151

"I":i.- Hole .-'f Substrates in the Growth of Self-Supporting Thin Films

D. Ramsay

;•..'.: .iru-vnt of Physics, Stanford University, Stanford, California '34305 U.S.A.

Tin • purpose of this report is to give the specific parting agents used in •_::••

;. :vp.ir.il: i.ni of sc 1 f-;-;upporting targets of some 40 elements. I would dlso 1 ik'.• lo

iv.'.\ :ie general ;:onc] usions that could be used as guidelines .and -.vhere ;. •oss: :.]• •

• :iv" tin' Li.ita tliat led to these generalizations.

With .".ill other conditions the same in a vacuum deposition, the choice 'if the

sub:, t rate •..•an makJ the difference between depositing a uniform cohesive film <•>>-

•r-niiin:-; i !i"; no film at. all. For example, when vanadium is evaporated onto a glass

..;lui-' •.-rated wit . a 2500 angstrom (80 iJgms/cm ) layer of potassium iodide, an ov-r.

(•(•ntir.uoii.i film is formed, which can easily be floated off in and is self-

supporting at 500 angstroms (25 ;jgms/cin ). However, if iodide is the? sub-

strate -outing layer, then the deposition is erratic and will not form a continuous

film even if the charge of vanadium is increased by a factor of six.

The crystal structure and lattice constant of the material, which is to be

deposited, must be considered when choosing a substrate.

The formation process of thin films (Fig. 1) begins with the arrival of a

single vapor molecule at the substrate (a). If it condenses then it can either niigr.it o across the surface or re-evaporate (b). In migrating, collision can occur

,i:i:l .-ombi nation (c). With the subsequent loss of energy, stable islands are

f'-n'H'.J ,\t preferred growth points or adsorption sites, in a process termed

iij." 1 • ^jtion (d). With the impinging of more molecules, growth occurs (e) until the a.M-Lis start to coalesce (f). This is an oversimplification of the process, but it

•.-.•ij] serve to illustrate how the substrate directly affects the formation of a

: ; Jin. The condensing molecule must have an adequate number of adsorption sites

(!•'!!!. '?) and these sites must be spaced at an interval which will allow coalesce:,^1 and inhibit re-evaporation. In epitaxial film growth, the matching of the lattice spacing:-; is critical. In pojycrystalline thin films the lattice constant nay not

)..•..- critical but it is the determing factor for both the number of adsorption sites

.uid the distance between discrete sites. The condensing coefficient and adsorption fiiomy of tiie particular compound are also contributing factors.

For instance cadmium films will grow at room temperature on a substrate of

;^inc chloride." Both are hexagonal in structure and have lattice spacinqs of 2.08 and 'J.S2 angstroms, respectively.

The critical area for causing stress and creating crystal defects is at the sul strat.e/f ilm interface. Cobalt, which in the bulk solid is normally hexagonal,

.:!i!"ortwi in part bv the National Science Foundation- 152

D

'i

Fig. 2. Schematic of substrate surface.

can be induced to grow at room temperature in a cubic arrangement on a substrate with cubic crystal structure. Below 20 angstroms the lattice is strained to exactly match the substrate. Above 20 angstroms dislocations are generated to accommodate part of the difference between the cobalt and copper lattices. If a film like this were to be removed from its substrate it would disintegrate from internal stress. Ideally, in order to obtain a uniform film, substrates with the same crystalline structure as the depositing material should be used. Fortunately this is not always absolutely necessary. Many elements can be satisfactorily formed on amorphous parting agents such as teepol and fonwar. About one half of 4 the metallic elements are cubic and so are many of the water soluble salts. For some of the hexagonal elements like titanium, yttrium and thulium, there are hex- agonal salts, in this case . Cobalt, scandium, holmium and 154

!•; is' • :x\\r:v.. um • \ r o\\: wo 1 1 on a J uin I n um oxide, J v,':i i f •"; i is . i" Li.- J in/xaqoridi . Jn pracf: _e , -3 2

i_: !• ,i ; .raisiuiii KUbstr.it.o may be as thic.-i as 2. J * 10 cm (/ nqs/cm ) and is etneu

.- :" I' i y \i\ ;i:iq in s < ••

i • :...- n! y aluminum which has been expGS"d to air ] a:w onouqh for the oxide

cat :•.!•: !o i'yrn on the surface. In this method residua] contamination eould not

: i . :• i.-.it-. ii iy iookinq for q.-iimiia rays from r rot on-induced roacti ons. This C.;YCS

:n -:r:-i-r !i::i;t of :.<. r: ::• m-i/cm" to tlie aJumimi;:! conlent •:;•:' tJie tar(jdt.

:h"ie ai"" exf [ft i'.ir.'- to tji i J a;!proiii.-h. Altliouqii barium rhlorido i:; ort::n-

r : .• ii:: i i •, j i v,T(;rks we11 '! foi" bery] i i tiin w}]i';}i is hexagonal. .^ui -.'X^ la.nation may ;.-' ( in the '•,'•-]"/ snail ijiain si;.:e (dci-'idar structure) or the very iarqe lattici.- onn-

-.;-.u!t rjf iia) i urn chloride^ ('i.V> anqstronis) . Also r. u'.-ianc -;.».:-, which is "ubi.?, con-

'•• :!•••..: v.-r 1! us a roharorct fill-, on aluminun o:dd>', v/:iich i^; hexaqona!.

