTALS Fi{OM MOLTEN Z. FISK Los Alamos Narimwl Labormory. Los AIcIllIOS. NM 87545 USA J.P. REMEIKAt AT&T IJdl I.lIbUrtlWfic's. MurfUY Hill. NJ 071)74 USA ,. .,. Contt:nls I. I nlro<luclion 2. Dcs~ri'llion of lechni'lue 2. ' Examplc> 2.2 Choicc uf Ilux 3. Further ;sSp.:(h uf "pIal gro .... lh fcom melallic lIuxe> 3 . \ Diagnlhlics 3.2 COlllaillcr~ 3.3 1-1u\ incurporalion 3.01 Slulchiollll:lry ;) 3.5 s<,'Jraliun of cr)"slJh all.! Ilux 65 S5 3.b Solulion kinelin 66 63 63 63 4. Exlen>ions of Ihe lechnique -1.1 Lbo: of lemperalure gr;sdicnl 4.2 Evaporalion growlh -1.3 Trav"lin~ solvenl melhods -1.-1 Miscellaneous Appendix I {o-! Appen.Ji~ bS Rderc:ncc~ 55 62 67 67 67 67 611 68 69 2 70 I. Inrroducrioll In rt:se;rreh on the physic,,1 and chemic:11 properlies of illlermeiallic (an~ olher) compounds, materi .. 1 III the imlll of singh: crystals is preferred. Ceramic or pressed powders of a given compound. wherein Ihe p.. rticles lIrt: randomly oriented. not only present a much higher SlIrC:lce to hulk ratio than do single cryslills; tll\;Y "Is" prescnt prohlems of inlcrgranular composition and porosity. and prevent any lIleatlillgfllln\\.';I~uretllcnts of ;\Ilisotrupy. A generally higher level of purity is obtained in ~itlgle crystals. ,mil, IIIl.1rt: importantly. lIlany macroscopic and (parlicularly) microscopic probes can only be effectively utilized with single erys"ll mah:rial. FClllli surfacc cXl'erimellh. for example. using till: de Haas-van Alphen effect, arc in this catcgory. as are many neutron ;rnd X-ray scattering t Do:ceascJ . _ .' _ .... ' r.... _. •.. ~ Z, FISK ~ ",J J.I'. Rt.:~lI' I"'\ e~peril1lents . Th i~ m;lkes the preparation of ~ ill g~e erys t;,ls a mailer uf conside r- ahle illlpOrl~nee for Ihe condensed llWlIer SClentl)!. The II1mt import;lIIt ;ttuJ )implnt technique for crptal growth i~ gruwth. from ;1 pur e lIIelt o( the ~uh~t;ll1ce uesi reJ . Even he r~ there ;Ire nutllen.u) e"per~mental tedlniques ;lvall;lhle, lor exampk Z\llle me ltlllg, cry)tal pulling alit.! Unt.!gm;111 cooling (sec Cit . SO by Abell ill Ihis volume) . Uul such techniques ;Ire ge nerall y limited to cases whe n the material is congruent ly melting. And even in Ihese cases the re can he complications; the male rial lH;.!y melt ~II an inconveniently high lemper;l\ure 01 Ihe vapor pressure of one or more of .the .comrone~Hs may be pruhihilivcly huge al the mehing t elllpcrat~ r e . In such S tt~ ;!II.ons ;lIId In the more usual casc where the compound mclts ,"congruend y, II IS useful and often nccessary to h'lvc other techniques aV;Jilab[c for the preparatio n o f single crystals. O ne such techni(IUe which is not widely used is Ihc growth of single cryst;,ls (ro m 'molten metal tluxes. We Iwvc no explanation fo r the relalively smit1l ;Bte ntion given \0 this techniqu..:, which requires wlwt amounts 10 the determina· tio n of pan of the approp riate phase diagram. ;IS do many cr),stal growlh investigations. [t is ou r purpose here to describe somc o f our experie nce in usin~ Ihis teehni(]!le ;md give sUllie genera! gu.i\J.e lines a~ to how we h'l\'e galle about uetermini ng a ppro pria te sul \'ents (or n·ux.:s) in p;Jrticular cases. This choise Ill" solve nt is d early the p\;lce where expe rie nce and imagination arc pUI to the tcst. and it is si mila r 10, bUI simpler Iha u, the cas~' wi th oxide s),stems. II is nOI our imc nt ;on to provide an introduction to the theory and practice of crys t'II .BrOwt h. Such information ca n be o~ t ai l1 ed from a la.rge nu mber of .good rderenees (for example , ~ee E!wd[ and Sched 197)) . It IS rather our aim to illliiC;Jte Ihe general usefulness of the lechnique of growth from molle.n mel:l!s nu ~es from Ihc pwctie;![. how to do it. side , and to tabulate a number of IIlSlanccs where it has been successfu l. In this last , we arc by no means (:.unill:'lr with alt the wurk III this al\:a . Compuler searches of Ihe [ite r:llure actually turn up only literature refe renceS that we arc alrc;{dy fami liar wilh. The most effecti ve way 10 reseil rch whe ther a mClal rlu .~ method has been successful fo r single crys tal growth of a pani cul"r compou nd is 10 se.treh Chemica l Abs!faets unlle r th;Jt comro~n~. We ha\'e not done this compound by compo und search, and wh;J t we present IS III laq;e pan drOlwn din.'ctly frum our own rese;t rch work. This em ~ha s i s on our. persulI:.! 1 experience is in nll w;ly me'lllt to draw atlentllJll frolll Ihe Imp(ln"ll~e til other wo rk ; r;Jlher it rep res<.'nts how we came to the ~ubject, and it serves our pa rt icu lar purpose in this d l;Jpler. Wltal arc lite atlv;ml;)ges of this particul;tr technique? The re arc IWO gen<.'ral ones: (I ) m<lleri :..ls l';m he grow n often well below Ih"';r melting points, and this tlften proow:es m:.tler;~1 wilh fewe r defects and much less therm:! 1 strain. and (2) t1lOlten metals Ilffer a ckim environme nt fo r growth. since the OIolte n met:ll nux of len ge tl ers implllities which do not subsequen tly ;lppl';l r in the crystal. In litis rc~ peel. a molten tllct"ltlux e;Jn ;tel. purity-wise, like a good vilcu um . In addi tion tll these adv<l nt:.t ge~. there is the point 10 be made in f",\'o r of metal nu .