Improvement Of The Radiation Hardness of the npn transistors with ...

parkagendaElectronics - Devices

Nov 2, 2013 (3 years and 9 months ago)

80 views

P1

Bytkin



1

Radiation Hardness of the

-

irradiated up to D

=2x10
8

Rad Dielectrically Insulated ICs,
manufactured with Different Types of the Preliminary Radiation & Thermal Processing,
Measured in the Temperature Range
-
60

C…+125

C.

S.V. Bytkin,

Computer Automatical
Systems, 330091, Ukraine, Zaporozhye, p. b. 2069


Abstract

In the present paper the author reported, that RTP,

based on alpha
-

or electron irradiation, may improve bipolar
TTL IC radiation hardness up to 2 times at +25

C in the
case of

-
irradiation
(D

=10
8
, 2x10
8

Rad)
. RTP, based on
electron irradiation, may improve bipolar TTL IC radiation
hardness up to 5 times at +125

C. No improvement of the
bipolar TTL IC radiation hardness was achieved at
-
60

C
for the both types of the RTP techno
logies.


I
.
INTRODUCTION


The forecasting of the influence of the RTP on the
radiation tolerance of bipolar transistors after

-
irradiation,
was reported in [1].

RTP is the process of irradiation of wafers with IC
by the flow of

-
particles with energy

4
,5 МeV after
creation of the npn transistors by diffusion, but before the
deposition of bonding. After irradiation the isothermal
annealing of the wafers is carried out for the obtaining of


the relevant npn output transistors beta
-
current gain
values, h
2
1E,

ensuring reliable operation of the IC

After realization of the RTP, i.e. irradiation of the
transistors in the certain range of dozes and annealing of
the same devices in the certain range of temperatures the
transistors were exposed to the

-
irradiat
ion by the identical

doze
(


=10
7

Rad)
. In the other words the 2
2
full factor
experiment was carried out

for the forecasting

of the
influence of the RTP on the radiation tolerance of bipolar
transistors. The mathematical model, adequately
circumscribing r
esponse of the system (Y= h
21E
(


)/h
21E
(0))
was obtained. The change of the Y was calculated with use
of the experimental data for the
-
irradiated (
60
Co source,
1x10
7

Rad) npn structures at the variation of the

following conditions of the RTP mode: integral flow of
alpha
-
particles
, cm
-
2

and temperature of the
isothermal annealing
t
ann
.
,

C. The matrix of the planning
and the selected intervals of the variation of the factors are
adduced in

the Table 1.


Table 1

Matrix of the planning for the experiment of the 2
2

type for

the calculation of the radiation stability of npn structures

after RTP.

Level and interval of the
variation of the factors

Integral

-
particles flow,


,
cm
-
2

Temperatur
e of the isothermal
annealing, t
ann.
,

C.

Interaction of the
factors

Zero level, X
i
=0

2,5x10
11

300


Interval of the variation,


2,5x10
11

50


Lower layer, X
i
=

1

1x10
11

250


Top level, X
i
=1

5x10
11

350


Code label




Experiments: 1

-

-

+




2

+

-

-






3

-

+

-




4

+

+

+


For the investigation were used very simple IC
test npn structures with dielectric isolation, manufactured
by usual diffusion technology as shown in Figure 1.

P1

Bytkin



2

Advantage of the oxide isolation, received by
thermal oxidation, used instead of the process with
implanted oxygen to form the isolation layer is practical
exception of the leakage currents. Devices wer
e made on
CZ n
-
Si wafers with

=0.3 Ohm x cm, interstitial oxygen
concentration
N
oi

=7x10
17
cm
-
3
, carbon concentration
C
S
=2x10
16
cm
-
3
: The obtained samples were subjected to
irradiation by

-
particles from radioisotope source (the
surface of the source was l
ocated at a height of 1mm from
the test IC wafer) with an energy


4.5 MeV and to the
isothermal annealing in nitrogen atmosphere. Integral

-
particles flow



and the annealing temperatures were in
accordance with Table1, npn transistors h
21E
was
mea
s
ured

before and after

-
irradi
a
tion. Results of the Y=
h
21E
(


)/h
21E
(0) calculations are in the Table 2.



Figure 1. Vertical structure of the test transistor.


Table 2.

Experiments

Relative change of the h
21E

of ïðï IC test transistors after

-
irradiation

for the samples
after RTP ( in the accordance with the combinations of the factors of the Table1)


u

Y
1u

Y
2u

=(Y
1u
+Y
2u
)/2

1

0,385

0,383

0,384

2

0,476

0,469

0,473

3

0,464

0,452

0,458

4

0,474

0,467

0,471


At identical number o
f parallel experience on each
combination of levels of the factors the reproducibility of
the process was determined by the Cohren criterion:

,

where


is the dispersion, describing scattering of the outcomes
on
u
-
combination of the levels of the factors;

-

number of the parallel experiments;

is the
greatest value of the dispersions in the lines of the plan;

-

tabulated value of the criter
ion at 5 %
significance level;

is the number of the
independent assessments of a dispersion;

is number of the degree of freedoms
of each evaluation.


