■■"", ■ ' '■'■ -"■■ ■ ■ ■'

^mmmmmtmmmm mmmimm j urim

AD-A 010 8 02

STUDY OF THE ELECTRONIC SURFACE STATE OF 111 -V COMPOUNDS

W. E. 5pice r

Stanford University

P r c p a r ed for:

Army Electronics Command Defense Advanced Research Projects Agency

15 October 1974

DISTRIBUTED BY:

mi] National Technical Information Service U. S. DEPARTMENT OF COMMERCE

- - -—— . . t—

"-—

171174

STUDY OF THE ELECTRONIC SURFACE STATE OF III-V COMPOUNDS

(M C

SEMI-ANNUAL TECHNICAL PROGRESS REPORT

W- E. Spicer, Principal Investigator

15 October 1974

o <

NIGHT VISION LABORATORY U. S. Army Electronics Command Fort Belvoir, Virginia 22060

Sponsored by

DEFENSE ADVANCED RESEARCH PROJECTS AGENCY DARPA ORDER NO. 2182 PROGRAM CODE NO. 4D10

CONTRACT NO. DAAK02-7Ä3069

Effective: 1973 September 01 Expiration: 1975 August 31

lMthVleWVnu c?"clusions contained in this document are those of thm aut or and should not be interpreted as necessarily represening the off1c1al policies, enher expressed or implied, of the Defense Advanced Research Projects Agency of the U. S. Government

R«produC*d ! y

NATIONAL TECHNICAL INFORMATION SERVICE

US D*p«rtmenf of Commorc^ Spr.rgf.eld, VA. 22151

SOUD-STPTE ELECTROniCS IRBORRTORV

STRIIFORD ELECTROniCS LRBORHTORIES STRHFORO UniUERSITV - STRHFORO, CRLIFORRIR

1 •mm^^mmm^m

The i • urn,:, ampul ««a placed In a copper aide are ol aboul ■ c« total

length, aa shown m Pig. i. uter ptmp down and bakeout of the aytte«

the aide arm »aa cruahed to break the aaipul. The cs reaialned in the

«ide arm alter the ampul »aa cruahed. it was neceaaary to heal the aide

»« "ith heat tapea or a propane torch in order to evaporate ceaiu« at a

BUfficlenl1\ rapid rate.

The ceaiu« was evaporated onto a copper substrate whin, had been

— 1'"' to approximately l- K with a liquid nitrogen cooled cold strap

■• Fig, I)- After aeveral hours ol cooling the substrate temperature

reached a minimum ol 1 K. All data shown in this article was measured

at a temperature between | R ,„d I K on films which had never been

wa naed above l • K .

It took approximately one hour to evaporate a sample sufficiently

thick for measurements. It was not possible to measure the thickness ol

the cesium layer, but it was visible as a dull, dark layer on the sub-

strate. A white-green layer ol es r apparently much thicker than that

°n the substrate, foraed on the cold strap. During evaporation the yield

»aa monitor,., at . eV. The yi.id firat rose as a monolayer of Cs

»•a depoaited, and then fell as a bulk cs lave,- fomed.

During cs evaporation the pressure roa« to about > lo" Torr.

Dunn« measurement the pressure was approximately } X lO*9 Torr. The

ayaten was pumped by an Orb-lom pump and . small vaclon. it was neceaaary

to turn the Orb-ion off dunn« measurements because the liKht from its

filaments caused photoemission from the Cs

The cs ••■ oxidiaod ehile cold by admitting high parity (K ppm

impurities) oxygen into the ayatem through a varian leak valve.

U

Th6 PrMence oi Cs ,', th« »yt- -used la,-,. revrw current.,

Wlm" * UC0<1 l0" "" '^ '^^ i" th. EDCs, .UtUr to thoM shown in

"'• ' '" R,'i"'"'1- • Th- rever.. current peak. h.v« b««, r^ov«, rro»

th« FlK's Ihown in i his papec.