~ growth that il is <l "poor man's" tedHli4ue : it can be done with the simplest equipment . often e\'ell ..... ithout :t temperature cOfltrolkr . although OJ conrroHer is necessary fo r Ihe growth of I:uge cry~ l .. b. This enables a 1II0de~tly e(luippeJ ht boralory 10 get GI\(IW'111 (I I ~ 1 t"(;I. I . c/(YSr,\I.S 1·ltO:.t MOLTI, N MET,\ I. .... IJXES inlo th.: l.lUsi n e~s ,If ~i ngle ("f)'~t;t1 growth wllhout l;J lge r;JpilOlI equipment expenditure. an impllrt:'1111 :'ldl';lIItage ~1Ill"e Ihe mO~1 reliable ~uurce for ~ ~ingle crysta l is from your own <)per'llion . To the ahu ve po i nl ~ can h<.' added tlwl O\x;Js;onally ut her I<.'chni(/ues will not yield cr)·,lal,. Ihal v:lpl1r pre,~urc rrohlern~ (:'I ~ in a number of Yb ~l1d [u compounds, I'or ex ample) arc often greatl y ledueed in ~olutiun ;Hld thaI in it num be r of cases cumpletely new compounds have b<;<.'n discovered growi ng fro m met.]1 flux s}'s lem~. There arc. to hc SUfe. a number of di~;Jdvat1tage s to the tcch nil/ue . The first and fo remost is IIt:'it it is not always an applicable me thod: an appropriate mclal nux from which the desired com pound will crystallize may 110t be found. In addition, difli cu lt ie~ arc encou ntered with so me flux choices, when the flux, u r some component of it. enters the cryswl;Js an impurit y. A further problem havi ng to do wit h the ;Jctual growth siluation is tha t !lux inclusions often occur - the cryst:J1 grows arou nd a pock<.'t of flu .~. In thi s case speei,tJ care mUSI he t:lke n with the grow th ratc conditions. The prohlem is reall y on<.' of excc~s i\'e Jl uclc;lt ion which takes place either due to ](10 f;Jsl a COOling rate , or supe rcooling of the mclt by slow COOling with subsequen t mu[t iple nucle.l1io n and fas t growlh of large bUI imperfect c r ~'stals usual ly co rt1 ;Jining inclusions. It is under !h<.'se conditions thaI the large crystalS arc imperfect. but aflcr equilibrium is reOlched cryslals nucleate at a much slowe r rate . :lnd ilre of much higher quality. The rlux gro wth method :llso s uffers from :t problem common to o ther growth methods, namc ly the container problem. which will be discussed at some length b<.'low. Findi ng a container 10 hold th e melt at the tempera ture desired can be difficult, as secondary reactinns oftcn occ ur which one docs nOI irnrnedi:ltc1)' expect. And fin ally. Ihe .. hilit )' IU se p;tratc crys tals from thc flu x at the conclusion of growth nec tJs special consideration, :md is also discussed at lengl h helow. 2. I)cscri[ltiull ur tcchn ill uc 2.1. £XQmp/t'l' Whcn <.:ry~tal) grow tmnl .1 mdt. we learn somet hing :lholl t Ihe re!cl'a nt ph;lse diagfOlm. Ctlnvehely. knowledge (If Ihe relevant phase diagram will allow (I nc to SCI ca refull y some o f the conditions for eryst;ll growth. We seldom havc, needless to sa y. Ihe pllilse diagrJ m information necess;tr y, and il is often muc h (as ter 10 optimize growth eondition~ without le;Jrning much in detail :lbOIll .Ihc phase diitg ra rn . This is in pa rt becaus<.' fJelOrs othe r Ihan tho}e cont:'li ned in thc plwse diag ram J lso contribute to the proce~s. Neve rlhc!css, at:eess to the tabulations of bin;lry ph'lse diagrams (i.e. Hanse n 1<)58 , Elliott 1%5. Shunk [ %1) , Jnc Moff.llt I<JH I ) is extremel}" useful when expe rimcl1ling with possible !tuxes for growth o f Ol p;lrticulJr malerial. The simpiesl W:1Y til introduce !he concepts :Ippropri;lle to gfuwth of crys tals from metal fluxcs is with a simple bin;Jry eute~·tic system. Suppose we wish to produce crystals of Si. We krm w pe rfec tly welt that there is an <.' llIire indus try bui lt Z. FISK .. nd J.P. 56 Hl~tElKA (ilHlWIIi 01 SINGLE CHYSTALS fROM MOLTEN METAL !'LUXES on extremely high quality Si single I:rystals. Nevertheless. we might wish to Jlerform some experiments on small crystals which we could easily produce ourselves. 5i is chemically rather inert with respect to acids such as HNO., • Hel and basic solutions such as NaOH. This would allow. us to extract Si crystals from a metallic flux which is attacked by the above reagents. The simplest metals to consider as solvents are those with relatively low melting points: AI. Bi. Ga. In, Sn, and Zn. Zn has a high vapor pressure and is a poor solvent for Si. as we find out in Hansen. Bi is also a poor solvent. It dissolves about 5 alo Si at Hloo°C. Ga about 20 alo and Sn something in the vicinity of 5 a/o. AI appears to be best in this respcct. dissolving about 45 alo at HXIQ·C, and the cuteclic composition . contain~ 11.3% Si (fig. 1). This makes the AI flux especially attractive for growing Si single crystals: largl! solubility al a reasonabll! Il!mperalure. ease of removal and rc:asonable composition range belween euteclic composition and solubility at staning temperature. It is clear that if this last range is small, liltle material will bl! available for crystallization before the entirl! melt solidifies. The small scale laboratory procedure for this AI flux growth of Si crystals then involves wr:ighing out Ihe appropriate amounts of Ihe elements corresponding 10 the templ!rature where you wish growth to initiate and placing them in a suitable crucible. AI will attack Si0 2 when molten and SiO, cannot. therefore. be used. AI 20) crucibles ure readily available' ii1id will not b~ allacked by AI until, in ?ur experience, temperature!> near 1600°C when some solubility of AI~O) in AI begins and where the vapor pressure of Al becomes an additional problem, partly because the thin oxide skin. which often covers molten AI, breaks up and loses its integrity. resulting in a much increased evaporation of AI. A protective atmos'600~----~----~--~~---,-----, 1400 E 1200 w a: :) !< 1000 a: w Q" ~ u.I I- 800 -1111/0 Si 400 0 40 20 alo AI ..' . . ~ eo 60 Si .: 100 Si .. ' . Fig. I. T.:mpcrOilure-comp..,siliun phase diagram rur Ihe AI-Si sy" lem. (Arter lIan~en 1958.) .. '.. ~. . . 