Biotechnology and Materials Science Cbenr,istt'y fu tlte Future

il[ary L. Good,Editor

Jaqrcline I(. Barton, AssociateEditor Rtdy Bautn, fusistant Editor Ivars Petercon,Assistant Editor t'tt

i;i i ' i

IrlancyHenderson, flssistant Editor

I :

't{ .1

'l

1

":i , . t. ,: 'I.

;.i .aI .t,

i+-| .i I

,,t1.4 1. t I 't ':. 1 r ' r .t ) | r :

,?.t

',d

.

t

.iiiI ;t:i .i. I {;l .'-t".! r l

:fi . ,ii''{ ''-

:

j

'.':i-.1 r i. ..:1

.J'':

. ";' '' r' -.1' :i , .r.t' .,, .1:'

American Chemical Society lVashington,DC 1988

'l

.( t

- .r 1! \r : i

f

-(

. ,;r I,

+J r Fl

I 1 '.1 + b

Ub nrv o f C,on grcss Catalo gin g-in- Publi cation D ata &otechnology and materialsscience:chemistrv for the firrure \larr L. Good, editor, |acquelineK Barton, associate editor...[et al.]. P.

cm.

Includcsindex.

lsBN 0-8412-1472-7 (pbk.) ISBN0-84u-I473-5 I. Biotechnology.2. Materiais. I. Good, Mary L., l93l-

w248.2.P55t2 660'.6-dc19

. II. Barton, ]acquelineK

t988 88-14544 CIP

Copyrighr o 1988 J \

American Chemical Socieq' All Rights Reserved.The appearanceof the codc at the bottom of the first page of each chapter in rhis volume indicatesthe copvright owner's consenr thar reprographic copies of the chapter mav be made for personal or internal use or for the personal or internal use of specific c[ents. This consent is given on the condition, however, that thc copier pav thc statcd per copv fee through the Copyright Clearancc Center, Inc.. 27 Congress Street, Salem, MA 01970, for copyrng beyond that permined by Sections 107 or 108 of the U.S. Cop.vright Law. This consent docs not extcnd to copving or transmissionbv anv means-graphic or electronic-for any other purpose, such as for general distribution, for advertising or promotional purposcs, for creating a ncw collcctive work, for resalc, or for information storage and retricval wstcms. Thc copying fee for cach chapter is indicatcd in thc codc at thc bonom of the first pagc of thc chaptcr.

, ll .

,l

'...., t,.

r.ia.

l', 4. ,.t L-. -a

.t,.L i-..:::

Thc citation of trade names and,/or names of manufacnrrers in this publication is nor to be construcd as an endorsement or as approval by ACS of the commercial products or services refcrencedhcrcin; nor should thc mere refcrencchcrein to anv drawing, specification,chcmical process,or other data be regarded as a licensc or as a conveyanceof any right or pcrmission, to the holdcr, rcadcr, or any othcr pcrson or corporation, to manufacrure,reproduce, usc, or sell any patented invention or copyrighted work that may in an,v way be related thercto. Registercd namcs, tradcmarks, etc., used in this publication, evcn without specific indication thcrcof, are not to be considcrcd unprotectcd by law. PRINTEDIN THE TINITEDSTATES OFAMEzuCA

itl t: I \li ;l

i'

r;i

'. .:

i

, t,

,t..

t;T -..l "

f rr - , r'I

'It ..jl' .'j / .j.

@Ptt"

g.

Materials for Advanced Electronic Devices GeorgeM. Whitesides f,, n,pical intcgratccl-circuitchip looks like an I tarrnorcdinsect.It has .r hlrtl pl.rsticor ceramiccase fionr u'hich an arrav of nrct,rllcgs p"'prvlrude. Inside this arthropodal packaging sits a rcnr.rrkab,lc collcction of n r a t c r i a l st h a t a r e c a r e f u l l vc n s i n c c r e dt o c o n t r o l t h e nrovcnrentof electrons.