Th. cs .1.0 io.#Ped th(, (.,Illt(rr to traani r(,sistant,r to ((nly abou)

' «*».. we ..r. no, :ih,(. to r.lM the r..i.t.nc. by h..tlng th. «itter.

Tins lou res.stanee n.acie U unp-ssihlo to take EDCs at photon enerKleS

for «rhich the rorward current uas belo. i ~1' amps.

RESULTS

A. Introducl ion

Plfure . .ho*, the photoelectric yield of the oxidized ceelo« ftl.

as a function oi oxv.en exposure. Pro. it we see that the oxidation pro-

c... up to . x 10 ung^lr. (n, i- ' rorr-^ond) ^ he divided .pproxi-

— ^ """ four region., m r^o« i there is onl, a slight ineroase in

v-ld »ith exposure. At ahou, 1 ,, a .o-e apid rate of increase is ohserved

»P to .bout 100L, a, •hid, point the viele, heBins to rise rapidly. At

I- •• »ntor reCion IV where the ylold drop, slowly with exposure.

EDCs ol OTidLed ceeii» are shown . n Pig«. - through l( . Two sets

01 EDCs a,- presented for hV . 7.7 aIu) llV . , ,, eV! ^ ^ ^ arc

ttken fro- two different runs ol the experl.ent. The EDCS were quite

reproducible from the two runs. The EDCs are normalized to that the

a"- under th. curv. equals the yield at that ph()ton energy. Notc ^

there is a scale charge between Figs. 3 and k. h and , 7 and 8, and ,„, ,

B. Low Oxygen Exposure (Regions I and U)

EDCs for .lean Cs are shown in the hottom curves in Figs. J and J.

PoakA .s due t-unscattered electrors from the Cs valence band. These _ 1

w ** mfm^*mmm*mmmmim

tDCs agree very well with previous work on clean Cs ' " except thai the

weak si i-ii. i in-. Laheled I! j in >n\r curvea is auch weaker than the peak In

the same position In the work ol Smith el al. Since shoulder B Is at

the sane location where the strong oxide peak B develops, we believe thai

the clean cs Investigated by Smith actually had a small oxygen contamina-

tion. Our curve for 1L exposure looks more like the EDCs ol Smith than

our EDCs Iron, clean Cs . We believe that our Cs was very clean because

the Cs was distilled under vacuum once before the ampul was placed in tin

experimental chamber, md because the long path around two bonds which the

c> had to traverse on its way to the substrate (see Figure i) caused

further purification ol the Cs . On the other hand. Smith's apparatus

featured ■ line-of-sighl evaporation from the Cs ampul to the substrate.

Pe«k Bj lias previously been attributed to valence electrons which

have been scattered by ■ surface plaamon. '- Our data docs not preclude

thai explanation; we still sec a weak shoulder B- before oxidation. The

shoulder Is fairly broad compared to the oxide peak B , which tends to

suggesl that the shoulder Bj Is not caused by oxygen contamination. The

sh "ll(,"r could ^ ' '"-'■'l by the surface plasmon as originally proposed.

'"' '' could be due to small quantities ol contaminenta In the Cs

In Pigs. ■ and it is important to note that the magnitude of peak

A deos not decrease for oxygen exposures up to 200L, although the growth

«* »tructure below eV is very dramatic. Peak A la caused by unscattered

electrons emitted from the conductiun band ol cesium. The growth of oxide

emission while the peak characteristic ol the metal remains constant is

similar to the behavior seen In Sr , and suggests the same explanation

for the Cs-oxidation as lor the Sr oxidation.