57 phere must he pruvided for Ihe mel!. and inert gas (HI! or Ar) is preferrl!d over vacuum bccau:.e more AI cV<lporalion will occur in vacuum. The inl!rt atmosphere clln either be pruvided by scaling the crucible in a quartz tube. simply done at a glass bench, ur using a gas atmosphere in a lube (urnace arranged vertically to .lCcommodate the crucibles. Horizontal tube furnaces can also be used if their bore is suflicient to accept lhe chosen crucible size as il stands up. In this case, the crucible(s) CHn be loaded in un a ceramic boat. There is an important point to remember when scaling a crucible into a quartz tube: the tubl! as it stands vertically will contact the AI~O\ crucible bottom at its I!nd taper. AIlO) has a lIIuch larger thermal expansiun than quartz and will crack the quartz as the load is healed. This cracking can be prevented either b) flattening the quartz tube boltom so that the crucible can rest there without contacting the sides, ur a short stub of quartz can be used in the lube bollom to spring Ihe crucible away from Ihe nesting location. We also find some AI vapor attack on quartz tubes from the AI in the crucible load . This attack is slower. obviousl~·. when the evaporation kinetics are hilllil:red by an inert gas atmosphere ill thl! quartz tube. It is reilsollilbll! to OIdjusl the gus pressure in the tube so that it is of order 1 atm al the highest working temperature. although propl!rly seah:d quartz tubes of -I em internal di:tmetcr C,1n easily wkc 5 atm at tempera.ures up to 1200°C. While it is possible 10 usc quartz above 12(J(J°C. Ihis is the range above which quartz starts to soften. Then it is important both 10 keep the quartz tubes scrupulously free of fingerprints and surface dirt (locations wherl! devitrification starts) and to carefulI~ adjust the S,IS prcssurl! to be 1 atm at the operating temperalure to prevent eUher the collapse or Ihe elCpansion of Ihe tube. Il is gener>llly. bUI nOI always. Ihe case Ihat the larger the comaincr. the larger the crystills. It hecumcs dimeull. if you are not a professional glass blower, to seal off crucibles lurger than aboUi 2 em in diameter in a quartz tube, and for crucibles above this size it is much easier to work in a tube furnace equipped with a tube through which inert gas is flowing. Mullite tubes (an Al 10)-Si0 1 ceramic) are relatively inexpensive and can be used to at least 1500·C. a common upper temperature limit of operation for furnaces heated with silicon carbide heating elements. Mullite lubes becoml! somewhat porous at l!Ievated temperatures, so that a slight over pressure of inert gas is recommended at all times. It is also important to rl!alize Ihat there is usually present trace amounts of unwanted gases such as H~O and O~ in high purilY He and Ar deliverl!d from gas bOllles. The gas can be first run through a commercial gas purifier. or one can be conslructed by packing a quartz tube with granular Zr metal, which is heated in a small tube furnace to ~!OO°C and through which the gas is first passed. Evcn at the ppm levcl, 01 can be gCllcred from thc gas stream by the metal flux. It m'lkes senSt, therefore. 10 not Ilow more gas past the crucible thcn necess'lr),. This can be accomplished by cilUsing the gas !low to by-pass the lube. while maintaining a positive pressure within it. as shown in fig. 2. 11 is useful 10 know that the McDaniel Company which sells mullite tubes also sells gas lighl end-cap scaling fixturc:s for their tubes which requirc no additional preparation of the mullite tubes. . ..' . ~ . .:. . . . (iIHIWTII III , SI1"(;U: CJ(YSTALS FI(OM MOl.TEN III1;TAL FI.U .X Es Z. FISK an.! J.P. ItEMEIKA 511 FURNACE CRUCIBLE_&=--HH CHARGE EXHAUST GAS 'gm Fig. 2. Gas tlow arrilngemc:nl 10 milinlain posilivt cuver whil!: nOI pa~sing gas over hOI crucible and charge. Lower villve opo:n and upper valve closed allows for initial Hushing of spa~c. Illc revers.: .. rrangemenl being used 10 mainlain intcgrity of gas cover during heilling and cooling cycle. The: t:xil gas ~.. n cunvcnie:ntty be c:xh .. uscd Ihrough a bubbler, allowing monitoring of flow ratc. Once the crucible has been loaded into the furnace, and the protective gas covcr established, the furnace is then ramped to sufficient temperature t\l get the Si into solution. The kinetics of Solulion of various materials prove to be quite sloW in many cases, and some experimentation is often required to establish the appropriate soak period' required for complete solution. Then the furnace is ramped down: the slower the rate, generally the better the crystal growth. the fewer the inclusions and the higher the quality of the crystals. For the example given here, rates in the few degrees per hour should yield mm size crystals. although some crystals will be obiained if the furnace is just shut off at the soak temperature and thl!n allowed to cool ~t its own rate. Indeed. for exploratory work aimed at determining the feasibility of using a particular solvent, it is often useful to use coolings rales of lOO-200°C h -I. This normal rate of cooling for a furnace can be slowed somewhat by increasing the thermal mass present, such as packing the crucible in Si0 2 sand. Once the crucible is cold. it is then necessary to remove the AI. This is readily done in this case with NaOH solution . If thl! leaching reaction is slow, heating on a hoi plate will speed things up. The Si crystals will remain after Ihe reaction of the hydroxide with AI is complete. Crystals will in fact be visible on the surface of the melt before the leaching is begun. This is a particularly simple example we have given of the teChnique, but there is re:llly nothing additional involved in the general approach to crystal growth by the nux methud. only the pha~e diagr.ulIS becoming more complic'lled and orten 1I0t known. and the sepmution of crystals requiring more care. An exampl.: involving unly slightly different considerations is Ihe growlh 'o f VAl" crystal!. from AI (Iig. 3). In Ihis case there arc iI number of intermetallic phases present. VAl!. VAl). and VAI~. the last two melting incongruently. II turns out to be diUicult to obtain clean. single phase samples of VAil by arc melting. presumably becliuse the UAI" peritectic horizontal intercepts the liquidS lit a composition (-17% U) rather far from the composition of Ihe compound. BeclIus!! of this. even if one has no interest in single crystals of VAl}. if you want to know about the properties of the material