1‘(sdxa-[
>
LOE
lauuew
“‘I=‘I
pJem~o3
Japun
uoyauni-ud
(P)
PUT2pJt?MJOJ aqJ
lau
asJahaJ
XO!%aJ uoya[dap
aql sluasaldar
Japun n
svaraqm
n ‘aleJ
snql
Sefq pJemJo3
a]EJ
aamos
aleqsqns
6[[“!3ads
aJaM
Japun
pa3ap
pyos
+u uo
palsal
aq uw HIS
suoypuoa
wq
UO!%aJ uo!la[dap aql uo pasea
(NH) ap!Jl!u
u+d uoqs
sapotp
qanm aJe leql sa8ueqa
lau aql
uo~leu!quIoaaJ
SEWI TvXL3Iio!BLL
‘saJn]anJls
sluaJJn2
aql U!gJ!M
pun !u s u ‘suoypuo~ uoyw!qruoaaJ
put? 94 4 u ‘uoypuo:,
aql u! passaJdxa ‘(1 ‘alI
aql [E ‘z] sa!laug
P”=
‘uo!sng!p ‘awosaq aSJaAaJ uoy~aul
se!q
sluasaJdaJ
e st! qloq
‘aa!hap
uoJoq
%u!sn
nzue[d paleaLJqe3 .b/L;r ueql
aqL
a%el[oA luyalap
e se Jo a%ueJ ap!M e JaAo luawa[a
Ja[[ws
Jo3 luaura[a Jeau![ reauyuou
lu~~ol~uou~ luaJJt-0
~01
paleJqye3 aa!paga
ue
se pasn aq ue3 apo!p aql leql os[e pue a%uw w!q ~01 aql u! saleu!wop
1uaJJna
Z-.4 leql alwpu!
Jo pmMJo3 aq) u! Jaql!a aql qly
‘sapo!p
[eluaur!Jadxa -anJls --
OL
aql
n
(Q - u) -
01
spua[
(d = u)
(‘A = A)
uo!leuqxoJdde
asJar\aJ
= A) se!q pJem103
aql
01
a[dys s!qL
u%!s
%u!Mo[[o~ aql
puo
swq
dxa!u=d=u
uo!la[dap
aq uea
aql u!ql!M
11 ‘,Qp![dtu!s
[ewaql aql
Jod
pua (s,o) JayJw
aJi
a!su!Jlu!
(s‘s)
aau!s
arnldw sluafagaoa
aql s! !u pun ‘d[ag
Jo3 sluafagaoa Qsuap
pue
a3aqMLJaAa
%_o = 3
suollaas-ssoJ2
103 a[qeupqo
.Aqsuap
sa[oq 8-130
aJnldw
aql
aum[oA aql s! $y ‘(1)
aql Jo da%-p!m aql qly
LG.4 30 [aAa[ Waua
aql leql
EQunsse
pluenv
palaauuo:,
uroJ3 paugqo
asJaaaJ
log1
s!ql30
J!aql
pue)
swq MO[ s!ql u!
asodJnd
1uaJma .del
LlJaua
lurddnaao
aql s! 11
8 - .i aql a%ueJ
a8.q
0s qsnoua
30 suognq!Jluo3 s[alza[ r(8Jaua
sl[nsaJ
~U!A[OAU! saJn1 pue luamamseaur
(suo!ley~
e se pasn aq “~3 3[asl! aporp
aql spaaaxa
wqly
asaql
‘aporu se!q asJa.jaJ
apo!p %upol~uou~ luaJm3
sa!Jas se!q
“al30
%u~~ol~uou~ Jo3 a8ueJ
an[eh-q%q
e leql Moqs 01 Jaded uoyg!p
~uror~[i
sjsa3ap
aql 01 aso[a
f$~
= u = d uo!saJ
paumsse sarlwo[aA aqi
aJnldw
pw
aJe “3 pue d3 ‘sralua:, sJaluaa
30 uo!lez!Jalaenqa
paJa$unoaua [[curs dJaa r+euAp
uo!laun~-ud ‘luanna
~uoypuo~
snuy
u%!s sn[d aql aIaqm
aql 01 sagdde
.suogeJado
uopaunj-ud u!
