DTIC ADA262818: Specific Contact Resistance Measurements of Ohmic Contacts to Diamond

8 Pages·0.44 MB·English

by  Defense Technical Information Center

AD-A262 818
a.vi tm te (11o ckiaa
AGENCY USE CNLY re.j. b"IAI 2 REPORT DATS 3 ýi4' * : ..... ..
February 1993 l , qf-r
4 T71 a AND SUBT'LE.
SPECIFIC CONTACT RESISTANCE MEASUREMENTS OF OtlM1II( RC:Z Wlf
CONTA(CT TO DIAMOND PR: Z06)II 5)N
6 AU, OR(cid:127)S: ;WU: DN:I)O9052
C. A. Hewett, J. R. Zeidler, M. J. Taylor, C. R. Zeisse, and K. L. Moazed
7 PE RORMING OANMIZANTION NAMEiSI AND ADOESSiES) 8 PRF N- ORM. ,N ',C-
Naval Command, Control and Ocean North Carolina State University P.E ",
Surveillance Center (NCCOSC) Department of Materials Science
RDT&E Division and Engineering
San Diego, CA 92152-5001 Raleigh, NC 27695-7916
9 S, -NSORNL%4ON7QfThNG AGENCY NAMEMSA NDA DDRESSES)
Office of Naval Research
~
800 N. Quincy Street A -
1"1
Arlington. VA 22217
11 12, 1993
SJP~t~v-NAPY OTEAPR
2a. O!STR, S7ON:(cid:127)VA-A tAr; STA5STM: I 'T'B(cid:127)C!C".i
Approved for public release; distribution is unlimited.
"13
A.&STRACT(N ia.rr 200 wZcds)
We have demonstrated the applicrhility of the specific contact resistance measurement scheme of Reeves to semicon-
ducting diamond. Four sample types were usd: highly doped epitaxiai films on < 100> and < 110 > type hIa substrates, a
type UIb diamond 0.25 mm thick, and a type IPb diamond 0.05 mm thick. Measured specific contact rvsis-tances ranged from
2 x 1o- to I x 10-2 ohm-cm2.
Published in New Diamond Science and Technology, 1991 MRS International Conference Proceedings.0,
14 SUBJECT TERIVW 15 NUMI3ER OF PAGES
electronic devices
dseiammicoonndd uteccthonrosloLN 1 PRCE CODE
17 SEcunrrvy cZASiFFZATN 18 SECuRfTY Ct(cid:127)ASWSFATION 1g SECURITY CLASSJF.ATXON 20 UMITATnN OF ABSTRACT
OF REPOR oF nh4SP ACE OF ABSTRACT
UNCLASSIFIED UNCLASSIFIED UNCLASSIFIED SAME AS REPORT
WAR7 'AO 2&1,' /MT SlandaMT1o* ,2W.( FROIM
ZaNAMCIVýF 'i SF- .t. ... .
J. R~. Zt'idler I
A~cc -;Ion For'
I-
UNCLASSIFIED)
fSPECIFIC CONTACT R±.SISTNNCE MEASUREMENTS OF OHMIC CcNTACTS To
DIAMOND
C.A. Hewett*, J.R. Zeidior', J.J. Taylor*, C.R. Zeiss**, and VL.
Moazed**
*NJVvA I'tl :.yu ms. Center, Cod.' 'Ii atalI i n.t B1 vd.
;a n Die-ýp) 'A ',? I t2-.OOO
**Dp I MaIterials Science and Env.n--ner ;nq, North Caitolin.i
State Unlvl-rtsity, Raleigh, N( ' i
* ABSTRACT
* We have ' ctatdthe appl icaj,.' the speci izc <~-n
tact rce i. t.i- 7.c cit emcnt scheme~ o to fscn iconduct-
ing diamond. !'ý,u -- rimpl1e types, wer h qhlIy doped
ppitax1 iil , on 100r. and - 11f) ý , '1t at,; a type
Ilb diamond C)2. mmt thick, ind ,I type!t c!iond0 .0" Mm ttihik.