it is best 10 attempt the single cryslal growth. solely to gel single phllse material. Similar considerations apply for a number of Yb compounds where the high volatililY of Yb makes arc melting an unattractive melhod for preparalion of single phase, peritectic compounds. In the growth of VAl) crystals. we sec that we need to prepare a melt with U concentration between -3% lind - 10%. Additionally. especililly if a quartz tube is used w conwin the AI~O\ crucible. it will be advisible to keep the growth temper.llurc hduw I 200°(', Thi~ rest rielS the U com:entwtiun betwecn -3% am.I _7 t y(). This is it narrow range from which III grow crystals. but it is easily done. the peritectic temperature of UAI 4 , the When the melt has cOllled to furnilce is shut oH to discourage growth of sizable UAI 4 crystals. VAI 4 crystals. to be sure, can be grown from melts with V concentration between 1.7% and 2,6%. Again. extraction of the crystals can be accomplished by leaching out wilh NaOH solution. A point to nmke hcre is thaI it is necessary to pay attention to the chemical no·c. 1600 W II: ::> ..... < II: W Q, ~ ..... 6000=----~2~O~~~~--~~--~B~0----~100 AI alo U u Fig. 3. Th~ U-AI binary phas.: diagram, (Alter lIao~~n 19511.) GI{(l\\~nf OF SINGLE CHVSTALS FHOM MOLTEN METAL FLUXES Z . FISK and J.P. REMEIKA activity of spccies dissolved in the flux, in this case U. U is often quite r~~ctivc towards crucible materials, including AI 20). In this case. however. Ihe activity of U is suffociently reduced by solution in mohen AI so as not to be a problem . This sort of problem is generally worsened al elevated temperatures, and one can often detect attack on the crucible material from a discoloration of it. In some cases no product is obtained from a growth attempt because one solute in the melt has reacted out completely with the crucible. . • The actual composition and temperature range c~)Ovenicntly availablc for gruwth of UAI) from excess AI is quite restrictcd. We haVe therefore investigated other possible fluxes from which to grow this material. A generally useful flux from which to grow a number of U compounds has proven to be Bi. In fig. 4 we show the U-Bi phase diagram. There are twO peritectic intermetallics, UBi 2 and U )Ui •. UUi z melts at a rather low temperature, and the ability to grow intermetallic compounds of U from molten Bi is presumably due to the relative low stability of UBil' the U-Bi compound most likely to be in competition with the compounds one wishes to grow from Bi. In fig. 5 we show the AI-Bi phase diagram from Shunk (1969), two differing boundaries being given for the miscibility gap above the melting po.i.W of AI. What is important here for us in that something like 13 alo AI can '6e dissolved in Bi at the melting point of AI. 1~00r-----r-----r-----r-----r---~ \200 ..... w -I -I' 1200 l- w a: :l ..... < a: ~ 1000 ..... I Q. ~ W 600 t- 400 200 0 20 - iii :J ~ 800 ~ W t- 600 - 40 0 20 60 ~O alo U 80 100 U Fig. 4. The U-Ri binary pha •.: diagrllm. (Afler Hansc:n 11J5!i and Ellioll 11}6S.) .. .: - ~. 40 60 alo BI 80 100 BI Fig. S. The AI-8i binary phast diagram. The IWO lines shown .m from IWO differcnl in\'esligalions 01 1he syslcm. (Afler Shunk t969.) Our original attempts at growing U-AI intcrmetallics from OJ were aimed al producing UAI 2 crystals. For reasons we do not fully understand, only UAI crystals were Obtained, even when compositions with U-AI ratio as large as UzA were tried. This is an interesting fact since, as we discuss further below. it is mos often found that the congruently melting species in a binary phase diagram (hen UAI 2 ) is the must likely to precipitate from the ternary melt with the solvent. Till crystals of UAI) obtained from Bi flux were rectangular bars. with latera dimensions oC one or more millimeters and length in excess Qf 1 cm. Tbl extraction of UAI) from the Bi flux is done by solution of Bi in moderately dilutl HNO). It is a fact that many U intermetallics are quite stable chemically witl respect to HNO). It is curious that many of these compounds will dissolve readil: in Hel. There is an alternative way to remove the Bi flux which proves to be extreme I useful in situations where the crystals desired arc anacked by HNO). This is b spinning off the nux in a centrifuge, and is useful for any low melting flux. Whll one does is to place the crucible in a furnace which is set to a temperature ju~ sufficient to melt the flux, then place the inverted crucible in a centrifug containing a base of glass or quartz wool. or glass beads, in the bOllom. and spi out the molten nux. This is best done in an appropriately sealed glass tube wit I the coarse filter material in the bonom. but can be done in air when the rate c oxidation of the crystals will be small. We have been informed of situations wher iii .. , a: w I~.. .. , ~ 800 I ;:) , ::I TWO LIQUIDS 1400 ..... ,, TWO LIOUIDS a: AI 1600 61 - ~ .. ,. ' ... :: '. -:~ , . - b' . Z . FIS" ~".J J .I', REMclKA heat WilS directly :Ipplied 10 the centrifuge buckets. which wUl~I~ .a~low solve.:llIs wilh m.:iting poin" as high as that of AI (66()OC) to be spun ort . I Ills could be ,iI llangerous prm:edure. and should be used only as a last resOrl. In such .a c.. se.:~ 1\ would be lIel'cssary to replace.: thc SiO~ wool with AI~O ,1 granular matenal. which is nut ilttackl'u hy multen AI. A cOIH:i'e descriptiun uf growth of Rln} alld RSn, from excess III .. nd Sn. res,?ectivcly. can he.: found in Kletowski et al. (l9~5). 