.\:v \.\:Y fac"Lor in tiie pr^i'.tration of sel f-s'lpi'jort inq films is the tem; -eratar'j

. ••!" 'h- ::.:;..; t rate . If the substrate temperature if; very low, say 76 Kelvin (liquid

:. i !.!"•;:.••..•!.) i oiid,vn ; a t: on is so fast and the afci jnobi ] i tis.'S so low, that they can-

not r.i':i thr; usual positions they occupy in tli> crystal lattice. The filn is

:'•.• r, iinplelely disordered or amorphous. If the temperature is sufficiently hiqn thin i:i'il.,i l ; |-y of t!;e atoms is hiqii and sinq Le crystal films may be qrov.'n. I-Y-r

; ;.":•.'••:•',': til] im> si• 1 f-supp'Ortinq films the p.articular temperature is not as irripor-

''. in t a..; ::iru n t.ai :i i ::q the sarifj temperature durinq t'ne qrov.'tl; of tlie- film. 7 ":" you I] row a i"Ton film on a teopol-ooatea1 qlass slide, suspended by itself,

•:..• fi 1 ni will, shatter around _!U00 anqstroms (50 ..qms/cm"). If it is grown on a

; •:. :,.:_.i., . .chlfride .aitoil q],oss slide; JOB ivin tjet to about 4000 angstroms (100

• :•••••./ :n"; . if i i i :•: qrown on a thin Jayor of boron on a glass slide which lias

: • • :. •. x; . i:-.-.-d to .

':,: •-.'.••.;:; of i.i. • 10 cm (4:lJi aqms/onD. Iri th i s case, what appears to bo an

::, ' ,,n •• "f therm.il strain is not entirely true.

;:. the coins./ of depositinq a film, an increase in the temperature of the

••.•:' ! it'' comes from tiie radiation

! .'..:: !•!•• i.ondensinq atoms. If we i\ilculate the heat rise due to the latter for a

.' ;•• :.'••!•'." :,i l'/rij- fiJm urown on .) qlass slide (Fiy. 5) we find it to lie 1.5°C.

(.: 7 M C T ac; acj a-.: 1 qs •; s g s _ ^ C •- 'A C 'final

a q a1j q yy q.s

ma s s i n cj r ams

i: - silver

s ;. e ••.• i f i <_• heat

t ois;.'r; r a t u re °C

•i 1 ass slide

Fiq. 3 155

,: ! " i ::i.i>: imum increase from 15°C to 16.5°C. It is possible then to qrov: a

i :-i •..-I'iicMt introducing any large temperature changes in the substrate due to

:,.!"iisi]iq atoms.

'.'ifTO are two approaches for avoiding temperature build-up in the substrata

. •:. ; ,1.: i .it . Tile Eirst is to minimize the amount of heat radiated from the

:: • MM!.ion iniurci:. If the time of evaporation is short, or the vaporization

•; "r.itur" low (•' 1200°C) often no special precautions have to be taken at alJ .

:..• i i i ::is liko bismuth and lead are simply grown with the parting agent (in this

'.•••.: i mi! iodide and potassium chloride) on a glass slide suspended by itself.

:: '.-.': M- v.ipori zation temperature is very high (> 1200°C) often the area of

•.r'••:!- radiation can be reduced and effectively shielded or water cooled.

••In:- ; i I:;:; for example, with source temperatures of > 2700°C have been grown 9 : i! •:. slide coated with sodium chloride. The electrostatically focused

• -i :' .i 'iun was water-cooled and the area of exposed radiation was only 1/8 of

'.):•:, in diameter. Since the amount charged was small (~ 30 mgs) for thick i •. •.••V'MviI evaporations were made, but no curling or shattering was found even

i • :.i V.IK'SS of 4!3OO angstroms (1 mg/cm ) .

•':. • s.'ct'iid approach is to compensate for the radiant heat by removing hoat

-:i -..•• subslrat" at a rate which will keep the temperature constant. What is

••I-:.mi. is that tiie initial condensation layer must not occur at a much lower

•:•••!•.i' .:rc (".Ii-Tsi the final condensation. This would lead to excessive thermal

ii.i.-. ;u'ii;ii built into the film. Temperature gradients across a 1 mm thick i ; sli.-'ii. •.•an be rM° to 100°C, so for elements like palladium, scraetimes a

-.' ;!;:.:;t.!,:iv because of its much higher thermal conductivity is put on a

••-••.>oli'.l copper plate. In this case the cesium iodide release agent is

• !•.)•• i just prior to the palladium.

ih-- literature cites several instances where the substrate is pre-heated, 10 • illy to iOi)-400"C. This will increase the mobility of the atoms on th ;

.'• t .it-.- surface, the rate of coalescence and the grain size, but it requires

:;••• a substrate with the same coefficient of thermal expansion as the conden-

,- aifiTi.il. iJtherwise, severe strains are built up in the film when it is

11 •• i down. r'nly in cases where the film will not be removed would this .,e

•••! I.I:<1.-. The two titanium films in Fig. 4 were both grown on glass slides

. r !:n- -;,U:IP conditions but witli different substrate layers. The one marked

•;.•..! ,i f i In of aluminum 6000 angstroms thick (150 Ugms/cm ) deposited on i'.:

• • i i. i; .•!-_,' followed by the titanium of about half that thickness (3000 angstroms).