•«" ■ i «i ■iimaHOTK^wwii« j-"OT^«««qqpg«H 1 "immmmmmmm

In t.lu> Sr work it was tound tli.U , until oxidation neared completion,

the oxide was Located beneath the surlaee. The first step in the oxidation

procesi wae the appearance of dissolved 0 ions below the surface. The

[irst step in the oxidation ol cs appears to be identical to that of Sr ,

i.e.. ions are dissolved in the Cs metal. This is followed by pre-

cipitation ol crystallites. Here Sr differs from Cs in that it has one

dominant oxide whereas Cs has many oxides. Thus we believe that at least

up to the L exposure, there is a layer of Cs metal on the surface,

»1th the oxide lying in the bulk of the material.

The extreme sharpness of the peak B for exposures up to 30L is

also similar to the behavior of Sr oxidation, and aRain we believe

that the explanation is the same in both cases: for the first 20-301, the

oxygen at'mis are well separated so that the oxygen-oxygen overlap is very

small. Thus the peak B is an atomic-like oxygen level. From 30-100L,

peaks C and I) develop, indicating that oxygon levels overlap at these

exposures, i.e., formation of cesium oxygen compounds starts.

C. Regions of Large Oxygen Exposure (Regions II, III, and IV)

As can be seen from Figs. ••-' and 8-l0j there are very distinct

changes in the EDCs as the oxidation proceeds from 30 to ^OO0L< Based

on phase diagrams and estimates of vapor pressure, Ron Bell has estimated

that the phase important for activation of photocathodos is Cs0C 2 2

C« and (Our other compositions lie on the cesium rich side of this

composi t i on.

Another important parameter is contained in Figures '5-g). Thi 9 is

the work function eCP , which is obtained from the EDCs by ihe difference

in magnitude of the low energy cutoff and the photon energy. In Figure 2b

the work function is plotted versus oxygen exposure.

- 7

i^^mi^m mm* ^mm |

v' ' oxygen expoaure, the yield (Figure «) read ta a maximum and the

«ork lunrtion reachea a sharp minimum DI about I . ' v\'. In addition a

i.irlv dlatlncl DC is obtained at 1,. Kote in particular that a dis-

tlncl peak (cN La preaenl at -•. eV In Figure I, which la not «ell de-

fined (or any nearby oxygen exposure, Vhus it appears that a fairly

dlatlncl EDC occura for the Lo« work Function form of ecalua oxide. Our

next atep will be to Obtain EDCs from cesium oxide on GoAs and ccmpare

these with the oxidea studied in the present work.

IT. GaSb

Two GaSb samples. 1< cm" n type (in) and K 1'cm"" p (iTp).

hav. been Itudied. The In sample was misoriented by the manufacturer

along the (ill] direction Instead of the (llo) directiw», causing the

''■v-,al '" cleave at a Large angle to the crystal axis. This disrupted

the geonetry of the retardini potential analyser and reduced ivs rssolutio

However, the data obtained from this cleave (Flgurea 11,12) appears in-

teresting. The Banufacturer has supplied a properly oriented replacement

for sample 1 n at no additional cost. We plan to repeat the measurements

of the oxidation of sample l8n, presented below. In the near future an

the properly oriented sample. We also will study the effect of Cs on

this sample.

The cleive was made at a pressure of • X li Torr. The EDCs

showed a tail up to the Fermi level, which was caused by the reduced

resolution of the electron enerRy analyser. No strong emission from

filled surface states was observed. Our estimate places the surface Fermi

level quite close to the bulk position, that is, near the conduction band

- 8 -

^^BMMM»^ - - —^-^^^_-

^wmim ■^-< M m«mmmm -""'■ " .w\n»iiv™w~m^mmvmm

>™. Thl. indicate, that there are „o e«Pt, eurf.ce et.f. in tto ton*.

