Electron-microscopic and optical studies on the electrocrystallization of alpha-alumina (corundum) films on aluminium

ELECTRON-MICROSCOPIC AND OPTICAL STUDIES ON THE ELECTROCRYSTALLIZATION OF ALPHA-ALUMINA rCORUNDUM, FILMS ON ALUMINIUM* S. TAJIM.A.

Y. TANABE.

Tob)n Cii! .4bstract-EEe~tn,n-nli~rascopic

studies

RI. SHIML~RJ. and T.

on the electro~r\.t~llizntion

ol’ alpha-alumina

cnlly l’ormed tn bi-,ulphate melts hnve been made uytng Super-R:tffincrl mechanism

ol’ formcltton

niicrogr;tphs

ii diied

toyther

<)I’ the substrate

metal.

under \arioui

made \vlth the c~~n\ention~l wlphuric Oxide

tilnv+ unodicnll)

produicd

formed

by eleciri< nrcing.

Cr!,tals

grow directI\

Rel‘ructi\c mdes mewnw&it is immediiltelt

tr~nil;vmed

clectrlc urclng.

TIw

non-eyt,tence melt-oxide tionj

interl’xe

~barrier

plus

wI\enf

nclian.

por,w

2~1d ;Inadlztng

in I’uwd WII

itudie5

3nodi-

tilms.

Electron and on

31~~ comp~rwn

URIS

tilms.

s!,trm<

Ire

31’ all-barrier

and no huund~r\ tilm,

ol .fwn

,411 the\e fxt~

\iw

m;1! hold Ir

I.t\er,

t!pr.

hlacro-pore<

are ohser\rd

\ho\\btn~f nmo;phou>

and tincrllv to nlplln-aluniln;l

double rel.r.tctIcw. b\ migr.ltion

the o\idiltion

tvpe oxide anod~c~ll~ ftvmcd

o\idr

are

111’.\I’-

meihaniwi in yue‘ou~

hrt\\een them.

is tirst lilrrned.

\vhtch

hv local heating due it>

i\ not ~~bw-~r~.

suggest that tlw o\i&titwi

under the ilnod~s potrntlal. I;l\rr) .

clt’the wide

or elcctrol\si>;

(>I’the stripped

phenonwnon

The

tilms

per c’ent .Al). and the

and bk>tt,lm 01’the a\ide tilms.

on lrluminiuni

to gnnimn-aluniina

of micropow.

of the oxide.

condiri<~ni

139.399

penetrxtin s the a\ide to the vb>trJte metal. I~IIUI pares xc smclll and tilled up \vith WM o\idr and he3led: the p\ve itze becomes larger \rlth

man\’ but they are _ gxdu.~llv

time.

\\ith -+ame optx;lI

hu\e been tnhen on the lop iurl’ans

the surbce

~~ORI

TcrCyo. J;lpcln

l’nitcnily.