I

Tbc Sfitrcttu"eof an Intcqrntcd-Circttit Cbip T h c k c v l v o r k i n g p a r t s o f a s c l r c r i ct h r p . r r c r c a l l v t o o s n r a l lt o s e er v i t h t h e n a k c d e v c ,O n i r r l : c p a c k a g i n gi s r c a d i h 'a p p a r c n t ,b u t u n d c r ; . rs c J n n l n gc l c c i r o n n r i c r o s c o p c '\ ' o u \ \ ' o u l d s e et h a t t h c c h i l ' ' i r s e l t - r sm a d c o f a scriesof thin lavers,w'ith onc nrJtcrill colrcd on top of another (,rrcFigurc I ). Each lavcrirasa thiclurcsstirat mav be anlrvhcrcfronr a fbw'atomsto sevcrf,lthousandatoms d ec p . ,d c p c n d i n go n i t s f u n c t i o n .T h es e l a rc r s . c a r c f u l l v laid onto a chip'ssurfacc,tlicn etchedinto \JSt Jrravsof nricroscopicelcctronicsu.'itcltcs and gates. n ork toqether to shuffle electrons abor-rt.In place oi thc maze of u'ith scparatervircsand componentsn'picallvassociated elcctricalcircuits,thc chip'srviresand electronicdevices are integratedas lines and channclson its surface.Toda/s integrated-circuitchips ma)' carry hundredsof thousands of transistors,each nreasuringas little as a ferv micromctersacross. A qpical chip startsoff as a clean,polishedwafer of silicon doped (deliberatelycontaminated)uith a trace of eithcr boron or phosphorus. Doping with phosphorus which provides elecproduces an n-dopedsemiconductor, trons as current carriers. Boron produces a p-dnped senticonductlr,which provides positive, electron-deficient

a

'

't

. t

r :; . ':l

r

r473-5 / 88/0085$06.0A/ 0

@ 1988AnrcricanChemicalSocien' I

. . i I

l a

t-

86

BlorecHNol-ocy & Marsruars ScrsNcs:cueursrnv FoRTHEFt-n pe

Figurc I. Scnnning clccfi'on microscopic vicw of nn intcgroted-circtir chiq.

regions, called holes, as current carriers.Atop this silicon surface, chip fabricators deposit as manv as a dozen lavers,some of which, like silicon, are semiconductors, while others mav be electrical insuiarorsor conductors. Each rype of material plavs a role in pushing electrical charge through the chip's circuits. The conductor in an integrated-circuitchip ma,vbe a metal, such as aluminum, or silicon that is heavily doped with a conducting material.Strips of conductorsform electricalconnections within and benveencircuit elements.The insulatorusually consists of silicon dioxide and is used ro prorecr the silicon surfaceand separateconducting regionswhere no connectionsare desired. Figure 2 shou's an integratedchip crosssection. In the fabricationof intesratedcircuirs.the laversare

Cu-Snintcrmetallic PhasedCr-Cu Cr

3.8 pm SiO,

2.3 1tmAl4% Cu

Cu 0.85pm Al-4Vo 2.8 pm SiO,

0.15 pm Cr-CrrO,

Figwc 2. Cross section of an integratcd-circuit chip.

9. WHirusrors

Matn'iak frr Ad.vatrced Ebcn'onicDniccs

87

addedone at a time.A templatecalleda "mask" determinesthe patternfor eachlayer.Fabricationtakesplace by variousstepsthat combineoxidation,maskprotection, etching,diffusionor ion implantation,and vapor deposition. Finally,the chipsaresealedin a protectiveplasricor ceramic package.The packageis what many people envisionwhentheythink of a computerchip, but the chip itselfis acruallya flat picceof silicon,no bigger than a fingcrnail. Thcsctiny chips,carryingan increasingly heavyload of transistors and other devices,havefueledthe explosive growth in thc clcctronicsand computerindustries.Every year,nroreand morc cvcrvdayproducts,from toys and tapc rccordcrsto rcfiigcratorsand automobiles,rely in chips. some\\'avon intcgrated-circuit

ConQctitive Choices Nlicroelectronic technologl'has erolved so quicklv and has becomethe focus of such kcen inrernationalconrpetition that policvnrakcrsarc no\\ lett u'ondering whcre thc United Statcssliould focus its eflbrts and what role m a t c r i a l ss c i c n c cs h o u l d p l a ) ' i n t h i s r c s e a r c h O . n the vcrv comnrercial sidc, a hrsc fiaction of the gross nationalproduct has conrc to depcnd. nrore or less directlr',on elcctronics.Nlorc reccntlr',from the perspectivc of narionalsccurin',thcrc is a gron'ing perception and conccnr that manv csscntiajelectronic svstcmsin nrilitarv applicationsusc componcnts that are a','ailable onlv from ]apan. Thc growth of microclectronicsraisesm'o important issues.First is the basic question of how best to trv to defend the existing electronics industry, which relies heavily on silicon-basedtechnology. Silicon itself will remain an important electronicsmaterial for the foreseeable future, especially as researchersachieve smaller feature sizes and three-dimensional strucrLlres. New tcchnologiesbasedon gallium arscnideand the construction of hybrid devicesthat combine the most promising characteristicsof silicon and gallium arsenidewill find their places.Technologiesbascd on diamond or cubic boron nitride are also possiblc but much more distant. Nerv superconductingmaterialswill probably have a role in high-speedcircuit connections.

.,:,:

'!^: -' , .a l' 1

,:

':

;, '-ri

,: ,

i

: ; , .

,.) i

'r'.