Measured spe,->ir : contacýt. reiticrom 21½ to
jxlO'2 ohm-cý
IN'TRODUCTION
-IeThe unique mae~land electr P; roI I f 11 ,1, -,i
maeit a potenIt'I'l Iy very .mport~int c~s I ,,* for, ui-e isn h
power or ii rqh t requency app1 icat cnc;' 'I" IIs in hi qh tfemperýi
tore and corro';ilye coy irorimentn. ' ,vr ,I.. r; hive rf cent ly
reviewed the fotlvat ion aind pros~pect: t :zonduct i nq Aii m'r i
asdelectronic devic-es 12-4 . Whi', --- ciritici, I r11'-r:
research were etifjd we note thaIt hia.-: ; been MrioJ tn
key areas. 1`hc gjrowth of larqv homo-fn it films dopedWt
boron, a major ;tep to-ward the qrowth - e crystal Jii--s-st
films on sub!;trittes- other than diaric( ent''t, reporteId 1-,,
Regrowth of ri t.ondarnaqc layers, iri- y ion inplant. it * i
int diamond ha';, il';o been reported 7q- in torminq c,
,-rica I contact;ý thot h rectify rnq and :id 1i amondi tl- 'o
*been reported in the- literature 1-I paper
measurements, of the specitic conatact inccc o.f ohmic ln
tacts to diamond are presented.
Ohmic contac-ts; are charaicter1i.!vd L,.i n hco
resist ance ani oqit the cotc-sm b or)nt o.1.
No)r-maI i z in q th;, ntr act resi1;t anciýet , tIh -, nt*i ct ar ea qIi ves! r me,
to the specific- k:ontact resntotance, r,., ic he Ideal cas~e o?
un i fsrm current !! .4 perpendicuilar to, n-ot act we have
where A iný the contac~t ati it) t-iel it y howeveýr, the
~i
current flow ir; rarely perpend i:u r , ind the in ito res;ist anco-
of the sevmicond~ict or lead!; to cur rent r4cwt sq -1', Meann I e -
ment. techn iquit;ý have been devel1oped to 0- 11w Tthe ci~irrent ''rowd i nq
and bulk rp! jot ance, of the nem icon~tuct?ý t- t,,- dec: invoIvol f rom
P(,thereby uniquely determnin ing the sp ciof nt act res;istance
t15-16)j. Conceptua-.lly, the simplos;t technique Is that, of Cox and
Stac I,j, iii whiich iconitct s are m~id-c 1toth the, front Ind
back sides; 0f ftfieim The primal y isivintiqes of this- tech-
nique are the eaein procunc; ing theý cot ict-, anid the :;impl if i-ed
ainalysisj! dtif. to the, qeomot ry. Itiiwevei de I, the
semi-i n!u fat inq tiit ire of bulk, dilamonit Ith ers is- a lar qe rle
iN-- 0- -V'd A.,1lr l' R 0iI~ v-i,:ftv.iq
resistance contribution to the total res;istance WrO. the sub-
strate itself. Thus, in order to determine the specific contact
resistance one is forced to take the difference of two (nearly
equal) large numbers. A second technique utilizes an array of
contact pads of equal size, but varying separation contacting a
thin conducting layer. The contact resistance in these struc-
tures is analyzed using the transmission line model [15. The
primary drawback to transmission line model measurements is the
need to perform a mesa etch in order to reduce the analysis of
current flow between contacts to a two dimensional problem.
I'o techniques for eliminating the mesa etch requirement
from the transmission line model measurement have been demon-
strated [1,17]. One approach (17] is to simply extend the con-
tact pads across the width of the sample. forming an array of
contact lines. The second approach is to use a ctrcular test
pattern consisting of a central dot ard concentri"c ring con-
tacts. The measurements are interpreted using a circular trans-
mission line model. For two reasons we judge the second approach
to be superior to the first in analyzing diamond samples. First,
the use of contact lines requi-es rectangular (or square) samples
and is not adaptable to samples of irregular shape. Second,
natural diamonds are frequently non-uniformly doped and the use
cf circular contacts permits probinq over a smaller area. Thus,
we chose the circular transmission line model for determining r,.