wtctwn

\rhich mdicatc~ 1115 oc‘c‘urs at or near the

Ion5 through

Idttice imperfec-

01’ the bawler t\pe nr duplex cIectrvI\t~:i

!titli

tw \vithout

RksumC-Le me<3nisme de 12 formntioli anodique de AI,O, I >ur .\I W.YYY en milieu hisulfate rc>ndu es1 discute aptis wimens uptiques el niicroscopi~~~-~lectr~~niquej des tilms ~~btrnu~ J3ns dikerws ionditions d’&ctrul\se. Ces tilms wnt 1~x1s du type txirrierc: de5 nixro-pores son1 dtis a dcs arcs atteignnnt Ie mit3l 3 tr;ivers I’uu!de:: Ie3 pares initi;lu\. perils et n~~mbrcu\. wnt prc’gressi\ement combI& p.ir In t+.lctioii Jnodique. mai> Ies pores nou\z;,t,\ grdndiswiit en function du temps. Les cristau\ d’;ilumine se d2!elapprnt dirrctrnwnt sur Ie m&l. wn> cauches intermedi.iires. L’indice de rtil’raaion des tilms detach& accuse Ieur c.trxt2re tout d.;ltwrd amorphe: ils ?\oluent \erb AI,OJ 18. puis AlJO, Y, a In fan\eur de, Pchztutiements IJS~LIN qur pro\oqurni Ia 3~~s. Ce:s I’nits sugg+rcnt qus 1.1 migration des ion5 .413- d:lns les d&luts r~ticul3iresdc:I’c~\~dt‘p~~tIrr;111 ieulecxpliqucrI’e~3ississement. Zurammenfassung-Dir Elehtrokri~t3llis~tion van r-4l,O:,-Drck~chi~hten. dir anodissh in Bisulkttsshmelzen gehildet \iur&n. ist untcr \‘er\mrsh~niimus \vird in Zuiammenhnnc mit einizen optischrn Studien der Ox\dwhichten dishutiert. Die .\ul’nahmen \rurden hei \erschie&nen Ele~trol~rrhcdtn~un~sn an Jer 6hcrtl8chr. nn Querschiiitten und am untercn Ende der O\\dxhicht& SONIC .In der Ohertldche drs B;t>ismetJll, gemxht. Die Ergebniwr: \\ur&n niit denjeniyen kerglichen. \\elliaure erhnlten \trrden. Die in Salzsshmelren Cehild&n anadisshen O\)dr~hi~hlrn bind alle \am Sperrschichtt)pus. hlnkrLyxven cntitehzn durch Lic‘htba~elibildun~. und durchdrineen ~JS Oxbd bis zum Basismetnll. Die uryrti~~$iche~~ Ptwn Gnd hIti und z.lhlrelch. \%erden jber suhz&\; durch d;1s new O\\J nu~‘~el’ullt und heilen .iu\: di? Pnrenzrdw nimnlt nail der Jeit LLI. Die KristJlle \\xhsen unmittelb.lr nut’-dem r\luminium. und es \\erden kelne Grrnrichi Brrchuli”,GiiJew\ an ~~h@dsten Schichten zeicen. d;ljs runiichjt .Imorphrs O\\d fybildrt \\ird. &I, sxh sul’lee loh.ilrr’Erbitrung durch Lisllth,,~~lthilduIlg. Dw Erwheinung der “Forlil DoppLxlbrechung” u rrd nicht bwtxchtet: die5 * Prewnted

nt the I?rll

nwetins

of CITCE.

Bru.wli. 127

April

1461.

I28

S.

TMIMA.

I’.

T.\~\BE.

RI. SHIW~R\ ;Ind T.

~~ORI

INl-RODCJCTION IN

A pre\ioUS

aluminium

paper’.

it was o\idatinn

by anodic

reported that 3lph34luniiw in low temperature melt\

film can be formed on of bisulphate. The tilm is

also formed In some other fused salt swems such ;I> nitrate. The ~IWI~~IWII~H interesting Gncr alpha-alumina is said to be formed onlv ;I~WY IO(H) C. ivhile melting

point

of aluminium is 660 ‘C. and electrathermic

This

of electrochrmicnl surface

nnd is \\orthy

of further

is a topical

reaction>

in\e\tigation.

at

ewmple

Iaxlized

Thus.

the tilm i> porous. can be made >mall amount of ~3liini3-3lt1iiiiii;i. current

ij \‘erF similar

the pwinus

paper.

to that

it heemed

thicker. and

of the combination

tire33

of the electrode

electron-microscopic

optical studie< \\ere made for the tilms formed under \;lri~jus and comparison \\a> made \\.ith the conventional wlphuric

cnnditionj

acid

of oxalic-acid

o\idc

tilm5. as reported the mechanijlll of formation

and

IaFer).

manner. o\id,tion

through

.4s to the site’ at I\ hicll the o\iJalliol; \ ir\r3 ( I) ut metal’o~idr iiiterf.icr:

(21 nl or near

ions and

ths o\idc.

faces bv mutti;

The ne\i information 31~0 of ;~luminium ill electrolvte

of ,\I”+

ions ;rikl O”-

suggested

the reaction

recentI\

movement

of their

that

lar_cer 4ze

nwtnl;o.\ide

the

of O”-

of o\\ueii ._

been

atoms

or

of nluminium ions occuri at both inter-

.Anderson.l

at the mrtal/o\ide tlimugli

ha\e

nuide

SeeIlls to be of the

interface. i3 dificult

Hoar5 becauje

sugcpr>t that OH\\ here it decon1poa

and

intrrf;tce

ion.,

ion<.

wide fiIlNS 011 tbarrkr plus

occurs.~th~re

interfax bv. miprution h ami Berg” state that oxidation

same \.;e~.. u bile Schenh stated

;IC‘IU;III\

bv iiii~ratioii

the wlutioii’oxidr

Gunthrr~chulze

nii~txtioii

in ot

formed in aqueous that alpha-alumina

I1 i> kilo\\ n that tlierc e\ijt t~\\‘o l~pei iii the $ro\i tli foriii of anodic aluminium in ;qurous el~c‘trol~tr~. that is. barrier-t\pe nnd dupk\-tvpc contradictor!