tP. The .tructure in ,),. mc. also t9n6e6 to be „,..„,. uhu,h sllarp(,n(>(|

— ...u .,,., lighl oxldfttlOB Ho2L)i Tliis Minv ^^^^ i non_unifrtrm

-rk function aero., the aurf.ce, broadening the etructar.., an,, the '

re-ve. the aon-unifomty. Viljoen el ai. *„„« that tl](1 worIc rJtlo|i

oi cleaved QaSfa varied by a. nuch .■ • nv .„, ■ niiun a. . ev tcroa. the surface, and chia

variation vaniahed after a faa daya. upon oxidation ol sample in. the

rer.1 u.v.., aoved towa«|. the va^nce band wuri««. and at lo^L the »ove-

' -a. alao.1 hall a volt. At exposures greater than io\ the Fer.i level

novod ,,uav from the VBM and aft., I, L exposure had returned to .ithin - .3 eV

- its ordinal (unoxtdized) position. See Fl,. 12. The EDC at lo\ exposure

.hould be .ontrasted with the IVp sample at the sante exposure (Fig. ik) .

The I7p .ample appear, to be at a .ore advanced staRe ol oxidation, with

a higher peah at about - .v. The cleave on sample l7p was very rougi,

becauae the sample I. very brittle and has a tendency to chip rather than

- cleave. Thl. brittlene.. seems to he peculiar ,0 sample Wp; sample

1 H cleaved verv well even though it was oriented erong. Wt (.lüaved

"«Pie 1 p only one. because ol the bad cleave. We wiH attempt to make

change, in the - leaving apparatus which .11] lmproV9 „ur chance|J of

ettin, a good cleave before ee attempt another cleave on sample 1 p.

The 1 p fiata is shown in Figures 1- nul 1 -ri, • ri^uj.s 1 ana in. The surface Fermi level

*PPear. to eithir .xperiaental accuracy, to b. at the bulk pelt Ion,

«hxch 1. .bout .1 ev above lho vnM. This „^ ^ thcre ^ ^ ^^

.«rface states near the bottom of the ener.y gap. The oxidaUon behavior

—ptne different from the n type's. The structures were sharper in

th. dean B«. and the Fenni level showed little movement with oxygen

0 -

mmmm^mtii "' ■ "■l J'"——~- ii im.im mm* m in« i i 11 ■ UILH

cxn ..sure. Tlic oxygon peak built up more rapidly. The 10 L curvr looks

very much like the l( L Clirv« Ot t lit> 1( n sample. In Figure 1; we lined

up the leading peak for the two different dopings and it can be clearly

seen that the positions of the Fermi levels ru^cr by almost the lull band-

gap. This suggests that the bnndgap is 1ree of surface states. Future

work »ill include studies of effects of cesium on the clean surface.

III. GaAs

Since our last report, one additional cleave has been made on our

i) -3 ; X K cm ■ p type GaAs sample (sample 19p)• This sample was exposed

to oxygen. EDCs lor sample Tip are shown in Figure U as a function of

oxygen exposure. Exposures up to 10 L have little effect on the EDCs.

At K L, the Fermi level begins to move to higher energy relative to the

leading peak in the EDO. and by Ll L it lias moved upward by about 0.5 eV.

Also, at about K L a large oxide peak has built up.

The behavior of a 6 X 10 CB':I n type GaAs sample (sample ikn)

upon oxidation is shown in Figure T". The overall shape of the EDCs is

similar for the clean and oxidized n and p type samples. Figure 1

compares EDCs for heavily oxidized n and p type GaAs

However, the behavior of the position of the Fermi level is different

for the two samples. On sample ihn the position of the Fermi level is

a'm. st constant with oxidation, while the Fermi level moves up on the p

type sample. The position of the Fermi level with oxygen exposure is

shewn at the top of Figure 1). The behavior of the position of the Fermi

level on silicon is shown (or contrast at the bottom of Figure 1Q, which

is taken from Reference 7.