For the details of the analysis, the reader is revearned to refer-
ence rl!,
Several constraints must be placed on ohmic contacts to dia-
mond. First, they should have a low contact resistance. Second,
they should be able to withstand the operating conditions for
which diamond devices are intended; e.g. high temperatures.
Third, they should be strongly adherent to the diamond surface.
Fourth, they should be compatible with conventional device pro-
cessing techniques. Ohmic contacts produced via a s;olid state
annealing process have been studied extensively and are believed
to satisfy all four conditions ['91. In this process a thin film
of a tranwition metal carbide forming metal is deposited on the
diamond surface. Annealing at high temperature (9'50C) leads to
the formation of a carbide layer at the interface. This layer
provides an intimate contact to the diamond and promotes good
adhesion. Fig. 1 shows the resistance measured between two
Molybdenum-diamond contacts as a function of annealing time at
950"C in hydrogen. From the figure it is clear that annealing
the contact produced a decrease in Lotal resistance of several
orders of magnitude. Auger electron spectroscopy studies indi-
cated that the decrease was associated with the formation of a
carbide phase. Similar results have been observed for Ti and Ta
contacts. Measurements of the specific contact resistance are
required in order to determine the metallizations that provide
the lowest electrical impedance for device applications.
EXPERIMENTAL PROCEDURE
Four different diamond samples were used in this study. The
first sample consisted cf a diamond film with a high boron con-
centration and about 4 om in thickness grown epitaxially on a
i '00, type Ila insulating diamond substrate with the dimensions
of ' x 5 x 0.25 mm. The second sample consisted of a 6 Am thick
diamond film with a high boron concentration grown epitaxially on
a 'II0, substrate. These films were grown at MIT Lincoln Labs
'j a
us;ing the procedure outlined in b on naituralI iamond substratest
provided by Druker International, B.V. The third siample wa.; an
irregularly shaped type lib diamond 0.25 mm thick. The fourth
sample was identical to the third, except that it had been
thinned to 50 am in thickness.
, 'a.,.',
40 O
3o 50 \
4A
aT S
Figure I Resisutance between two Meu.diamond contacts as; i funct ion
of annealing tome it SSO'C ýn an H1 ambient.
The samples were cleaned using de ontam, deionized water,
and ethanol, sequentially, The samples were then coated with
photoresist and patterned using standard photolithographic tech-
niques. Samples were baked at 120"C for 20 minutes tollowing
photoresist patterning and loaded into an Ion pumped ultra-high
vacuum system (base pressure 5 x 10-' Tort). Electron beam evap-
oration was used for deposition of the carbide forming metal (ei-
ther Ti or Mo). The thickness of the carbide forming metal was
about 100 A. Subsequently, 150o A of Au was deposited from a
resistively heated boat onto the surface of the Ti or Mo without
breaking vacuum to prevent oxidation prior to annealing. The
pressure during evaporation was 2 x 10-7 Torr. Film thicknesses
were determined using a crystal monitor during deposition. After
deposition a lift-off process was used to remove undesired metal,
leaving contact structures with the geometry shown in Fig. 2.
The dimensions (in the notation of Reeves) were rl',l.65r 0 ,
r ý2.74ro, r '-4.4ro , and r2 '-. 45r with r ell.7 em.
1 2 0 0 0
Following patterning, the contact structurei; were probed
using a Keithley model 220 current soulce and model 196 D)M.I.
After probing, the samples were baked at 120"C for approximately
20 minutes and then annealed in a purified hydrogjen anowient at
150'C. Anneal times were 2 minutes for Ti and 6 minutes for Mo.
After anneal ihg , the samples were re'measured.