Sillce

tilms.

am.1 5ulpliuric-xid

th;lt the jtruc’ture

formation comes about in ;i quite diNerent contributes to the general theorv of anndic \villi or \\ithoiit x)l\eiit action.

porous

and some ofelectrolv4s

nnd i:. c~~n~powd of ;Ilpha-aluminn and n since the \oltnre’tims cur\e at constant

;Ilphn-aluntina film is similar IO that: of porous tape oxide tilw electrolyte \\ith wl\ent action. But the present rejs;lrch indicates

t\\o

is

the

buck to the w~lutionio\ide interface. occurs 31 c>r nar the wlution.‘o\ide

ion \\ill migrate throu!zh wide twvnrd the into O”- a11d H.- the ilobile H+ ion goillp Let\\ is ;Iiid PIunit+ ~oii~lude th3t oxidation interface Iw the mi~ratioii of .\I”’ ion> from

radioncli\ itv me;i~urenient of single or combiwd tnrtmte :inii 5ulphuric Ad till115 tsulplluric acid bring m;lrl,ed by radioxti\e !?‘“I. This is c’cGn&ient \\ith Vermilyea’> of anodis formation 111’ tant;rluni wide filni5. hIor& er. \ ie\\: of the mech;lilijm the solution, oxide iiiterface I iw cd11 31%) e\plaili \\t’ll the niecli;iiii*iii 0r f0rmntion

of the ~rw\lh of duple\-laver-t\pe oxide films. in which the iiiech;~ni5in oxide tilm is-e\pi:rinsd bv the n\id:rtion t-action ;\t thti plrre-hnje ;1rcta. ELECTRON-hliCROSCOPIC

OBSER\

E.vptv~inrtw/o/. ~luminium cent. \arious

AIXGI.

2.5

conditions

used for r\perilnenti 2.0 ciii2. 0-S iiiiii thick. .\ii~~di~

\rith

a titanium

c:~th~de

;IS sl~o\\n

of porous

.\TlONS

\\a\ Super-RaHindI fY9.999 per o\idntioii ~35 carried oul under iI1 Table

I.

Elrctron-111i~roscapic

and uptial

studies on the clcctroc~st3lli23til,n

or cllph;l-aluminn

129

The voltage rises steeply in the initial stages (a few min)-barrier layer formationthen the film breaks down and electric arcing appears on the anode surface: it mo\.es from one place to another on the whole surface,frcquently accompanied by a humming sound.

With the breakdown

of the film, the rate of voltage

rise slo\vs down.

Thrre-

fore. the voltage/time cur\‘e at constant current is \‘ery similar to that of porous layer formation in aqueous electrolytes \vith solvent action. but its nature is quite different in this case.

obser\ntion. Prepararion oJ’ rtylicas. For electron-microscopic prepared from the following parts of the anodized nluminium: I. Top surface of the o.tide film. 2. Cross-section of oxide and substrate metal (cutting angle 4S‘C). 3. Bottom of the oxide film. 4. Surface of the substrate The cross-section conditions:-

metal (counterpart

\vas, after mechanical

of the o
polishing.

replicas

were

bottom).

electropolished

under follo\\ing

Electrolyte,

Perchloric acid (d = I-61) 75 CU? Acetic acid anhydride I25 cm3 1 V. 3-10°C 30 sec. Electropolishing

wealed

clearlv

the undistorted

metal/o\ide

boundary.

due to

the rather insoluble nature of corundum film in perchloris-acid-acetic-anhydride bath, which cannot be expected from amorphous or ~nmma oxide films. In order to prepare the replicas of metal/aside interface. specimens were cut to appropriate size and then the tilm was stripped in phosphoric-chromic acid mixture at 80°C. hlacro-pores in the film made possible the stripping of the film. The replicas \vere prepared by the pamthn-coated Sump two-step method. proposed by one of the authors 8. The outline of the method is as follo\vs: A tirst replica is prepared from the surface of the specimen by the use of nitrocellulose film and then a carbon film is e\apornted on it. followed by chromiumshadowing

at an angle of 35”. The second replica is then protected a11c1 strengthened the nitrocellulose is dissol\,ed in acetone and the paraffin at its back with paraffin: is removed by heating. This second replica is cleaned se\erul times in acetone and