2.2. Choice of flux The general problem faced by materials physicists and chemiSlS needing single . crystals is. of course. the best way to grow the parlic~lar crystal th~y n.eed. There are several general observations' which can help 10 the exalmnallon of the possii>ility of successfully growing crystals from mol~en metal. Iluxes. The first is that it is generally easier to grow a congruently melting matenal from such a flux than an incongruently melting one. and that the higher this melting point is. the morc readily the crystals will come out of the melt. We saw an exception to this above with the example of the growth of UAI) from Bi, when UAI! was the material expected to grow from Di. It seems likely that at some presumably large concentration of UAI, in Bi. it w.ill'·~e po~~ible to precipitate VAl!. The temperatures involved ~il1 probably be quite high. and it would then be better to usc anuther growth technique. The second, fairly obvious, observation is that competing ph:lses formed by thc flux with the elements in the compound wanted can prevent growth of the desired compound . This translates into looking for potential solvent-solute binary ph<lse diagrams with fairly low-melting. incongruent compounds. An cxample illustrating these two observations is. the growth of rare c<lrth borides from molten AI. For the light r<lre earths. La Ihrough Eu, th.e hexaborides . grow. Although then: exist several AI borides. AlB! is the one most likely to cause trouble. and this melts incongruently at 975°C. These rare earth hexaborides mclt congruently at temperatures ovcr 2000°C, and in fact their solubilities in AI an: quite modest. For example. LaB b has. very roughly, a SOlubility of 14 mg per gram AI ncar I 400°C. The interesting point is that for Gd and beyond. it is the h.:traboride that comes down [rom the AI !lux, and it is with these heavy rare earth borides that the tctrab~ride be~om~s cong~ucntly melting rat~cr than the hexaboride. whi.:h is congruently melting m the hght ran: earth series . The cross over occurs near Sm. and it is possibk to grow either the tetraboride or hexaboridc of Sm by varying the composition of the charge. We also note that it is possible to scparate. below the ppm level, Sm from Gd by growth of the~e borides: Gd will not dope into SmD/> ill AI melts. as has been shown by spm resonancc experiments. The Gd comes down completely as h!traboride. An aside remark here is that Eu. which only forms an hexaboridc. is <I very volatile clement. resembling the alk<lline earths, and growth from AI is. a very convenient preparation procedure for EuB h • because the large disparity in melting points between Eu ilml 13 makes arc melting without large weight loss dimcult. b .l There ilrc further a~pcet~ Ill' the growth of rare e<lrth borides from AI which arc worth Ill\:nliollilll:!. One point is Ih:lt in Ill:llly neulron experiments. il is important t\) lise Ihe II B i~otupe ,ince the abslIJ ption of neutrons by '''n is so high. It is oftcn flluml Ihat Ihe "B isotope onc CilO obtain has much lower purity than desired. i\lld thai iI collVcni\'nt preliminary step of purification can involve a . buride ~rlIwth whidl lea\'e~ Ihese impurilies behind. This has beell done in the preparation of fille earth- Rh. B. compounds. A sccond point is that initial attempts to producc YbB~ crystals from AI produced insleild a ternary phase, YbAIB~. This proved to be an interesting phase in its own right. and it was possible 10 also produce this phase ....;ith several other rare earths. In order 10 grow YbD. from AI, the starting growth temperature can be in~'reascd by going to higher YbB. concentration in Ihe AI melt, so that crystallization StilTlS above the melting point of YhAIB •. Alternalivc1y. YbB. can be grown from molten Yb in a scaled Ta tube . A third general observation regarding metal Itux growths is that certain metal lIuxes work well (or compounds containing certain dements, Di for U compounds. for eXit/nple. For this reasun we have organized our listing or Itux growths in appendix ) by thc solvcnt Ilux. 3. Further OSI)C~·ts of crystal growth from metallic '''Ixes 3.1. Diagnostics In order to determine the suitability of various !luxes for the growth of single crystals of a pilrticulilr compound. iI number of small SCil1e and lairly rapid trials are generally required. Crystals produced in this way will most often be small , and it is useful in such l'ilSeS to examinc the melt microscopically. This is I!specially the case when Ihere arc problems associated with chemically separating the crystals from the Ilux. \Ve haw found it worthwhile in these situations to saw with a diamond wheclthruugh the crucible ,lOd melt. and then examine the cross-section after poli~hing it metallurgically . A cut vertically through the crucible gives the best r~prese/ltaliun IIf what is actuillly happening in the melt during cooling. II is perhaps not surprising thut crystals often grow much larger in melts placed in special locatioll~ in a givcn (urnace duc to the local temperature gradients, and aspects of this will be visible in such mctallurgical examin<ltions. Elemental analysis of these cross-sections in ilO appropriately equipped SEM is also useful. 3.2. COlllllillt'rJ Cont .. iners ilre iI central problem in crystal growth using metal fluxes. Aspects of this have heen mentioned ilhovc. A short list of useful crucible materials is given in appcndix 2. It is important to realize that solute reactivities in different melts vilr)' cnorrnollsly. Fur eX:lOlplc, it was found possihle to grow large rare Z. fiSK ;10.1 (,HOWl II 01 51:-;(.1.1: nnSTAlS IHOM MOl:lt:N Mt:TAL "LUXES JP. REMEJI'A JA. Swidliollll'''Y earth rhodium stannide crystals (rom excess molt..:n lin in scaled ljuanz tuhes. Some rare earlh allack on the quartz was observed. being greatest for Ihe light rare earth compounds. lIut this allack was sufficiently slow to allow large cryslals to grow anyway. In vi~'w of the exceptional slability of rare earth oxides. this . indicales that the rare earth-tin interaction is quite strong. reducing Ihe rare earth activity in Ihe melt substantially. By contrast, rare earths and U dissolved in mollcn Cu arc found to be quite