1063-7397/04/3302- © 2004 MAIK “Nauka /Interperiodica”
Russian Microelectronics, Vol. 33, No. 2, 2004, pp. 99–105. Translated from Mikroelektronika, Vol. 33, No. 2, 2004, pp. 129–136.
Original Russian Text Copyright © 2004 by Chumakov, Gontar’.
There remain some uncertainties concerning the
hardness assurance of integrated circuits (ICs) in the
case of one or more ionizing radiation pulses of arbi-
trary width and shape. Existing approaches apply only
to speciﬁc pulse shapes and are based on linear models
of radiation response [1–3]. They offer no straightfor-
ward ways of evaluating model parameters. Moreover,
they ignore the role of load in radiation-induced failure.
In fact, they do not work if failure is caused by a num-
ber of factors.
The models currently employed are based on the
(i) Radiation response can be described by a linear
(ii) Every radiation-induced failure considered is
caused by a single process.
(iii) The failure threshold of irradiation for an IC is
connected with its radiation response.
(iv) The parameters of a transient response are pro-
portional to those of the photocurrent pulse.
It is arguable that none of the premises is true for the
whole range of practical radiation pulse widths. First,
recall that radiation-induced failures in modern ICs are
mostly due to rail-span collapse [2, 4, 5]. Second, a
number of concurrent radiation-induced processes are
observed in some ICs; examples are single-event
latchup and single-event upset in digital ICs [6, 7].
Third, it is often impossible to predict the failure
threshold of dose rate from photocurrent magnitude,
because the criterion of failure for a circuit element is
usually deﬁned in terms of voltage:
is the maximum total voltage increment
caused by a photocurrent pulse and
is the switch-
ing threshold voltage of the circuit element (the latter is
about 1 V for most digital ICs and varies from millivolts
to tens of volts for analog ones).
In a linear approximation, we shall identify major
types of response to pulsed ionizing radiation with a
view to extending the results to a nonlinear case. Let us
take a simple equivalent network of Fig. 1. Its radiation
response should depend on certain relations between
the time constant
of photocurrent decay, the radiation
, and the network time constant
By deﬁnition, the network will be regarded as a slow-
or a fast-response one if
For a given
, we consider three timescales on which
a radiation pulse may occur:
(i) a short pulse, with
(ii) a long pulse, with
(iii) a medium-width pulse, with
If we exclude medium-width pulses, the radiation
response can be expressed in analytical form in the four
remaining cases. For short pulses,
For long pulses,
Predicting the Failure Threshold of Dose Rate for ICs Exposed
to Pulsed Ionizing Radiation of Arbitrary Pulse Shape
A. I. Chumakov* and V. V. Gontar’**
* Specialized Electronic Systems (SPELS), Kashirskoe sh. 31, Moscow, 115409 Russia
** Central Institute of Physics and Technology, Russia
Received April 28, 2003
—A computer simulation is conducted of IC response to one or more ionizing radiation pulses of arbi-
trary shape and width. The failure threshold of dose rate is examined as a function of pulse width. A method for
estimating the failure threshold is devised and validated. It depends on experimental evaluation of the failure
threshold for a very short and a very long radiation pulse. The theoretical results are conﬁrmed by experimental
data on the failure threshold of dose rate subject to pulse shape and width.
RADIATION-EFFECT MODELING AND SIMULATION
IN SILICON MICROELECTRONICS