I!0

S. TXJIMA. 1’. TW~BE. hl.

SHIMLIR~ and T.

benzene rcspocti\ely and mounted on n copper sheet meA

h10~i

ready for electron-micro-

bcopic observation. The shado\\ed carbon replica thus prepared forms a ne_entive of the orisinnl surface. Rtwrl/s. The plates illustrate the be\eral expcrimems in Table I. as follows: Fi.7. 1. Surfxe

of wide

tilm anodically

formed in IO per cent sulphuric acid.

The

black spots of about X0-300 .A diameter dispcrrsed in the \\ hole mntril indicate the pow in the tilm (duplex type). the a\erase number of pores beins 173 IO”/cm”. llearl> identical \\ith the results of Keller t’/ t~‘.~ by electron-microscope. e/ crl.‘” bv oas 3dwrption and of Rummel.lL b:; microscopic study. . L

of Burivell

F(q. _‘. Surlhcc of otide film ;lnodically formed in bisulphate ~nrlt. The chnracteristics ;Irc’ quite diKerent from those of the tilm in Fig. I. There e&t IarFer pores (about 2.300-17.000 A diameter) produced by arcing. The number of micro-pores produced by sol\‘ent action as swn in films formed in aqueous electrolytes is \ev much reduced. but they are noted loyally in ;I slight amount. Fi$p. 3-b. Challpe of‘ the wide surface \\ith time of tjrm;lrion in sodium bijulphate melt. It is MXII that the pore size produced by .~rcins is smaller \\ hen the time ofclectrolysi?; i> shorter.

For clampIe.

anodizing (Fis. 3).

tlw pore diilneter

is about 650-t 200 A after 2 min

.As the time becomes longer. the pore six

euanlple. tlw pore dinmeter is about I AM-X.SnU

becomes larger.

ti after 32 min anodizing iFis.

For 6).

\vhile lhr number of pores has become smaller. Thw pores are produced bv a different mech;lnism fram th;lt of;lqtwous rlrctrolvte snodizilre. Tlw sites of pores produced by shortsr ;II~ Ion_per electrolyGs are not the snm:. Pores ;lre produced ;It defects of the otide film and older pores are tilled up \vith WV. aside and healed. \vhile new pores are bein? continuously formed. Thus, aluminium ionr from the suhstratc metal miprate through oxide :IIIJ oxidation reaction actually occurs ;it or wrne\\ here nc;ir the melt/o\ide

interface.

F(y. 7. Cross-section (cutting angle 15 ) of the wide ami the underlying aluminium. Olide cr>stnls are grwving directly from aluminium without an_v boundary layers. The white part clearly indicates the typical f.c.c. structure of uluminium. The dark scale-like pattern’ are asides. F(y. 8.

Oxide part of Fi:. 7. The oxide tilm is observed as 311 aggregate of single

cryzt:lls ol’about 4.3&15.OOO A. The pores d~x to solvent action in Fig. I are warcely seen. Pcwrj due to arcills are now henlcd and L:CIIIIIO~be detected. Therefore. the oxide from the facts tilrns anodically produced in biwlphate mtlt~ are cowidrred. obser\sd in Figs. Z-6. as all-bfirrier-tvpe in structure. Fig. Y. Boundary

are;\ betlveen oxide and ;Iluminium under the same conditions ;ls that in Fig. 7. The upper part is aside. No particular layer can be a btirrwr-type wide is grooving directly upon seen bet\veen oxide and nluminium: ~hr >ub,trate metal. The oxide part is attacked 10 some de~rce bv electropoli~hing. The surFace of the crystals :lre more rounded 311d smoother tha;l those in Fig. 8. of formation

Fig. I//.

Bottom

of the oside

tilm formed

under the same condition

as that in

Fig. 7. A large pore produced hv arcing is xen: its size i> larger than those OF the initi;ll pores in Fizs. 1-5. The rounded and rr,ottled structures to be seen in the matrix arc considered to he due to single crystals \i ihich constitute the oside tilm and which are partly dissol\rd in a phosphoric-chrolnic acid stripping solulion.