reactive towards container materidls, and this might be guessed at by the low melting temperatures of compounds of rare earths and U formed with Cu. Crystals of rare earth-iridium C-15 Laves phase compounds have been grown from molten Cu. Ir dissolved in Cu will attack Ta. but the rate of this attack depends on which rare earth is in .the melt. This attack problem ",as found to be severe for the case of Celr!, butllot for Tmlr 2 • We were not able to find a completely suitable container for the Celr 2 growth. Ta. it should lie ~aid. is a convenient material to work with because tubing is readily available, and it is easily sealed shut in an arc furnace. When welding a mell load ~hut in Ta in the arc furnace. it is useful to hold the broad, flattened part of the tube shut in a clamp made from two Cu bars. This large thermal mass keeps the tube cool and prevents expansion of gases inside the Ta tube from spreading it apart in the region to be 901e'ia"ed. Mo and Ware also useful crucible materials. but arc not ductile enough to crimp flat for welding. These crucibles can also be welded shut in an arc furnace if a titted lid is fabricated. Some kind of copp<.:r clamp for holding the crucible while welding it shut is important: This hold!:r should be bolted to the <Ire furnace hearth for effective heat exchange. since keeping the crucible charge cooled is important for these high temperalure melting materials. Electron beam welding is, of course. a better alternative. but this is not as readily available. Many malcriills exist over a range of composilions. This variable sloichiometry is a source of probkms in /lux growth because the melt composition as well as phase boundaries vary during the cooling cycle. This sometimes results in a composition gradient ill the grown material. In a number oC cases. one composition boundary is a vertical line in the phllse diagram. This allows an excess to be added to the melt corresponding to forcing the growlh always to occur ar this limit of stoichiometry. II appears. Cor example. that the compound CePt 2 can form with deficient but not with eXCess Ceo We have grown crystals of this material from Pb melts. and found thaI we could approach quite closely the theoretical stoichiometry of 1: 2 using large excess of Ce in the Pb melts. It is important to realize that these excesses may be much larger than one might cxpect, as we have found that certain melts "hold back" certain elements - namely. the material that grows is very different in composition from the charge. The other important consideration concerns substitutional incorporation of flux in the growing cryslals. For example, it has been found possible to grow CeCu 2Si 2 single crystals from In or Sn !lux. Analysis shows that small amounts of In and Sn enter substitutionally into the CeCuzSi z. The propeflies of CeCu zSi 2 are particularly sensitive to such subslitutions. and for many experiments these crystals. therefore. proved to be unacceptable. There arc many cases of this kind of problem. and this !nukes it important to examine each possible nux from this standpoint. Chemical or atomic size similarity of the clements in the flux to one of the components Ilf iI compound should alert one to Ihis possibilily . 3.5. Sep'ITaticJI/ of crystals (l/ltl Jlllx 3.3 Flux ill corporation W.: have discussed in various places above the problem of nux removal. There are two general approaches . chemical and mechanical. Let us discuss chemical techniq ues fi rst. The stability of a compound in various acids and bases can be determined using polycrystalline material. and it is a good idea to investigate this before embarking on a lengthy attempt to grow crystals from a given flux. Specialized reagents do exist for certain elements. and it is often worthwhile to consult a practicing inorganic chemist for ideas connected with chemical etching. There are interesting effects. furthermore. which appear to result from electrochemical differences between fluxes and crystals. In the growth of rare earth rhodium stannides from excess Sn. the eryslals can be leached out with dilute HCI. As long as the crystals are in contacl with the Sn flux. they remain shiny and metallic. Crystals which have separated from the Sn, however, quickly form a black tarnish from which. microprobe analysis reveals. the rare earth has been extracted; lind the blaek material appears to be an amorphous mixture of Rh and Sn. Thus. removal of the crystals from the leaching solution as soon as they are free of Sn nux is desirable. As lInother case. we mention the growth of CeCu:Si: from molten Cu. It was One of the problems attendant to flux growth of single crystals is that macroscopic voids filled with the flux can often be found within crystals, and the presence of these voids is not necessarily evident from a simple visual inspection of the crystals. In the growth the U- and rare earth-Bell crystals from AI, lamellae of Al arc often found. We hav~ often had to separate our crystals from these lammellae by spark machining. followed by leaching in NaOH solution. Ar:y residual Al can be detected by looking for the diamagnetic signol of Al near 1.1 K. The prese.lce of incorporated flux is generally due to unstable crystal growth condilions which usually occur in the initial growth siages. as discussed earlier. It is frequently possible to recognize where this initial growth occurred and mechanit:ally remove' this part of the malerial. Problems of this kind can be ameliorated by providing deliberate nucleation sites. although this complicates Ihe experiment considerably. In the growth of rare earth 'rhodium slannides Crom excess Sn. we were never able to completely eliminate Sn inclusions. except in smaller cryslals from a melt Ihal had reached equilibrium . - - ' '.. ...... . . . _... ::- ' - - . .. . . . .'