The process is reproduced, if the inequality

is fulfilled.

For calculation of G size, the outcomes of experiments,
listed in table 2,were used (Y
1u
, Y
2u
-
outcomes of the parallel
experiments). Let's calculate the dispersions
:
2x10
-
5
,
2,45x10
-
5
,
7,2x10
-
5
,
2,45x10
-
5
; G=0,5853658.


Tabulated value

(0,05;
4;
1)
0,9065;
0,5853658
(0,05; 4; 1)
0,9065, i.e. the process is
reproduced.


P1

Bytkin



3

The dispersion of reproducibility (error of the experiment) is

determined under the formula:

,
3,
075x10
-
5
.

By results of experiments the regression
coefficients were calculated under the following formulas:



The significance of the regression coefficients was
checked up with the help of the Student criterion by the
formula:

, where


is the 5%
-
s' point of a Student's distribution
with
f
y

degree of freedoms;

2,78 is the tabulated value,
0,0077079228;.


Low significance of the regress
ion coefficients at the
factors means, that the given factor does not influence or
influences not significantly on the response of the system.
In this case
0,44625 is significant,
0,02525

is significant,
0,018

is
significant,
-
0,019

is significant.


Thus, equation of a regression for the researched
process will look like:

Y=0,44625+0,02525
+0,018

-
0,019
;

The adequacy of an obtained equation is checked
up with the help of the Fisher criterion. The adequacy is
justified, if the following inequality is designed:

,


is the computational value of the response in u
-

experiment;

7,7086 is the Fisher criterion at
5 % significance level;


is the number of degree
of freedoms of the dispersion of a
dequacy;

4
-

is the number of degree of freedoms of a
dispersion of reproducibility.


The computational value of the response in u
experiment is on an obtained equation:
0,384;
0,4725;
0,458;
0,4705;
0. So,
0<7,7086 so, the obtained equation adequately
describes process.


For the record of an obtained equation in natural
variables the transition fr
om code to natural expression of
the factor implements under the formula:

, where


-
is the natural value of the factor;

-

is the value
of the i factor on a zero level;

-

is the interval of a
variation of i factor

;
.

In a final kind:



P1

Bytkin



4

The obtained equation allows present visually
(Figure.2) the radiation stability of a test IC transistor
struct
ure with dielectric isolation, made with application of
various RTP modes after

-
irradiation (
1x10
7
Rad
)
.


























Fig.2

Obtaining the value Y=1 confirmed, that the
application of the preliminary irradiation and a
nnealing of
the IC npn of transistors allows essentially increase

their
radiation resistance. Than above used doze of the
technological irradiation and the below temperature of the
annealing, the less response of a system to irradiation (Y is
higher) and,
therefore, the radiation stability of the
transistor is higher:

The author reported about the

practical
experience
of the RTP, based on the different types of
irradiation, use for the radiation hardness improvement of
the digital dielectric isolation (DI)
IC [2].


The improvement of the radiation hardness of the
ICs, manufactured with use of the RTP, based on

-



irradiation, was equal approximately three times better, than
in the case of the irradiation by the 6 MeV electrons.


The RTP efficiency for the
radiation hardness
improvement in the case of

-
irradiation wasn’t clearly
shown.

The purpose of the present paper

was, first of all,
the
comparison

of the effect of the different types of the
preliminary Radiation & Thermal Processing on the DI
bipolar I
Cs radiation hardness after high doses of

-
irradiation. Secondly, if the improvement of the radiation
hardness may be really achieved, the opportunity of the
new technology use must be proved for the real IC working
conditions in the wide temperature rang
e.




II.

O
UTCOMES
O
F

T
HE

IC, M
ADE
W
ITH
A
PPLICATION
O
F
D
IFFERENT
R
TP,
T
ESTS,
A
FTER

-
I
RRADIATION.


A. Details of the experiment
.



The author investigated the distribution of the
«logical zero» signal, Uol, on 12 outputs of six IC
before
and after



-
radi
ation from
60
Co source (D



1x10
8
, 2x10
8

Rad) of the simple logic DI TTL IC 4NOT
-
AND,
manufactured by the technological process with RTP
(



8x10
10

cm
-
2
, t
ann

350

C, annealing duration



20 min,

e

8x10
15

cm
-
2
, t
ann

350

C,



90 min). The measurements
wer
e carried out at the temperatures +25

C, +125

C,
-
60

C.
Electrical modes of the U
OL

measurement, the logical
diagram of the tested ICs was the same as described in [2].