sluaJJn2 Jols!saJ
‘l[nsaJ aql se!q (h
( >
(z)
‘aJo3aJaql ‘(ud),A leql ‘(%)
Bu!puocis~J~oa
‘suoqaa[a
-aadsaJ
!urJad ~!su!.y
uba UI ‘!s u! [aha[ sap!au!oa
(!u + d)yl+ (!u + u)S = n
(I)
‘[c-r]
Qensn
are
aql 30 Jalua:, SJaluaa
&US)
[[eH-p”a@a[yaoqs aql pa[[e:, K[[waua% ‘saymduq Jo/put! si3a3ap [aiza[-mo[ [enp!saJ 30 diallea e 01 anp sasy luauoduro3 8 - 1 aql sas!hap lsour UI .suo!%aJy[nq [erlnau aql u!ql!m luauodmo3 UO!Sngp aql30 pue ‘uO@aJ uoya[dap uoyunj aql u!ql!m u1oJ3 %U~XI!%!JO ‘luauod -wo3 (8 - 1) uo!leJaua%/uo!leu!qio~a~ aql 30 sysuo3 Qensn amlwadmal WOOJ le suo!pun+d UOX[!Su! luann3 aql weaur [euopahuo2 1(qpamseaw aq 01 MO[ 001 ual3o s! [aha[ luanna aql )eql JaqlJn3 pue suogmgdde r(ue ~03 pasn uIop[as s! apo!p aql a%um smq s!ql u! leql aq 01 maas s!ql 103suoseaJ a[q!ssod ~a8uw Aru m[ I aq] U!ql!M ‘K[Ja[n~pred‘a%J[OA SE!q MO[ )e W!lS!‘alXleq3 sq 01 ua@ uaaq seq uoyualle q3nw lou laql madda 1~la6 pue sacyap ~opnpuoyuas [[z %uomt2a.mlmJis pa!pnls K[a+ualxa pue 2!stzqlsour aqi s! aporp uoyu$-ud aqL
(r’” - Ud)"3'3'N ‘asLou I/i JO Jaqz!y aql JO [ahal a[qeldasge ue samsua
osle uogelado 1uaJma MO[ d[aluaJlxa wo$s!saJ leap! ]uale+ba u! ayou [euuaql aql ueql ura[qoJd e 30 aIom6ue asod IOU saop ‘uogelado 30 a%ueJ sy$ u! ‘aslou )oys apo!p aqJ .a%uel wq htu 001~ aql u~]!M sapo!p uogaunj-ud uoa!ps %ugelado dq pazgeal aq ue3 sJols!saJ qans leql umoqs s! 11 .smalsbs )uawamseatu hew u! safiueqr, I!aql pue sjuann:, [[elus %upo~!uom 103 pal!nbal uayo ale (u r,01-601 Qes!ddQ sJo]s!saI yueudp an[ekq%!H-_)awsqy
iilyayun
‘,Q!slayun
Y’S’n ‘ZO891 Vd ‘Wd aql ‘luaurlredaa Bu!iaau!%u~ [eagaa13
aleIs e!ueyhuuad
MOCINV~
‘N
3f)NVX SVI8 (bX’)P+ SXO,LSISBlI 3IiWNAtl
‘d
pue sva
‘8
‘dJo$eIoqe7 aay.aa
aleIs p!los
w
BHL NI SNOILL3NlWUd 9NISl-l Etfl?VA-H9IH d0 NOILVZI?VERI
M. B. DAS and P. M. SANDOW
308
respectively,
where
k is Boltzmann’s constant, T is absolute temperature, 4 is electronic charge, A, is the junction area, and X, and X,, represent the junction depletion width in the forward and reverse bias conditions, respectively. Using these equations, the diode differential conductance in the reverse and low forward bias conditions can be written as
dZ, dZ,o- L) m/g
Note that an empirical constant m is used in place of 2 which would result from idealized eqns (4). For practical junctions M is usually smaller than 2[4]. From eqns (6) and (7) it is apparent that when Z, = I, = 0, g, obtains its minimum value and g, obtains its maximum value. When Zr,,= 1.2pA and m = 1.65, this corresponds to a dynamic resistance of 3.5 X IO’Q If the diode is to be useful as a dynamic resistor in the low bias range, it is important that its noise performance should at least be comparable to that of pure resistors. Considering first the shot noise source, and the fact that noise associated with the reverse saturation current is uncorrelated[5] to the forward current, the mean squared equivalent noise current, for the bandwidth Af, can be written as -2