; ! l

: . ; i ' i ' , a '

'i1 ,

j.

...:..

l:

rt , . l

t\. .i ..: i'

.i

i.

88

Brorr,cHNolocY & Marr,zuats Sctencr,: CHr,ivllsrR)'FoRTHE Fu-rt'nr

The second issue is to determinc in what directions the technology should be pushed. The electronicsindustry is continuing to move in the direction of miniarurization. Featureson silicon chips are becoming so small that they are easilydwarfed by a human hair, and miniarurization can be carried funher still. To achievethe next level integration, chip dcsignersare venturof ultra-large-scale ing into the realm of three-dimensionalstructures and exploring exotic approachessuch as la,vcredsuperlattices. Nerv techniques of fabrication promise unprecedented densitiesand device geometries.The new generationof chips would have fearuresthat are lessthan a micrometer across;millions of transistors rvould be packed onto a scrapof siliconor some other semiconductormaterial.Bv srudyingthe completerange of propeftiesof electronsin all npes of materials,scientistsprovidc the fundamental knowledgeessentialfor large-volume,reliablcproduction of microelectronicdevices. As chip designersadvanceto the staqeof ultra-large' scaleintegration,thev u'ill trv at the nrolccularlo'el to design the propertiesof chips. Then it u'ill be especialll' important to be ablc to predict thc response of a particular combination of matcrials and a particular circuit design.Chip dcsigncrsrvill also need the abilirv to assessvarious stratcgies fbr grorving thc ntatcrials requiredfor the nricroscopicstrucrurestho' dcsirc.

Packng h r.gSenti con du ctor"s The plastic or ceramicpackagingthat containsa chip's links with the outside world and protects it from damagc is now becoming the bottleneck that restricts efforts to increasethe speedand shrink the sizeof integrated-circuit chips. Although packaging has traditionallv been much less exciting to work on than the chips themselves,new developments in packaging may be the k.y to breakthroughs in chip technolory. Most electronic devices are placed on alumina (aluminum oxide) supports, or substrates.A picce of alumina is made by compressing alumina powder into an appropriate form, and then sintering (heating without melting) it to createa densemicrostructure (raaFigure 3). There are important correlations berweenthe form of the particulate alumina powder and the electrical properties

9. WHrr.stuEs

Dnices Materiak for AdtancedEl^ccn'onic

Figut .J. Sinrcrir4 oJ' nluminn. Lc,ti, ifunl pncl:iiltt; r'igltt. rlrnsc rrticrostrtrcttrcobtnitn'r{nt 1.100"C lin'ottt ltotn'.

89

and dimensionalstabilin'of the final assembledproduct. One important goal is to pcrfect the final product by using the best availablealumina pou'der, such as the highlv regarded matcrial providcd bt' the Japanese compan)'Sumitomo. That kind of pertbction is likell' to tcchnologies. Some ne\\'processing lcad to fundamentally more complicated are also trving to fabricate rcsearchcrs alumina structurcsthat not ottlv support the functions in an integratedcircuit, but also pertornt important functions of their o\\'n.

Atontic Viers a.

To control the materials and the chemical reactions involvedin fabricatingsophisticatedsemiconductorcir' cuits, scientistsneedthe abiliw to detectsurfbcedetailsas small as individual atoms. To achievethis notoriouslv difficult task, chemistsare begintting to adopt techniques developedin the field of vacuum physics.The scanning tunneling microscope enables researchersfor the first time to image atoms directly, one at a time. In a runneling microscope,an extremelysharp, metal needleis brought within a few angstroms of the sample's surface.This distance is small enough for electrons to leak or tunnel across the gap and generate a minute current. As the gap between the tiP and the sample As the probe crossesthe increases,the current decreases. across its surface, its forth and sample, moving back vertical height is continually being adjusted to keep the

l.-

.''. .t;

't i'a,.

:1. ta ; l, tl j:

.'

rJ. :i:. h . fl 'Yrf-

l . Ir,'.

',,) i(:.

!i, '. 1 |

..

- !

.';1.

1( ti

.- ).'-

. .

, < . t !

t"'

-''i,

l:':

'r: | .' f i' t a'.'t

, .\. '.|.t.': t.\r

90

Btore,cHNolocy& futareruarsScrence:CHr,ursrRyFoRrsr Frm;nr

. , . r ..

l.-:r.

i i;i

Lt ", '.. .i' ' liiA l:';, f'{j- .r

current constant.In essence,the probe traces out a contour map of the sample'ssurfaceatoms. The inventionof the scanningtunneling microscope, and subsequentrefinementsin its design have given scientistsincreasinglysharpviewsof atoms perchedon solid surfaces.Most recently,the microscopehas providedpicturesnot only of siliconatomsneatlyarrayedon a siliconsurfacebut alsoof the bondsholding the atoms in place (seeFigure 4). Normally, the voltage applied benveenthe sampleand probe staysthe same.To observe the bondsberweenatoms,scientistsat the IBM Thomas |. WatsonResearchCenter(/) held the probe still while varvingthe voltage.The resultwas a map of how the current varies at selectedpoints over a surface.The information was then used to show'where electrons bondedto surfaceatomswerelikelvto be. It's amazingthat scientistscan make such minute observations and can beginro usedericeslike the probe to pick up indil'idualatomsand movethem ro someother part of the substrate. Such a feat was inconceivable 20 yearsago.