77
ro= 11.7 um
r, =20-4m -ur
r =-- 32.5 m -rn
r2 = 52.3 pm(cid:127)
t2 = 6513 ,m
Figure 2 Circular contact pattern used for this work, The dimen-
';:cn; are r-l'l.61,ro, r?2.74rc 1 , r 4.,r' 1, and
! 2" .' ro with r0 =l1.'7 m.
1icw me ur-ements consisted of placing tungsten probes on the
inner dot, the central ring and the outer ring. The tota1l
resistance between the central ring and the inner dot, and the
central ring and the outer ring were measured. The end
resistance was determined by passing a current throuqrh the cen-
tral ring and the inner dot and measuring the voltage between the
central ring and the outer contact. This result wai; checked by
switching contact pairs and remeasuring. Using thene three
results, the specific contact resistance can be (:calculated in the
manner of 'I).
RESULTS AND DISCUSSION
Ohmic contacts between a metal and a semicsndlu.tor may be
formed either by lowering the effective barrier height, or by
heavily doping the semiconductor at the interface. meavily dop-
ing the semiconductor induces a decrease in barrier thickness,
allowing tunneling of the charge carriers. Lowering the barrier
height is difficult since diamond has a barrier height essen-
tially independent of the metal work function (18). Increasing
the doping in natural semiconducting diamond is a non-trivial
task, but may be accomplished rather easily during epitaxial
growth of diamond. Table I shows the values of specific contact
resistance determined in this experiment. It is interesting to
note that contacts to the epitaxial layern (epi-'100t and epi-
*It0o) showed excellent behavior as deposited. Contacts formed
by placing tungsten probe tips on the diamond surface showed only
a slight deviation from linearity. Contacts formed by the Ti
films were highly linear. This may be due to the highly doped
nature of these layers. The measured value of specific contact
resistinces after annealing are more than satisfactory for device
fabricatiion, Lightly doped diamond (represented by the "thin"
and "thick" samples) are much more d fficult to contact. Non-
ohmic behavior was observed for the at(cid:127)-deposited contacts in
these samples.
Several factors influence the accuracy of these measure-
ments. Of primary importance is the thickness of the conducting
layer. Correct interpretation of the transmission line
measurements requires that cur rent flow between contact pads be
two dimensional, i.e., no vertical flow of current in the dia-
mond. For the epitaxial layers this condition may be assumed to
be approximately true. For the 50 "m and 250 om (bulk) layers.
however, there is almost certainly a large component of vertical
current flow in the sample. This adds a large spreading
resistance term to the overall contact resistance, which in turn
greatly adds to the complexity of the transmission line analysis.
The possibility of using an approximate geometric correction fac-
tor for the thickness of the diamond is being explored.
Table I
Specific Contact Resistance Result!; for Various Samples
Sample Cond-icting Metal NA (cmm-) rc (il-cm2) rc (.l-cm2)
Layer at 100 K As-deposited Post Anneal
t(hhiimc)k ness ______
epi-<100> 4 Ti xioli Il.- 2xsix0l-o''
epi-<ll0 6 Ti 7xi0O 1 .xt(-' 2xl0-"
thin 50 Mo (aj) 1014 over-ranced 1. x-I (
1(a)
thick 250 Mo 1014 over-ranged I i.4l10 0
(a) Representative value from other type lib diamond sa mples.
It i al:;o of interest to note the spacific contact
resistance of the SiC and Ta carbide contacts on type [lb natu-
ral diamond substrates studied by Fang, ot al [l1], They
obtained an average value of about I x [o53 ohm-cm2 for both
cases. The conventional linear transmission line was used In
their study leading to the necessity of forming a mesa. This was
accomplished using a CF4 plasma etch. A thin conducting layer
was formed by inducing ion damage in the diamond. Thus the dif-
ference in rc betweer the two studies may be due a differei:'e in
doping level, a difference in contacting lightly damaged diamond V
versus undamaged diamond, or an effect ot the plasma etch.