~,.sllot forward
= 4kTAf;
and
(11)
For the flicker noise the equivalent resistor is as given in eqn (10). The diffusion current in the planar structure arises mainly due to hole injection in the lightly doped n-region (ND = 2 x lOI cm-‘) and may be expressed in the manner
and
gr=dv,=
N, is the effective value of r-g centers representing a distribution of energy levels, (Y is a numerical factor smaller than unity, E is the absolute permittivity of silicon and f is the frequency. For the reverse bias case the conductance g, should replace g, in eqn (9). When the diode functions as a current monitoring resistor, as shown in Fig. I, its open-circuit noise voltage must be small compared to the measuring voltage. This can be represented in terms of equivalent noise resistor for the shot noise in the following manner:
(12) where
D,, is the diffusion
constant
and Lh is the effective
(g, + g,m,,,)
and
(8) iLotlreverse= 4kTAf;(g,-,,, _
+ g,)
where g,_,,. is the value of g, when I, = 0 and g,.,,, is the value of g, when Z, = 0. The flicker noise or l/f noise associated with the diode can be a serious limitation for its use as a resistor for d.c. and low-frequency a.c. measurements. At the low forward bias range the mean squared equivalent noise current can be written as[6] ‘2
L.nicker =
4kTAfM;
L -_..__1
where
(b)
R
_
F
’
@Ntx;,a
kT 8
Aif
’
(10)
Fig. I. Measurement of I/V characteristics in the low bias range: (a) forward bias and (b) reverse bias arrangements of the test diode D2.
309
High-value dynamic resistors usiqg pn-junctions diffusion length for holes in the n-on-n+ base region. In the low bias range the diffusion current is usually negligible compared to the depletion region r-g component of current. RESULTS AND
The point of intersection gives V,c = 0.415V and is related to m, I,,,, and Zd0,according to the equation
(13)
DISCUSSION
For the purpose of the measurement, p’n diodes D, and D, were selected and arranged in the manner shown in Fig. 1.The low bias test diode D2 was selected to be of larger area than that of D, which acted mainly as a current source in both arrangements (a) and (b). Measurements below lO_“A become rather difficult because of large time constants of the meters (Keithley 602 and 610) and high impedance of the test diode. The results of both forward and reverse current measurements are graphically shown in Fig. 2. The saturated behavior of Z, in the low reverse bias range provides the value of Z,0which can be regarded equal to Z& since both the forward and reverse bias voltages used were small compared to the built-in barrier potential (-0.8 volt). Using this value of Z,,, ( 1.2 PA), curve (2) was drawn to determine the value of m as indicated in Fig. 2, curve (3) was constructed to match rhe diffusion current eqn (10) at V, greater than 0.5 volts.