I i " i"' ' ' i . ' l.: .(. l.j-l

li 'i l, :- 't ?. irl .I'. \...lt

h,i l. i l.t'tr, l'^i ;'

Lt..,. -1. a'.-, -:+:. '.

t,

t. .,. t' l . ;,..:!

l. . .. l)
.{ .\? I a l 9 1 . ,

Sincethe earll' 1970s,scientistshave been promoting gallium arsenideas a fasrcr,more efficient substrate materialthan siliconfor makingintegrated.circuit chips. (Figure5 is a scanningrunnelingmicrographof gallium arsenide.) However,rhe vast majoriryof chips are still made from silicon,which is abundantand cheap.The most importantadvantage of gallium arsenideis speed. Electronstravelaboutfive dmesfasterin galliumarsenide than they do in silicon.Gallium arsenidealso hasa high resistance to electricalcurrent beforeit is dopedwith any impurities to form circuit elements.Consequendy,a gallium arsenidewafer, or substrate,is semi-insulating, whereasa silicon wafer is semiconducting. That fearure simplifiesgallium arsenidecircuit fabricationconsiderably. Gallium arsenide also offers a wider range of operating temperaturesthan silicon and much higher radiation hardness,which is a decisiveadvantagefor military and spaceprograms.Another major advantageis that galliurn arsenidecan be doped in such a way that it emits light, which makesit useful for lasersand lightemining diodes.

t-

t

l " a ;t.i

rii- r:

(,....,." l',.,{ |( .. . . , ' - a

i.1,, , !, , . 1

ii':. i'\rr L.. : -..f . t l; ' I :< h - .

)t

' i+i ).). tlla.

L.t '1'l'l"l''

. rt'' l,':i.'

lt; '. .$;.. ,1+l. '. r.!'

. 1,.._ -

! ' f

kl' -

fta F.F

.

lr'..!.'

/ {:,

-

- ' . : { t

t ..:>' i :'1+'

' ' , , L . ) ! . ' c l . $ '

j

i

, i.l.

,!

. :.li ''

I l

.

,.,--!.

1;.'

":

I ,'' -' ' i l c

,;., i I

i),r'

'

i'i'

' '.rr.. .r-

.

._,-il.i:

1 I

a

i-

:' .

r'.' ..;..{. . !:li

-'.: .-(;, ..,:-, ii : J

, , 4

. ,i-:, t-.'i -': " 'YJr:

.

.,

,

.

..

..

,.a.V .,Ii

:'' rYf

.'

\

i: f t...

9. WHrrr,sroes Mata'ials for Advarced Electronic Dnies

1...' !rl t.-! lr l.:' l.r

9l

t'; t.', t.t'

l:1r tr 1..,

l:,'.

lr:,

l:;

l:l F .: f' -' t. l\ :'r l,

t,;

lt ..; : : ' Il ( . , I ..; l . ; i

li.'. t,tr

f.i ' r,.'. f'!

t, i.^

l.i' ; t

i'l

j I

. d

.,;-

Figrrrt 4. 7illo ricn's oJ'tltc silicorrsurlhcc ,,r,onrtr*,;:rt.rurclirn

:.

nticroscopt,. Tlr lann. is n blortrp of tln

'

92

BtomcHNoloc)' & M.qrsruarsSclsNcr,:Cne,ursrRyFoRrrs Fwunr

Figure 5. Scanning tuttncling nia,ogrup h oJ'galliun anenidt.

The problem with gallium arsenide is that thc material is exceptionallv difficult to grow into largc, defcct-freecrystals.Much care is ncededto produce from thc elementsgallium and arsenica verv precisel.v tailored just compound with the right properrics and the right proportions.Large silicon crvstals,on thc other hand, arc relativelv easv to produce, in part bccausc onh' onc elementneedsto be controlled.With gallium arsenidc, rwo materials must behave properlr'. Onc of those materials-arsenic-is toxic and volatilc ar the high temperaruresneedcdto grow crvstals.It tends to bubble out of the high-temperaturemelt. Despite the development of various methods for overcoming theseproblems, high-qualiq'gallium arsenideis still relarivelvexpensive and hard to get. Furthermore, silicon is a better heat conductor, and it allows more transistors and other devicesto be packed into a given surfacearea.