CONCLUSIONS
Specific contact resistance for th,- Ti and Mo refractory
metal carbide contacts to diamond have been measured. The values
dareete rmin inreeda sofnroambl et hea gjcreoenmtaecntt s wtioth thteh osbeu lkp utbylpise' he[dl b (cid:127)-edviaimouonsdly samIl1p1l3es
for SiC or Ta carbide contacts, indicatinq the usefulness of the
circular transmission line rcsndrl in avoiding I mesa etch and
allowing spec"ific contact resistances, to be easily determined-
rWW' "
ACKNOWLEDGEMENTS
The diamond samples used In this study were provided by Dr.
Michael Seal of Sigillum, BV. and Drukker International, 13.V.
Preparation of the epitaxial films by Dr. Michael Geis at MIT
lincoln I^ibs is greatly appreciated. The authors gratefully
acknowledge the support of the SDIO/IST through Mr. Max Yc(cid:127)er of
the Office of Naval R'esarch.
REFERENCES
i.G. K Reeves, S)l id-State Electronics (cid:127)3, 487-490 (1980).
R.F. Davis, Z. Sitar, B.E. Williams, H1.S. Kong, I1.,]. Kin,
- W. Palmour, J.A. Edmond, J. Ryu, J.T. Glass, %nd C.1.
Carter, Jr., M.ttrials Science and Engineering 31ý),J 1-104
( 1988).
3 K. s;henai, R.S. Scott, and II.J. Baliga, IEEE Trant;. on Elec-
"tron
Dev., 36(9), 1811-1813 (1989).
SN. Fu jimor, H. Nakahara, and T. linal, Ja5pin. 1. of AppI.
ihys. , (cid:1)2 (5) , P24-827 (19'90) .
IH, . S~hi'ci, Y. NiQ.-hibayashi, and N. Yu),imori, Jipýin. J, of
Appi . ihys. , .0 12), 2103-2154 (19R)
SM. W. Ce ss, in Dlia mond, Borton Nitr ide, 0.; icon Carbide and
re'l.ated materials, 'dited by J.T, Glas:o, N. Mes.;icr, ind N.
FuL mcri, Miter. Pos. Soc. Proc., Pitt!burqh, PA lh89.
SB.V. Derjaguin, B.V. Spitsyn, A.E. Goridetsky, A.P. Zakhla-
roy, L.L. houilov, and A.E. Sleksenko, J. Cryst . Growth, 31_,
S B. .u, . - ;andhu, N.P. I'arikh,. I., S.wanson, .ind W.Y.
Chu, ' c!. Instrum. and Meth., B41,_ 422-423 (1330)
.Maz-.i, .1P.1. ,'eidler, and M.J. Taylor, J.Appi . Phyf.
*3(cid:127) 3'], 2,4',-12.4 ( 1'i0i3
1. riLr, .T. Phyn. D: "\pp3 , Phyf_, 2?, 1'62- 1'04
11 F. ýr, C.A. Hewett, M.G. Fernandes, and S.S. Iau, IEEE
1, <' (cid:127)-,n ( ,.vi c eu, .34( 9) , 1783- 1 "8 6 (129q9 ).
:2.';h. GC 5 ldenhlat , 3-A. ,rot , C.W. Fiat Ield, A.R. Bad, ian,
,,nd T. ladzitn. [FEE Flectron Dev. Iett, ;_ (9)), I11-312
(!'.90 .
S. (',hen and G, Sh. Gildenblat, in VLSI Electronics Micro-
structure Science, Vol. 13, ed. N.G. Einopruch, Academic
Pre!..s, Inc., New iork, pp. 87-133, (1986).
14 :'.-Y- Ting and C.Y. Chen, solid-State Electronics, k4,
431-438 (1"71) -
1'. IfF. fierger, !:olid-State Electronics, 15, 145-158 (19,2).
1i P.H. Cox and N. Strack, Solid-State Electronics, 10,
21 3-1218 (1967).
If Il. Shiomi, H. Nakahata, T. Imai, Y. Nishibayashi. and N.
Eujimorni, Jap..J. Appl. Phys., 28(5), 758-762 (1989)
18 C.A. Mead, Solid-State Electronics, 9, 1023-1033 (1966).
4 V