Used RTP mode was far from the optimum from
the point of view of the obtaining transis
tors peak radiation
stability, but it has allowed use of RTP without change of
the existing process of diffusion during IC manufacturing.

P1

Bytkin



5

Experimental data were processed by [3]. Obtained
histograms were fitted by well
-
known

distribution density
function,

but in the case of the two extremums the
superposition of functions was used. For example,
experimental data for the

-
irradiation based RTP
technology ICs set after

-
irradiation with D

=1x10
8

Rad,
measured at +125

C were described as in Fig. 3.



Fig.3


B. Experimental results
.



In the Figure 4 the 4AND
-
NO IC U
OL

distributions
after

-
irradiation (D

=10
8
, 2x10
8

Rad) at +25

C are adduced.
The radiation hardness improvement for the RTP
technologies, based on

-
irradiation ma
y be roughly
estimated, in the comparison with the standard ICs, as the
relative shift of the distribution functions peaks: ((a4

a1)/a1)/((a3

a2)/a2)

2.4 (times), where a1, a2, a3, a4 are Uol
values, at which peaks of the distributions are observed.
The sa
me for the electron based RTP is 2.015 times. So,
with the accuracy, enough for the practical use, it may be
considered, that at +25

C both used RTP modes improves
radiation hardness 2 times.
P1

Bytkin



6


Fig.4



Fi
g.5

f4

f1

f6

f5

f3

f2















92

f3

f1

f4

f2

f5

f6

P1

Bytkin



7


In the Figure 5 the 4AND
-
NO IC U
OL

distributions
after

-
irradiation (D

=10
8
, 2x10
8

Rad) at +125

C are
adduced. The radiation hardness improvement for the
electron based RTP technology, may be approximately
estimated, in the comparison with the stand
ard ICs, as the
relative shift of the distribution functions peaks: ((a3

a1)/a1)/((a61

a5)/a5)

5.54 (times), where a3, a1, a61, a5 are
Uol values, at which peaks of the distributions are
observed.

Evaluation of the same for the

-
irradiation RTP
may be ca
rried out only qualitatively. The total U
OL
distribution for this RTP mode may be fitted (D

=10
8

Rad)
by the superposition of two Laplace functions:



where


Increasing of the TID results in obtaining of the
“mathematically more cor
rect” extreme distribution:


where


Appearance of the distribution “tail” indicates on
the possibility for the certain part of the ICs set to reach
the same Uol value as for the control set.

The situation with th
e RTP, based on the electron
irradiation is vice versa. Extreme distribution



changes by




Increasing of the left peak of the f6 distribution
indicates on the possible imp
rovement of radiation
hardness for this type of the technology. So, with the
accuracy, enough for the practical use, it may be
considered, that at +125

C used electron irradiation based
RTP mode improves radiation hardness 5 times.

In the Figure 6 the 4AND
-
NO IC U
OL

distributions
after

-
irradiation (D

=10
8
, 2x10
8

Rad) at
-
60

C are adduced.

Obtained results were unexpected. After D

=10
8

Rad radiation hardness of the ICs, made with use of the
RTP were less, than the same for the standard ICs. But the
TID in
creasing followed by the tendency for the improving
of the distribution form for the ICs set, manufactured with
the RTP.

C
ONCLUSIONS
.


1

RTP, based on alpha
-

or electron
irradiation, may improve bipolar TTL IC radiation
hardness up to 2 times at +25

C.

2

RTP, based on electron irradiation, may
improve bipolar TTL IC radiation hardness up to 5
times at +125

C.

3.

No improvement of the bipolar TTL IC
radiation hardness achieved at
-
60

C for the
both types of the RTP technologies
..
P1

Bytkin



8



Fig.6


R
EFERENCES


1.

S.V. Bytkin
,
Improvement of the radiation
hardness of the digital bipolar IC with dielectric isolation,
manufactured in accordance with the RTP technology,
Components fit for Space Seminar (17
th

February 1999)
Proceedings, Royal M
ilitary College of Science,
Shrivenham, Swindon, UK, pp. 99
-
108.

2.

S.V. Bytkin, Comparison of the Radiation
Hardness of the Dielectric Isolation IC's Made in
Accordance With Different Types of the Preliminary
Radiation & Thermal Processing. Military and
Aerospace
Applications of Programmable Devices and Technologies
Conference (MAPLD 1999) CD
-
ROM Proceedings,
September 26
-
28, 1999, The Johns Hopkins University
-

Applied Physics Laboratory, Laurel, Maryland, US

3.

StatSoft, Inc. (1995). STATISTICA for Wind
ows
[Computer program manual]. Tulsa, OK: StatSoft, Inc., 2325
East 13th Street, Tulsa, OK 74104, (918) 583
-
4149, fax: (918)
583
-
4376.

f
5

f2

f3

f6

f4

f1