A Matchfo, Silicon. Until now, siliconand galliumarsenidetechnologies have developedsomewhatindependently. One way of dealing with the silicon/ galhumarsenide trade off would be to marry the two types of components.Putting gallium arsenide semiconductor circuitsatop a siliconbaseis a bit like mating a Ferrari with a Honda. The components seemincompatible,but if the match were to work, the result would be an attractivecombinationof high performance andcconomy.

9. WHrrssrogs Iv[aterials for AdvarrcedEbctronic Dnices

93

Hybrid integratcd-circuitchips of gallium arsenide and silicon may now be feasible.Researchersat the Universiry of Illinois at Urbana-Champaign have discovered a way to deposit gallium arscnidelayerson top of silicon wafcrs without spreading the crystal defects that ruin the electronicpropertiesof thc materials. The trick is to find a wav of aligning the silicon and gallium arsenidecrystal laniccs.Normally, the strucnrres do not quite match. For a rorv of 25 silicon atoms, onlv 24 atoms from a gallium arsenidclayer are neededto fill the same space.Aligning thc nvo materials produces a largc numbcr of defcctsw'hcrc thc nvo larticesmeet. The mismatchcan bc overconrcif thc silicon baseis slightly tiltcd. A gcntlc slopc of about 4o provides,at thc atomic lcvcl, tinv stcps that take carc of thc problcm. If thcsc stcpshavcthe right oricntation w'ith rcspcctto the silicon cn'stal latticc,thcn thc inhcrcrrtbumpincssof the slope docs not producc dislocationsthat thrcad their way into thc galliumarscnidclavcr. The oricntationis thc ko'. For a sclllarcsiliconchip u'ith an Llppcr surfhcc prrallcl to a facc of thc crvstal latticc,thc slopc nccdsto risc fionr its lon' point at onc corncr to its pcak at thc diegon"rllvoppositccorncr.Thc conrlrinationof light-crnitrinqerlliurn arscnidcchips and c o m p l c x , t i g h t l v p a c k c d s i l i c o n c i r c u i t sc o u l d m a k c i t p o s s i b l ct o c o n n c c t c i r c u i t s o p t i c a l l vi n s t c a d o f u s i n g u'ircs.In sonrc of todai's nrost aclvanccc'l chips, far nrore po\\'cr alreadi'gocs into drivins thc rvircs that conncct c h i p s t h a n i n r u n n i n g t h c c o r n p l i c a t c ds i l i c o n c i r c u i t s thcmsch'cs. With hvbrid chips,thc ll,iresconnectingonc dn'icc to anothcr could bc rcplaccdbi, an efficient optical svstcnr,pcrhapsusing optical flbcrs. Becauscall partsof an intcgratedcircuit necd not bc cqualll' fast, it mav o'cntualll' bc possiblc to deposit gallium arsenidcat onlv thc points on a silicon circuit u'here the chip must operatcquickly. Rcccnt work at the Universiryof Illinois (2) will probably acccleratethe pace of h,vbrid-chiprcscarch.Continuous lascrs and optical intcrconnectsnral' bc dcvelopcd soon. More and more rcscarchgroups arc activc in the ficld, and severalsmall companics havc bccn cstablishcd to develop the tcchnology. The usc of matcrialsthat rcspond to light suggcsts the potential for a major shift in technology from computing and communications devices based on the

\--

94

BrorscnNoLoc)'& lvl,,rrEzuars ScreNcr:Cur,lusrRyFoRrns Fun;nr

movemcntof electronsto devicesbasedon the transmission of light. Opticalcomputationand communications are not yet a major commercialtechnology,althoughthe communications part is becomingimportant.One major interestof researchers is to developclasses of materials that enableone to manipulatelight wirh the speedsand characteristics that are requiredof new generationsof chip-to-chip and continent-to-continent communicarions. Perhapsfirrurecomputerswill do their computationsby manipulatinglight pulsesinsteadof electrons.There is much opporrunin'for important inventionsand major nervapplications.

Layersuporr.Layers

.:

.... :.' . , i

t. .. r l ' l

,' z r', . {

i

-1: :'r :> '.j :

s

't:.