Figure 2 indicates m = 1.63, Z,, = 1.2 x lO~“A, and Ido = ’ 1 x lo-” A. _. Using the values of ZrO= 1.2 pA, Ai = 6.7 x lO~‘cm’, .I~, = 7.7 x 10m5cm, C, = 7 x IO-* cm’/sec in eqns (3) and (5), we obtain 7” = 9.75 x 10m6set and N, = 2.9 x 10” cm-‘. The value of C,, taken from published literature[7], is applicable to a high temperature process induced defect close to the mid gap. Thermally stimulated measurements were carried out to determine the nature of this particular defect[8]. Using the extrapolated value of the diffusion current from Fig. 2, when V, = 0, the effective value of minority carrier diffusion length Lb can be calculated according to eqn (10). This yields Lb = 50.5 pm which is quite realistic in view of the fact that the n-type base region consists of an approximately IO pm epitaxial layer on an n+ substrate. The noise performance and values of dynamic diode resistors are indicated in Table 1. The effects of shot noise in the diode are seen to be not significantly greater than that of the thermal noise present in ideal resistors of similar values. The flicker noise equivalent resistance was calculated assuming the value of fi, to be that of N, and the value of a = 1. In reality, the &-value that can be achieved is much smaller than that of N, identified in this investigation. Even with this exaggerated value of N,, the rms value of the l/f noise voltage for a frequency range of 1Hz to 10 KHz can be calculated to be approximately
2i
Fig. 2. Forward and reverse NV characteristics. Curve I: Forward current as measured; Curve 2: Extrapolated forware current; Current 3: Diffusion component of the forward current; Curve 4: Reverse current.
SSE Vol. 20 No. LC
/
Fig. 3. Forward and reverse I/V characteristics in the low bias range. Curve 1 is the forward current with representing theoretical results for m = 1.63 and A representing experimental results; Curve 2 is the extrapolated behavior of I,; Curve 3 is the diffusion component of the forward current; Curve 4 is the reverse current with -representing theoretical results for m = 1.63 and A representing experimental results.
’
M. B.
DAS
and P. M.
SANDOW
Table 1. Noise performance of dynamic resistors
an
5.‘h109
ii.9lslO9
40
1. 15x101°
1.53x10
20
2.16~10
10
?.R5xlO
IO
10
_. 40 PV which is negligible compared to the voltages that the diode is often required to monitor in measurement circuits. The assumption of an ideal l/f noise spectra applies only for an effective density of r-g centers occupying a distribution of energy levels within the band gap. In the case of a single level r-g centers, as identified in this investigation, the detailed nature of the lowfrequency noise spectra is somewhat different[9-111 and the value of the rms noise voltage for the same frequency range can be shown to be lower than the calculated value of 4opv. REFERENCES I.
C. T. Sah. R. N. Noyce and W. Shockley. Proc. IRE 45, 1228 (1957).
2. W. Shockely, W. T. Read and R. N. Hall, Proc. IRE 46. 973 (1958). 3. C. T. Sah. Proc. IEEE 55, 654 (1967). 4. P. Ashburn. D. V. Morgan and M. J. Howe\. Solid-St. Electron. 18, 569 (1975). 5. R. D. Thornton, D. Dewitt. E. R. Chenette and P. E. Gray, Characteristics and Limitations of Transistors. SEEC Vol. 4. p. 134. Wiley, New York (1966). 6. M. B. Da\. IEEE Trmc. Elwlrrm DPL. ED-Z 1092 (1975). 7. C. T. Sah and C. T. Wang. ./. .4[@. Phy\. 46. IX7 I IYi). X. P. Sandow, M. B. Da\ and J. Stach. I.tr\ \‘cxo\ j2fc,ctirr,v of r/w EI~c~rroc~hemid Sohfy. 17-22 Oct.( lY76). 9. P. 0. Lauritzen. Solid-St. Electron. 8. 41 (1965). IO. M. B. Das. fEEE Truns. Electron Droicrs ED-IY, 33X (1972). Ii. C. T. Sah. Proc. IEEE 52. 795 (1964).