- - Z. FISK amI J.P. REMEIKA reporled that the Cu stoichiometry of CeCu~Si! crystals .was very n':.levant h) its superconducting properties. We therefore felt that growing Ct.!CU!SI 1 Crom pure Cu would provide the limiting Cu stoichiometry for. thi~ material. The p~obl~m came with the Cu removal. since. although an examination of the crystalhzatlon runs :.howed the presence of well formed crystals. Cu and the crystals Jre chemically allacked at nearly the s,lrne rate. Ultimately. we were able to slowly cxtract crystals using acetic acid-hydrogen peroxide mixtures. the process taking approximately one month. . A less often used. but useful. technique is to electrochemically remove the flux. using it as an anode in an electrochemical cell. By suspending the melt anode in the bath, the crystals can fall free as the etching proceeds, and if this electrolyt.e docs not attack the crystals. they can be easily recovered. The advantage of thiS technique is that the voltage used in the erosion is easily controlled. There ~an also be "protective" voltages developed between crystals and flux of the kmd pointed out earlier in Ihe chemical etching of stannidc melts. . Another tcchnique of some use is the solution of one met..ll by another after the growth is complete. As a case in point, we look at the gro'"":th of Ce.ln) crystals from exc~ss In. The left over In after the growth can be dissolved In Hg. The . crystals can then be removed frol)\. 11te liquid (at room temperature) I~-Hg solution, and the Hg film remaining on the Celn) crystals can be pumped IOto a trap in a vacuum system with gentle heating. The usef"lness of centrifuging off low melting fluxes has been discussed above. Such a process can easily be used in the Celn)-In case JUSt mentioned. There is the additional possibility of flux removal of high vapor pressure metals by simple evaporation after the growth process in a pressure containing vessel. There is also the less elegant method of cuning crystals from melts, exactly as is done from zone refined rods. (it()\\"lIt o. stN(il.l: CRYSTAI.S 1'1(0/1.1 MOI.TEN METAt. I· LUXES 61 solubility t~~I~ of a !:lIl1lpUlIlld in .. given nux. This can b~ done in an approximate manner by suaking arc-melted pcllets of the compound in givcn amounts of nux at lixed temper'lIures. 4uenching to room temperature. ilnd subsequently measuring the weight los~ of the leached-out pill of compound. 4. Ext~nsiolls of the (cdmit,ue 4.1. Use of tempf?rtltllre gmt/it·", A variant of the growth technique utilizes a temperature gradient. This is most easily discussed with an example. In our discussion of the growth of single crystals of rare carth hexaborides from molten AI. we noted that hexaborides of the rare earths Gd and heavier did not occur, rather their tctraborides grew. We did finally succeed in prepilring single cryswls of GdB •. by the following method. B has very low solubility in Ga. even al 15UO·C. We therefore made a flux of ~S alo Ga-5 ala AI in which B will have very low SOlubility. An arc melted pill of GdB~ was preparcd, and its densily is such that it noats on this Gu-AI mixture. Our AI 2 0) crucible conwining the !lux and GdD" pill was placed in a furnace so that the crucible bottom was coidesl. the temperature gradient along the 5 cm) crucible being about 50·C. The top of the crucible was maintained at ISQOoC for several dilYS; mm-size crystals of GdB .. were found growing on the bOllom of the crucible after shutdown. Our idea in this method was that we could force everything out of solution at lower temperatures that we put in at thc higher temperature through the dynamic equilibrium. In a case where the compound is more dense than the ftux. the temperature gradient can be reversed. 4.2. £vaportttiot/ growtJr 3.6. SO/lIIion killetics Some flux growth attempts appear to fail because of problems having to do with slow solution kinetics. There are cases where such slow kinetics can be used to advantage, but more often it constitutes a nuisance. As an example we can look at Cu-B solutions. There are no compour.ds in the phase diagram, and there is a broad eutectic melting near lOOO"~ If we can believe this reported phase diagram, this suggests that Cu should be a very promising solvent for growth of borides. A number of ternary superconducting borides have in fact been grown from Cu. A problem. however. is that B seems to dissolve into molten Cu very slowly. This means that a long soak time at the maximum running temperature is required. In addition, if the growth charge consists of the elements of the compound and Cu, it can happen that the elements other than B will dissolve in Cu first. and then react out on the B, possibly making further solution of B dimcul!. This com be avoided by pre-reacting the compound and using it in the charge. rather than the clements. It is al ..o sometimes useful to make ilctu,,1 An alternative to precipitation of crystals by slow cooling is to supersaturate the flux solution via cVilporation of the lIux. This clearly requires the fl'ux to be the most volatile species present. We hiwe grown crystals of UPt l from Bi by such a method. The disildvantage associated with Ihis method is that the supersaturation occurs at the evaporation surface, and it is difficult to prevent the formation of a large quantity of nuclei. In the case of Bi flux, it can be quite useful for high melting point materials which might be diflicult to separate from Oi chemically although the centrifuge separation method is usually effective. 