One interestingn'pe of microelcctronicdevicenow being studiedis a hetcrojunctiondn'icc nradc up of multiple, alternatinqlaversof gallium arscnidcand gallium aluminum arsenidc(srr Figurc 6). Each lavcr is onlv a couple of atoms thick. With such snrall stnlcrures,,a host of remarkablephcnomcnabegin to cnrcrgc.Onc is negative resistivin':When the voltagc is incrcascd,thc currenr goes dorvn rather than up. Another phenomenon evidcnr in exccedinglythin semiconductingfilms is ballistic transport, w'hich allorvs an electronto passfionr one sidc of a barrier to another w i t h o u t s t r i k i n g a n ) ' a t o n r s i n b en v c en , t u n n el i n g through the barrier like a ghost passingthrough a rvall. Under normal conditions, electrons do not race from point to point like speedingbullets;rho' srumblealong more like drunken sailors,Flowing through the circuitn' of a chip, they constandybump into impurities, rebound off walls, and slow down as they pass through the electronicgatesthat signifr' on or off in a microprocessor. Each collision cosrsdistanceand time. Ballistic transistors are designed to be so small that an electron can shoot right through the device with scarcelya single collision. The idea is to make the length of the region that electrons have to travel comparable to the average distancethey go before colliding. Over the last decade,there have been a number of spectacularadvancesin the construction of semiconductor heterostructu.reswith a specified band gap. These have led to the discovery of the totally unexpected rwo-

Fiqurc 6. Hctcttlurrcnlil stt'ucturc.

9. WHrresroEs lvlatcrinkfrr AdvancedElec*onicDniccs

95

dimensional belravior of electrons, as well as ro the development of novel electronic and oprical devices. Manv of the advancesare a direct result of the develop. ment of ncw crystal growth rcchniques allowing the formation of laycred semiconducrorsthat are perfect on the atomic scalc. A number of technologiesmake possible these astonishing strucrLlres.One is metallo.organic.vaporphase deposition, also called vapor-phaseepitaxy. Ir involves taking appropriarclv prepared organometallic compounds and allorving thenr ro react in the vapor phascto dcposit rhc desircdinorganic strucrureson an appropriatelyprcparedsubstratc.Epitaxyis the processof grow'ing crvstalline senriconducrorfilms in rvhich the substratcdetcrminesrhc crvsrallinin'and orientation of thc thin filnrs gro\\'lt on rop of rhc substrate.Among thc tcchniqucsthat havc bccn rr idclv uscd arc liquid.phase cpitaxr'.,chcnrical-\'apor dcposition.,and molecular-beam epitaxr'. I n l i q u i d - p h r s cc p i t a r r ' .t h c c p i r a x i . rlla v e ri s g r o n ' n b v c o o l i n g a h c a t c dn r c t a l l i cs o l u r i o ns a r u r e t c dw ' i t h r h c cornponcntsnecdcdto grou thc lavcr,u hilc that solution i s i n c o n t a c t l l ' i t i r t h c s u t r s r r a r ci.n c h c m i c a l - v a p o r d c p o s i t i o n ,t h c c p i t a x i a ll r i c r i s q r o \ \ ' n f r o m a h c a t c d strcamof gascousclcnrcntsor comp()Llnds, rvhichreactat thc surfaccof thc substrarc.i{eccnrlr',researchers have made considcrablcprosressin qros.inq lasersand othcr q L l J n t u n r - u , ' chl le t c r o s t r L r c r u r c u s ,s i n q m e r a l l o - o r g a n i c chemical-\'apor dcpositionfirr cpitaxialgrouth. The nrost aclvanccd scrniconductorheterostrucrures rcquirc specialfabricationtcchniqucsunder controlled, hiqh-r'acuumconditions.Thc processof molecularbeam cpitaxv is a bit sinrilar ro painting u'ith sprav guns containing differenr-colorcdpaints.The materialsto be lavered are heated in scparateovens rvithin a vacuum chamberuntil their aroms begin to boil off A compurer. controllcd shutter then opcns and closes ar precisely timed intervals,releasingthe proper quanrir,vof atoms, first of onc matcrialthcn anothcr,from eachfurnace.Thc atoms strike and adhere to the base plate, forming alternatelayers. Nevertheless,the precise control of the chemical reactions that take place at surfacesand especiallythe control of the puriry of the materials is one goal that heterostructure fabricators have not yet attained. Who-

96

BIorr,cHNoLOGY& IvlqrezuaLsSCIENCE: CHeUISTR}'FoRTHE Furrunr

ever first learns how to achievcsuch control-whether it happensin Japan,thc United States,or elservhere-will dominate a substantialpart of electronicsprocessingin thc fururc. The challenge shorvs a verv real nced for technologicalinnovation.

Warnrcr Superconductot's

,:.