'. :••. [••>'•: .-nip leti on, tlie film sliattered leaving the pattern shown. No effort

!:-..i'i' to prevent the slide from heating up and it reached a high of ~- 7C°C.

, ,•:•::!•,in' (:1K.1 geometric pattern of fracture with slide (b) . This titanium

::i is- l.!;e same thickness and was deposited at the same evaporation rate, 4000

• -t. r. .p\:-i ;•.•!.- ninute. However, the undercoating is calcium iodide. At the; 156

(a) (b)

Titanium films of similar thickness,

bat different substrates.

(a) a 1 umi. num (b j i ale iiini iod i de

•n: U't ii,n ot the run it way about 70"C and the; film was "intact, but as it

••"'Li] l.o room temperature random cracks appeared. In tho first case the stress in fhe film was due to much more than a difference in coefficients of thermal

:•:: ansif'n. The strain was mainly because of a hexagonal structure growing on .'.

Mi'ic suiistrat!'. In the :_:ecorui case tile random Irearimj of tile film is truly diu

'.i' !;:•• difference in the thermal oxj^ansion, by a factor of two, between tile

:la-;s and titanium. Because it is relatively thin, the tensile stress over-car

';.•• ohr-sive s t i'L'nq ti;. In a much thicker film it uiay have remained intact, but i! removed woulrl curl info .i tiqht cylinder. I recommend erhausting tlie list

••'" different substrate materials, before going to a proccdu' e of growing a thin film on a deliberately heated substrate. Table 1 is a list of the partir-f sub- ti'iiL'cs usixl in produciivj sol f-supporting films for targets in tandem accel-

•ral.or ex;n'rinents at Stanford University. 157

Table 1

Substrates used in the growth of self-supporting films.

E] ernent SubsLrate Element Substrate

Aluminum Teepol Manganese Aluminum Oxide Antimony Cesium Iodide Molybdenum Sodium Chloride Beryllium iJeodymium Barium Iodide Bismuth Cesium Iodide Osmium Sodium Chloride Boron Boron Oxide Palladium Cesium Iodide Cadmium Zinc Chloride Potassium Hexadecylamine b) Calcium Hexadecylamine Praseodymium Aluminum Oxide Carbon Teepol Nickel Coppe r Chromium Potassium Chloride Rhodium Potassium Iodide Cobalt Aluminum Oxide Ruthenium Aluminum Oxide Copper Tcepol Scandium Aluminum Oxide Erbium Aluminum Oxide Silicon Potassium Chloridt Germanium Barium Chloride Silver Teepol Gold Teepol Tellurium Teepol Holmium Calcium Iodide Thulium Calcium Iodide Indium Formvar Tin Teepol Iron Copper Titanium Calcium Iodide L^ad Potassium Chloride Vanadium Potassium Iodide Hexadecylamine Yttrium Calcium Iodide d) Magnesium Sap on Ytterbium Copper

Teepol 610 (sodium secondary alkyl sulphate) Shell Chemical Co. b) S. H. Maxman, Rev. Sex. Instr. 35_, 1572 (1964) Formvar 15/95E (polyvinyl formal) Monsanto Co., St. Louis, Missouri d) J. L. Gallant, Nucl. Instr. Meth. 102, 477 (1972) 158

References

1- K. D. Leaver and B. N. Chapman, Thin Films (Wykeham Publications Ltd., London, 1971) . 2. L. I. Mirkin, Handbook of X-ray Analysis of Polycrystalline Materials. 3. L. 1. Maisel and R. Glang, Handbook of Thin Film Technology (McGraw-Hill Inc., New York, 1970). '4. A. Taylor and B. J. Kagle, Crystal log raphic Data on Metal and Alloy Struc- tures (Dover Publications, New York, 1963). ''>. M. Harchol, Nuc-i. Instr. Meth. 40, 158 (1966). 6. D. N. Braski, Nucl. Instr. Meth. 1£2_, 553 (1972). 7. A. n. F. Muggleton and F. A. Howe, Nucl. Instr. Meth. ^3_, 211 (1961).

H. J. R. Erskine and D. S. Gemmell, Nucl. Instr. Meth. 24_r 397 (1963). J. R. F. Casten, J. S. Greenberg, G. A. Burginyon and D. A. Bromley, Nucl. Instr. Meth. 80_, 296 (1970). 10. S. II. Maxman, Nucl. Instr. Meth. 50_, 53 (1967). 11. G. V. Samsonov, Handbook of the Physicochemical Properties of the Elements (IFI-Plenum Data Corp., New York, 1968).