Electron-misrl~scopic

and optical

studies on the ele~trocr~stalliration

ol’alphn-alumina

I31

Fig. II. Surface of the substrate metal (counterpart of the bottom of the o
film the

basis metal. The square pores in the aluminium nwtriu are etch pits produced by the phosphoric-~~hromic acid stripping solution’“. each side beins about I .SOO-3.000 .A. OPTIC.\L

The chanse

of refractive

indes

OBSER\

.-\TIONS

of the oxide

film

with

time

of formation

vvas

measured for the tilms stripped in phosphoric-chromic acid misture by the immersion method usins methvlene iodide. sulphur and alpha-bromonaphthalene. Alphaalumina behaves optically as uniaxial nqative crystal, doubly-refracting (CV = l-768. E = 1.700). while gamma-alumina is an equiauitil. isotropic crystal. rrfracti\e indes 1.69-1.71. III the present measurement of refracti\.e inde\: of the films. the crystal grains of the oxide tilms \vere so tine that only the mean values could be measured for the specimens

of shorter

anodiriny.

but for thicker

iilms,

the mean. ma~imurn

(Nr’)

and

minimum values were observed at the same time. But the values off,, and P of alphaalumina could not be measured. as the speL‘inwns vvere mi\turrs ofalphaand sammaaluminss as shown in Table 2. It is known th;rt the anodic ouidc films formed in ayueous electrolytes ivith solv’ent action are amorphous. or fine cgrain _eamma-alumina. and sho\v. III spite of their isotropic nature. double refraction’“: the tilms are microporous and regarded as a mixed body consisting! of parallel cylinders of an isotropic material penetrating throu_ph the whole thickness of another isotropic material. The phenomenon is called The present bisulphate melt tilm sho\vs. in spite of the ‘form double refraction’. porous

nature.

no form double

refraction.

as the pores produced

by arcing are larger

than the vvavelengh of li_ght. This indicates that the tilm is of barrier type. NnHSO, u/rt/ KHSO, hcrrlrs. Of the three bisulphnte melts. the sodrum salt gave the best crystallinity of alpha-alumina and the specimen formed for 300 min showed a certain decgree of brightness (some:\r.hat lwver retardation) under crossed nicols due to double refraction by alpha-alumina formation. The t\vo hnths produce a similar tendency in the refracti\,e indes of the film: the mean value increases from

I.7 IO toabout

allv approaches the \aIue of alpha-alumina. m~limum values approach those of alpha-

1.7-M in 60 min formation and then gradu.\t the same time. the maximum and and gamma-alumina

reason \v hv the mean value is nOt constant hut approaches probably that. vvhen the time of formation ib short. the prim alumina are smaller and mutually hinder the measurement grains

grow

with

time.

and the masimum

and minimum

respectively.

The

that of alpha-alumina is of alpha- and gnmmaof correct values. The values approach

the true

values of alpha- and _eanima-alumina. NH,HSO, ha//r. It is to be noted that during 4ectrolvsis in ammonium bisulphate melts. arcinS is less frequent than in the other baths. The refractive indices of the films are lower than those from the sodium and potassium bisulphnte baths. The According to microscopic value gradually approaches that of gamma-alumina. observations in polarized li_eht. the film is sli$tly doubly-refracting. regardless of the time of formation. gamma-alumina. amorphous

Alpha-alumina is here in a cryptocrystalline state in a matrix of The minimum refractive inde, is l-667. due to the existence of The coatins ratio of amrn,~nium bisulphate films is generally alumina.

MIHSO,

I

I

1635

.\‘dni:

I+395 I,hYO

I

1.711

1 ~

I .;.I’)

I .715

I.740

I.752

I .;‘h

I +J’) I

1.6Y5

I

I~‘10

(

I.755

I.716

I.731

’

I.722

I .740

I%R?

I ,7!7

I .hh

I

1,740

I .7-s I.752

i-2;

1

1:::;

I.690

I ,fl!4 1 I .7Oh -__.

NH,HSO, I A,‘dni!

I .7x3 ~__

KHSO,

I

I .75?

I .717

l6Yl _~_ I -\‘dni’

I.730

1

;::j;l,

1,719

I .7bh

I.750

I .76h

I ,72s

I.740

I.710

I

I.710 I.71 I

1 I

1.70’,

I l.h77 H,SO

_~ I

, wnc‘.