: t -

For more than a decadc, rcscarchcrshave been toving rvith the idca of building intcgratcdcircuirsfor computers using superconductingIoscphsonjunction srvitchcs. junction consists of nvo tliin lavcrs of A Josepltson superconducring nrctals,such as lcad or niobium, whiclr act as clcctrodcs,scparatcdbv an cvcn thinncr insulating Iavcr.At liquid helium tenrpcraturcs, thc mctal'sclcctrical rcsistanccdrops to zcro, and clcctron pairs can tunncl a c r o s st h c i n s u l a t i n g j u n c t i o n . A u c r t c r n a l l v a p p l i e c l voltagc can stop thc currcnt fkxr', and thus this dcr,'icc cau be uscd as a su'itch. Dcspitc thc disadvantaqcof having to u'ork at tcnrpcraturesclose to al'tsolutczero, loscphsonjunction circuits havc appcarcdlrtractivc fbr conrputcrcircr.rits bcc:rusctlio' su,itciron .rrrdofT fastcr artd enrit ortlv onc onc-thousanc'lth as nruch hcat as scnriconductor transistors. Until rcccntl\',hon'cvcr' rcscarchcrshad had vcrv littlc succcssin finding matcrialsthat bccorncsupcrconductorsat higher tcmpcrarurcs. Thc best thcr coulclfincl \\'erccertainnrctalallovsthat abruptlv losc thcir clcctrical rcsistanceat ternperatLlresbrelow' 24 K. ln 1987, thc situation changcd dramaticallvrvith thc discovcrv of specialceramicsthat rcmain supcrconcluctors at temperaruresno\l, as high as 90 K. Bccauscthat tempcrarureis greatcrthart thc boiling point of liquid nitrogen.,though much lorver than room temperature, much less costlv refrigcrationtechniquescan be used to cool the ceramic rnaterialsenough to rurn thcm into superconductors. Therefore,the number of potcntial applicationsis grcatcr. Ar thc momcnt, scicntistsarc optimistic for nvo rcasons.Thc.vknorv thcv can shift thc critical temperaturc at rvhich a matcrial becomcssupcrconductingby varving the composition and thc stnlcrurc of thesencw supcrconducting ccramics.Thcy arc also certainthat investigations of how thesc matcrials achievca superconductingstatc

...cltcnristn s rtd ennincn's havejoincd thc cxbilnrnting qutst to tntdcrstand hightctrtpn'atu.t'c sttp ct'con d uctors, iu tp t'at,c tbcir propcrties, and puslt thun hfio prncticnl con trtrzl'cinl applicatior ts.

\\'urrEsrors

Ivlaterials fbr Advarrccd Ebcnotic Dn'ices

97

:-' 'a

will lead to the elucidationof fundamentallynew mechanisms fbr supcrconductiviq'.Both advanceswill likelv lead to devices-from electronic circuits to electromagnets-that opcratc at rclativelyhigh temp':ranlresbut posc no resistanccto elcctriccurrents.For the world that we live in, these advancesare Potentially as important as advancesin molccular biologl'. From a materialspoint of view, the greatestalnount has been focusedon yttrium-barium-copper anention of oxides. However, thesearc not the only ceramic materials that show superconductivin'.Researchersare beginning to look at other possibilities,especialll'asthev learn what specialelectronicproperticsto seekin specific materials. A h o s t o f c h c m i s t s a n d e n g i n e c r sh a v e j o i n e d t h e suPerexhilaratingqucst to undcrstandhigh-temPerarLlre into them push conductors,improvctheir propcrties,and practicalcommcrcialapplicrtions.One earlv application mav be for magneticficld dctcctorsand simple electronic

jl",'

-:.:, :.\

(

device s.

Dinn m tils Sparklhry Poten'tial A fien' sparkleisu't all that nrakcsa diamond so e,vecatching.Its hardnessand its abilin'to conduct heat and to act as an elcctrical insulator make diamond an attractive matcrial for elcctronic circuits designed to sun'ive high temperaruresor wrthstandintenseradiation. Although it is hard to imagine a way to fabricate diamond into thin shcctsof the sort used for siliconbased devices. some researchersbelieve that a furure generation of electronic devicesmay be based on dia' mond-it thev can overcomecertain problems. What is needed is an economical, practical method for laying down and then etching thin diamond films on silicon and other surfaces. Diamond is attractive becauseit carries electrical pulses extremely quickly. Its transparencymeans that it can transmit optical signals.Becausediamond is the best known thermal conductor, it could be extremely efficient in diamond-basedelectronic devices.Complete impermeability to oxygen and similar speciesgives diamond many of the properties one can hope for in an almost ideal device. The basic process for generating diamond coatings

I

98

.{.1 il

'il

:i!l ' l

-'.1

. ..1 l '

. . ,I rL I

1i{

'-iI t,,i ;'l { l

,i'l : i'

i ., : ;