10 -

^•»■««^■»■■^^»«■«"■■ISBIiBP*^^»^^^"* ^^■5"^|M|«11* ! ii ■

I" .lllc«, „K. Per« u,.! i. „inn,,, ,„ bnth , „,, p tw

"■"■ 0■Vi"a,1"■■ "—S ö" -""" ■»•". .-C, „.n the reml .„.,

»» Si „id, „„„m^ Mt„ mhici p.n the F(!rni. iwi

The brt.,1., .„ GaAs i, «„.,„, tlian ^^ ^ si ^ Ti]( ^^

level is ninned on elean n tvno r,-,a> I ,

»• «. explain tMl bohavl„r ,„ terins ,)f ^^ ^^ ^ ^ ^^

(»0) Surface U,at .. Pave Pn,lMtl, prescnted. ■ „ ^^ ^ ^

^. fl,.. bin.s i,, the As .tM. ., tho surfiu^ The fuicd ^^

«ate. are ...„c..^ .lth thc „ ^^ ^^ ^ ^ ^^ ^^

"" valence ..„„ „«„., an,, d„ n„t p.,, th. P,.rmi .„.,. ^ ^^ ^^

»vE™ esp,)sure „„e. „„, affect the pi„„.„e „f th. .,.„„. .ut#> ^

abou, lo\ e.p,lsure, „ penevc u.af t„e As sites are fule.,, ana ad-

ditional „..v^en p„nd„ t„ surface Ga at»s. 0xy(;en „ands t„ the Go

..« «ould re„„re „rea.in« Ga-As b„nds at the surface, and .„uld thus

=««. interface states. T.ie interface states .„uld then Pe responsible

to. tU pi„n.nB „f the Per« ievel „n the „eavilv „xldlzed p tvpe sampIc,

^WOrt in n« in ,rneree!i „„ . „„„ liKhtlj. ^ ^ ^ ^^

do « -1). Bef„rc ehl< S8,.ple .. (,loaved; an attempt MII ^ ^ ^ ^^

=1"« It, as a preli ,inary experiment to help us prepare for the heat clean-

'- « -res „f GaAs other than the (11,,, c,eavaBe face. Wo also plan to

stud.v el«, on «,. surfacc of lhls sample ^ ^^^^ ^ ^^ ^ ^_

^vioro, c. ™ »«.pie iop rCportpd ln our previous ^^^ ^^^ ^

11 -

■ ■■ ■ ■

^m^mmmmmmmmFm^^mm^mmumi nmmmemmmm^mmm'^mmimmmmmmi^^^m^^mmmmammnrwrmmf^mr^^^im'm] n iiu i\iiut< mmt « HI I

Will also nttempt to correlate our bulk Cs-oxide studies with Cs-nxide

on sample 1 p.

We plan to oxidize sample I'p with special attention to carelully

studying the behavior between K L and If L. the exposure region where the

movement oi the Fermi Level and the change in shape of the EI)Cs is seen.

12

. .- - . „._ ._...

■ ■■«Ill III M pwpp*l«W"w^^^^"»T»»^"P«p^wi^wr<W""P «"i «'I m\im\mi^m*a^^*mi^*m^ß*n

rLOLLECTOR k.y4(vj /

1/ COPPER

1

I

COLLECTOR c^AN

UJF

/V/\i.<JuM CMV€ UJPE

yKovPfkH SU0srf</\M

^

TOP VIEW

COtVEK SOßSTKATt

CD

COPPEJC SIDE MkAA

V

* WVA/\-\, X ". W\V\

\ I

AMMOULE.

'VACUUM

SIDE VIEW

FIGURE 1

- v. -

■ ■-

mm^^^m

("%*<<*/5uOj£>9iy 013 IA

' '" I ■■«■1 m '"■I « -.-^—^-W- 1

I L. -

T T

•\ •

rn

'•

• * /

J

^d) (VOU^hJ )«> /rm

t o

-2

er 5

< -J

Ui

8 -~ Ui g

, 2 _. O w

O

o

2

U

■■

mmmmmmmm'^mmmwimmmm 1 - ... -.»^

FIGURB .'r r

2.0

> v i

o

t O

Jii

(

1.5

1.0

0,5

0

K.y- 7 7 € 1/

-r -4 -3 -z -i ENEKGY 6£L0W FERMI LEVEL C^O

- i -

0

'" " I—

^

^.0 -

c o u c a.