I A;dm:

_____

I j

I.710

_ I +#7 I.715

h0

I

I

A\:dm’

* hluch

I+P-J

I

H :SO, plu> Okum

I

-_I 60

I.700

11~s uvsullint‘

loiver (about

than DC

tilms. 30 min. according

to X-rat

un;llb&. .

1.4) than those for the other two timj

(about

1.80). Mhich indicates the

possibility of dissolution of amorphous or gamma-alumina in bisulphate melts. In the case of AC film formation. the mean rcfractite indes is I.728 after60 min and the crystallinity is inferior as compared with DC tilms after 30 min. These results were confirmed bv ,X-ra! analysis. S+larric nvin( wnc.) or~tl s~rlplrwic aciA p/w cAw~rr hark. The oxide films formed in sulphuric acid (cow.) and sulphuric acid plus olrum baths gave refractive indices of I.687 and I.715 respectively (Table 2). though the appearance and chemical properties are \‘eT similar to the bisulphatr films as reported in the previous paper.L The oaides formed in cont. acid are amorphouj. \\hilr those formed in the acid plus In both cases. the existence of alpha-alumina oleuni are mainly samma-alumina. could not be detected. From these optical observations.

it ca11 be concluded that by anodic oxidation in fused salt baths. amorphous oxide is first formed: it is immediately transformed to gamma-alumina and finall> to alpha-alumina by local heatins due to electric

Electrol~-nfieroseopic and L,ptical ~lud~¢s ,.',n the elect~,,cr,,slJIhza[ion or :alpha-alumina

133

a r c i n g . T h e e , ; i s t e n c e o f i u t e r t t t e d i a t e m o d i f i c a t i o n s su,:h :is z e t a - a l t m ~ i n a I , -- 1-730) is n o t d e t e c t e d . T h e ~ e f a c t s a r e c o i n c i d e n t ~ i t h t h e r e ~ t d t s o r X - r a ) a n a l y s i s in t h e p r e ~ i o u s p a p e r , in w h i c h o n l y t h e i t u c n s i t v o f a l p h a - a l u m i ~ a l i n e s i n c r e a s i n g ~ i t h time or formatio~t was reported. .~e~,o~h,,l~,e,t~,,t.~--The author~ arc griffel'ul [o Messrs. N. Baba. T. Ki',osa~a. '~. Shibata and A. Otaki for their as.,isla~ce in the e\per~ments. Some p:~rls of the experimented ~ork ~er¢ performed w~th the z.~-,istance of the Go~ernmenl Fund for the Promt~tk',~l of Scientific Rc,,e.lrch. Ministry or Fducation. Io which their thanks arc due. REFERENCLS I. 2. 3. 4. 5. 6. 7. S. u. I0. II. 12. 13.

S. T~,.JIMX, M. S,~oa. T. I~l~rl and N. B.~,B~,. El¢ctrorhim..4cta I. 205 t It,~5q). A. G u n r H e r s c ~ t r t z e ,rod H. Be rz. E/et'lrt~lrl~otld~'l'.~ll(,r~ll. Kr.h'n. Berlin. f 1'~37j. S. AM)Ersatz. J. AtTpl. Phls. 15. 477. ( It)47t. M. SC'HtnK. ll'erA sto/]'.4ht,thlht,t alhl aehtt' .41todi~','he O~ u,httio,, p. 609. Franck. Bern (1948). T. P. Hob.r, "Ailodic Beh~l~tour o f Metals" in ~hrh'r, .q~pt'cts o/ Llec,'~,cht',tistrr. Vol. 2, Butler~orlh, Loltd,,n 11958J. J. E. LF~ls and R. C. PLUMB. J. Eh'clro('heJ,. S,r'. 105, 4~5 IlUSSI. D. A. Verr,.L~eL .4('/,t ~h,t. 2, 482 t 1'.-)5-41. Y. T ~NABE. %ipp, OI hill-_oktl. Ga~k. 211, 663 t 1')5(~t. F. KiLLer. M. S. HUrdlER and D. L. R~'~BI~St,r'.,.J. ~/ectrorhe,t. 5),c. 100. 411 (1953). R. L. BLr~ELL. ~,. SML.DSKI and 1". P. M . ~ , J. Ittlet'. Cltt',l. S,)c. 69. 1525 I. ItJ47L 1". Rt.M~Et. Z. Phys. 99. 518 119361. Y. T , , n ~ [ . unpt~blished ~ork. K. HL.B~r. J. Col/ofil Sci. 3, 197 (19481.