4.3. Travelillg sO/l'em methods The use of molten metal solvents in the traveling solvent technique is clearly possible. The general aspects of the technique are discussed by Wolff and Mlavsky (1974) und Cilll easily adapl to the various "ulles discussed here. This method is especially impurtant when large crystOils arc desired. Z . FISK iln..! J.I'. REMEIKA GIWW'III OF SI~G!.L 4.4 . Mis("dltllleOllS CHYSTALS nWM MOLTEN METAl. I'l.UXES Flu\ InShe .. ) We have m:lde little mention of the uses of seeding melts. or of providing deliberate nucleation sites. These would clearly be useful in m~lny situations. For the most part we have discussed only the simplest procedures. but it is clear that almost ::11 of the lechnology adapted to produce laige single crystals of oxides and similar materials can be suitably adapted to the growth oC crystals from molten Illl!tal t\uxes. This includes. in addition. the pulling of crystals (rom such met "I lIuxes. There is no need to repeat in our context the detail discussions of such methods as discussed. for example. in Elwell and Scheel (1975). The generalizations to this case are ea~ily made. In «('u.s,. S,i.' Ge(aJ Itlll .Iel 1'1> IU'I,. \'bPI. (jdSb(al Sh ZIlSiP,(~). C"'SiP:(iI) HSi, Sn RCu,Si, 1{.lth,Sn,. 1<1-'" ,I',,(b) ZnSnl',(a) GaSb(a). GaP(a) ZnSil'.(a). C,",SiP,(a) Appendix I RSn.(~) We present below a short list of representative intermetallic compounds grown from various solvents. Bin"ry compounds grown Crom one of their constituent elements have not been induded: For example. these include USn) from Sn. UGa) from Ga, Celn) from In and Ti8t!rfrom the middle of the Ti-De phase diagram. In our list, R means rare earth. although the rare eanh compounds referred to cannot always be grown for all the rare earths. Au" C,ysl~I' AI TiB , • ZIB, RAIB. RB" UB •. NpU,. AmB, RD. RBe". UBe". ThB",,, PuB"" Si(a). Gr:(a) Bi Zn . luSb( .. ). G~Sb(a). S1131. Ge( .. ) InA~(a) ·1{~I.:r~n~e>; (iI) Luzhn")a (l'.IbI!); (bl Mci>nr:, (IYII2); (e) Klr:lowski el .. I. (1'.1115). Appendix 2 Containers for \'arious llIulten metals. A more complete list cOIn be found in Drown and We~lbrouk (11)67). TIle free energy of form'lIion • .:lG. ill kcall mole 0) for various oxidr: crucible materials is shown in fig. 61 aflcr Reed 19711. Tht: values ure for reaction with one mole of O~ . UPt,. NpPt. PIMnSb. Ni~lnSb UAI, COnl~iner UIr, " GaP(a) ZnSil':. CdSiP,(a) Cu RRh,B, RCu,Si, V,Si Rlr, Ulr, Fo: Ga BPla) ZnS(a) Si(a). Ge(a) GaSb(a) ill"iI" IIIclal> Ta. al!.alinc callh lIIelal> AI. G~ Mg In Tit. Il'~phlle fur Ua, slcel AI,O,. MgO. Dca MgO. Ta. glaphile ur ~Ieel AI,O, graphile. MilO. AI,O,. n AI,O" 1.rO" ThO, AI,O, AI,O .. Ta laic callh. Til. Mol. W. lleO Mu Cu. Ag. Au Fr:. Cu. Ni Zu.C.... 1I1l Site:! 11,. Su AI,O,. SiO,. graphile !ib Sia"~ graphile 1"0 ",r /./,\...... 1/"",lh,,"1.; ,m ",,1 CIo,.",m'.1 ul/(,,'r ,.,li'r,ll>.1 X.A . (i,d,,,r,,I,, ..,. J" "tld L , t::P'''1I CO £I'r";r' S,'i..", '.. "IIM"/,,',, n ,I'.. 11I1i1l Z. FISK all.! J .P. REMEIKA £"",,.. Vul. I.! Melling poinb of m.i.!e materials'. Mellillt: ruin! (Oe) AI,O, 2015 11.:0 2SOO MgO SiD, ThO: 21lOO Clwpler 82 PHYSICAL l)ltOPERTIES OF RzFe'48.BASED ALLOYS 172(1 (suhcn~ -1;(00) 331l1l E.BURZO ·SiO l • AI,O •. a ...J DcO have t:",l<Ithcrmal ,h.x:k rl!)istam·c. ThO , ane! MgO intermedlatr. Ihl!rmal shock re)i)lilnce:. ilPe! Y10, ane! ZrO, poor the:rm:.1 shock re:sislance. -&O~--------~----------,------- __ Cemrll[ [IISli/lllt' 1'0 Box MG·(J6, Bucharesl. Romallia H.R. KIRCHMAYR Insrillllt lor £xperimtllfai Physics. Technical University Vielllla Wiedller H{/l({Jll'Ir. 8, A·JO.JO Vienlla. Alls,ria ~ -120 ~ 0/ Physics. Contents -160 o 1000 2000 3000 TEMPERATURE (K) 1. In!roduelion 2. Phase Ifiagrams and cryual SlruelUres 2.1. Phase diagrams 2.2. Cryslal suuclurc: of R,Fe .. B .:ompounds 2.3. Lalliee p"rame:lets of R,Fc: .. 8 compoun\/) 2.4. Preparalion of Ihe alloys 3. Magnelic properties of R,Fe .. D com· pounds 3.1. The Curie: h:mpcr;lIure 3.2. Magnetic mORlenlS 72 74 3.3. Anisouopy fields 3.4. Pr.r4lmagnetic behaviour 3.5. Magnelic properties or R,Fe"B hydrides 4. Rare·eanh-iron-boron magnels 4.1. Sinlered R-Fe-B magnels 4.2. Magnels obtained by splal· cooling " .3. The slability of R-Fe-B perma· nenl magnC:ls References 74 76 78 110 82 82 91 103 106 107 107 120 125 126 811 Fig. b. F,ee energy of formation of oxides. for reaclion wilh one mole: or 0,. Symbols l~ererer.ces Drown. A .• and J .H. Weslbrook, 1\167. in: IntermCI"lIic Compounds, cd. J.H . Weslbrook (Wiley, New York) p. 303. ElliulI, R.P. • 1%5. Con5lilUtion of Binar), AI· : luys. Firs\ Supplement (McGr;lw·HiII, New York). Elwell, D .• and H.J . Scheel, 1975. CrYSlal Gro"lh from High.Temperalure Solulions (Academic PreS5, New York) . Hansen. M . • 19~8, Constitulion or Binary AI· loys (McGnw·HiII. New York). Kletowski. Z .• N. Iliew, Z. Heukie and D. SlaliJ'ski. 1985, J. Less·Common Mel. 110. . 2)5. Luzhnaya. N.P.• 1968. J. Cr~l. Growlh l.U. 97. , Meisner, G.P., 1982. Ph.D. Thesis. University of California. San Diego (unpublished). Morfall, W.G .• 1981. Handbook of Dinary Phase Diagrams (Gencral Elecltic). Reed. T.B .• 1971. Free Energy of Formalion of Binary Compounds (MIT PreS5. Cambridge. MA). Shunk. F.A .. 1969, Conslilulion of Binary AI· loys. Second Supplemenl (McGraw·HiII. New York). Wolff. G.A .. and A.!. Mlavsky. 1974, in: Crys· tal Growlh. Vol. I. ed. C.H.L. Goodman (Plenum Press. New York) p. 193. B. B, n~. e e" 8; Co. C U' c•• ti, d,..,. d. E f ... g, g", ",mancnl indueliun )aluralion inJuclion "~'>Ial tic'" p:uamelcrs Curic .:onstant CUlic COn)lanl of Fe iOIl Curic conSlan! of R ion slifhless conSlimts critical diameler dislance between i,..11 ions grain diameler Young's modulus ani!>Ouopy energy spectroscopic splilling 'X 'Xu, 1/ Ji .. Il,ullll .11, ,.. II, II" 1/.. II.. J •• (i, j .. n, Fe) X,. K, K,. K" KI f~ctur M. spcClroscopic splillin~ factur of iron "',n M •• 71 one ion h.. miltonian cry)lalline .:icClric field hamiltonian external lic:ld anisotropy field exchange field coercive ficld intrinsic coercive field hypcrfinc field nUdC::llion fidd molecular field molecular field codlkienlS ani~o\topy constanlS saluralion magnelizalion efreclive moment iron momenl