BrorrcHNoLoG)' & MarEmarsScrErucr: CHeursrRyFoRrHr, FrrruRr

involves passing a gascous mixture of methane and hydrogen moleculesar armosphericpressurethrough a microwave bath. This process breaks up rhe molecules into hvdrogenand carbon aroms,which can then settle onto a silicon surface.This chemical vapor deposirion technique is not unlike thar used for gallium arsenide $rucrures. The presenceof hydrogen appearsto be necessarvro ensure that carbon atoms end up in a tetrahedral diamond crvstal arrangemenr rather than in a planar graphite strucrure. Hydrogcn aroms seem ro pick up "dangling" bonds on a frcshlv laid carbon surface,which preventsthe carbon'sstructure from collapsing into the form of graphite. Moments later, carbon aroms replace the hirdrogen atoms, and the crvstalline diamond film continuesto gro$'. It takesabout an hour to lai' dou.n a l-micrometcrthick diamond laver. Each film consisrso[ a random arrav of individual diamond crvstals abour 200 angsrroms across.Rescarche rs are no\\' trr.inq ro specd up the dcpositionrateand to build filrns that consistof a singlc d i a m o n d c r v s t a l . T h a t a c c o r ' p l i s h m e n rs h o u l d m a k e diamond fabrication verv sinrple.Thc nc\\. proccss is potentiallvcheaper,cleanerand more 'crsatilethan hight e m p er a t u r c , h i g h - p r es s u r c t c c h n i q u es n o \ \ ' u s ed t o producesvnthcticdiamonds. A diamond film's firsr applicationmar-bc in microelectronics.Becausediamond conductsheat like a meral' tinv diamond slabscould be used as basesfor elecrronic circuits that need to sur'il'e high remperarures. conl'entional siliconchips usualli,canrloru'itlistandrcmperarurcs greater than 300 oC. How'ever, diamond.b,ased devices could be used as sensorsin enginesor nuclear reactors. Furthermore,becausediamond does nor overheateasily, more circuit elementscould be packedonto a diamondbasedchip than on a siliconchip. In the United Srares,scientistsat the Naval Research Laboratory in WashinEon, DC, and at MIT,s Lincoln Laboratory have long worked on designs for diamond semiconductorcircuits. Until recenrly,th.y lacked marerids on which to resr their designs.New researchefforts to produce diamond filrns ar Penn State;North Carolina state universiry in Raleigh; and at the Researchriiangle Insdrute in ResearchTiiangle Park, NC, will now provide the essentialmarerialsfor that rvork (3).

9. \\'Httr,stor,s

I

J

I

I I

l i 1 t

hlateriak Jbr Ad.varrccdEbcnnni.c Dnices

99

Early in 1987, the Japanesecompany Sumitomo announcedthat it had succcededin dcveloping a diamond semiconductor.The diamond film is doped with a small amount of phosphorus, rvhich rurns it into an n.rype scmiconductor.The new process brings scientistsone step closer to creating true diamond transistorsand other elcctronic devices. Researchersworking at the MIT Lincoln Laboratory (4,5) havecreatedauthentic,though highlv primitive, transistorsin a thin diamond film b1' sprafing ions in pattcrns in the presenceof nitrogen dioxide trapping devices. The Japanesecompanv lvlitsubishihas alreadycome up with one commercial application for diamond in information s1'stcms. Thc cornpan\/now manufacruresa verv fast Winchesterdrive in u'hich the magneticmedium operateswithout a lubricant.\\'ith thc hcad in verv close proximin' to the dcvice. a dilmond-film rvear-resistant barrier prevents catastrophiccrashe s in the event of occasional.inadvertentcontact benveenthe head and the spinning drsk.

LoolzhryAJtend Recent advancesin high tcmperaruresuperconductiviw and the fabrication oi thin diarnond films are rwo significant signpostspointrng tow'ardtechnologiesthat m a y s o m e d a vp l a ) ' c r u c i a l r o l c s i n m i c r o e l ec t r o n i c s . iv'Iicroelectronic tecirnoloqiesare changingverv rapidli', and anv nation that expccrsto remain at the forefront of nen, technologiesshould investbroadly in invesrigating materialsthat have the potential for dramaticall.vchanging the world. The ctn'idendsmav not come this vear,or next vear, but perhaps25 ycarsfrom now. Nevertheless, thc investmentis necessarvto ensurea sccureeconomic

furure.

RefYrences l. Binnig, G.; Rohrer, H. Angcw. Chem.,Int. Ed. Engl. L987, 26,606. 2. Morkoc, H.; Fischer,R Eur. patentapplicationEP 232082 A2. 1987. 3. DeZries,R C.Annu. Rm. Mater. Sci.,L987, 17, 16I. 4. Geiss,M. W.; Rothman,D. D.; Ehrlich, D. I.; Murphy, R A.; Lindley,L.T., EDL-*, 1987,8, 341. 5. Geiss,M.W; Efremon,N.N.; Rothman,D. D. J. Vmuum Sci.Tbchnol., in prcss.