\

z

fu

01 >

I c

1 n r

LU 0.5

ITOOL

0 - —

kv>- 7. / ( /

FIGURE

L. L2^

i ' -

J

"^wmmumtmmi^' ■^^^'mmmmmK^mmmmmmm mmmmmmmmi^^^

2.5"r

Z.OT

1.5

> i

c

\

10

♦ •O

UJ

0.5

-8

- . i -

PIOOKK

B

Kr« lö.2eV

' --J- i —i ,_

"7 "6 "* '4 -3 -2. ENE^G>f BF.LO^ F£I<M| UEVEL (eV)

laaai—a^^

mmmmmnm^mtm

ISr

0.5!-

o F rciiii:

h If = .' 6, ^ ,. V

s-—• x 10 OL

-8 -7

30 L

v\ -4 -5 -4 -3 -I

ENERGY BELOW FERMi UEVEL CCV)

-"■■-— .. . -- -._.

wmmamm «>> " .■»■P-^W^ mmm^m^mmmmm ■

I Kil'ltK

^\r= fo,i cv

HOOL

'4 -i BNtRGY iClOW FEKMI LEVfL (cv)

ö

m ■ i ■"'I»"»"" ■"■'■■ ™ "" "* '*■ » ■ l"M ■•■

-- . -

2.0 kT= lö.ZtV

-8 -f -4 -5 -4 -s -2. ENEtiGV b£LOiv FL^MI Lev^L (eV)

—

"I D

. »Ill H«i I HWIIMII ^^m^^m-^mmm^mm^rw^m^wmmmmm m^^v^iH^mw*^^*^**'^*^*^

-

/ /

10

■i (/0 ßt* rattii \ »

o*. i on r/ o/v or I 7 p <*A SB

/0. 1 *.V

CLEAU

lo' L. 10* L

FIGURK 1'.

■

HI

* o

'0 c^s

P- < 5Pe

5" f -3 -t- o

FIGURE 15

■29-

"T nppiv«HP^9P^^|.qiw*N.M " ' ' miiiwfiw w^mmmmm, . ■. IIIJ HI«

-

^/o ■

Ep-foriO6!.

Ener^ ^|^ Fe rm, le^t FIGURE 1

Cf^n EDC

wmmmmmmmmm WiiipiiiiRUpam iiiipani ■ ii■ MfiP^IIHWIVMB '■■■— '-''

- 1

„ -3

4H

3-

I-

A \

/ \

CLSAN

105L

lOaL

/

/

K"*,

i ^

\s?

I V

t

I u

\ \

\

Ni —r_.

-4 ^B -4 -i

r u^ fir

-^ -I

w FEKMl LEVtL (eV) FIGURE I?

mt^^im^^m 1 m' ' "' ii ■■■

- aa -

FIGUKK 1

-6 5 "4 '.? -2. I eVCR<S»y BELo>w peg.Hl Le\yEL C£V)

■ ■

"— ' III I I I I ■ pijnupiwr^MMWwmi^^Mmvv»« i "

> O)

X <

Q 'Z. < ÜQ

LJ O

1.4

1.2

1.0

0 8

0.6

04

0.2

0 ^ CLEAN

GaAs CBM

18 I7xl0,ön-TYPE

,14 -3 6x10 cm n-TYPE

3xlOl9cm"3p-TYPE

J L_ i VBM

20 CBM

0^cmo n-TYPE

2xlO,4cm'3n-TYPE

I020cm3p-TYPE

CLEAN 10 8 I07 10 OXYGEN EXPOSURE (Largmuirs)

VBM 10^

FIGURE ig

3 3