scispace - formally typeset

Proceedings ArticleDOI

Laser induced single events in SRAMs

21 Mar 2013-pp 253-256

Abstract: This paper is aimed at emulating the errors in semiconductor memories by space radiation with a pulsed laser that acts as an ion. A sensitivity map of the memory is performed identifying potential error areas and how many errors simultaneously occur.
Topics: Laser (57%), Sensitivity (control systems) (50%)

Content maybe subject to copyright    Report

Laser Induced Single Events in SRAMs
C. Palomar
1
, I. López-Calle
1,2
, F. J Franco
1
, J. G. Izquierdo
3
, and J. A. Agapito
1
1
Dep. Física Aplicada III, Facultad de Físicas, Universidad Complutense de Madrid (UCM), 28040 Madrid (Spain)
(email: carlos.palomar, fjfranco, agapito@fis.ucm.es).
2
European Space Agency, ESA/ESTEC-TEC/QEC, 2200 AG, Noorwijk (The Netherlands) (email: isabel.lopez-
calle@esa.int).
3
Centro de Láseres Ultrarrápidos (CLUR), Facultad de Químicas, Universidad Complutense de Madrid (UCM), 28040
Madrid (Spain) (email: jegonzal@quim.ucm.es).
Abstact - This paper is aimed to emulating the errors in
semiconductor memories for atmospheric radiation by
pulsed laser that acts as an ion. A sensitivity map of the
memory is performed identifying potential error areas and
how many errors occur simultaneously.
Keywords Laser, memory, errors, SEU, MCU, sensitivity
map.
1. INTRODUCTION
Today Single Event Effect (SEE) characterization
of electronic components is carried out at heavy ion
and proton facilities. In recent years, SEE laser test
systems have become available to deviate irradiation
tests performed at particle accelerators. Laser SEE tests
are useful for component screening purposes and also
in conjunction with EEE component hardening efforts
(e.g. by detection of sensitive nodes). It has been
demonstrated that the pulsed laser technique can
reproduce SEEs in most EEE components [1 - 3]. The
test method is also less costly than irradiation tests
carried out at particle accelerators. This paper is aimed
to emulating the errors in semiconductor memories for
atmospheric radiation by pulsed laser that acts as an
ion. A sensitivity map of the memory is performed
identifying potential error areas and how many errors
occur simultaneously.
2. LASER FACILITY
The experiments were performed at the UCM
Ultrafast Lasers Center. The system is based on a
femtosecond pulsed laser source with pulse rate of 1
kHz. The laser system is flexible with the following
tunable parameters for laser irradiation tests:
wavelength, spot size, pulse energy, single shot, and
multiple shot. Typically for SEE laser test facilities, the
pulse width and wavelength are fixed parameters but,
at the UCM-Spain, the wavelength is variable and can
be chosen from ultraviolet (300 nm) to infrared (3000
nm) while keeping the same pulse width (60 fs). In this
case, the laser wavelength was set to 800 nm in order
to create free charge in the surface of the device. A
more complete description of the laser can be found in
[1].
3. EXPERIMENTAL SET UP
The laser radiation process requires very delicate
steps that are summarized below:
-Decapsulation and memory sizing. The laser radiation
requires direct lighting device, so that previously we
must perform its decapsulation. Decapsulation
techniques depend on the package or integrated circuit
packaging and you must choose the right solution.
Since the selected memories are plastic encapsulation,
the more effective technique for decapsulation
packaged plastic polymer is by etching. For the type of
polymer typically used in commercial encapsulated the
suitable acid is nitric red acid at a temperature of about
75-80 ° C. Because this acid attacks the polymer and
the metal, the decapsulation by the top of the chip is
much more destructive because, the acid also attacks
the metal contacts, pins and even integrated circuit
tracks. The device decapsulation is also necessary to
study the "lay-out" or integrated circuit design and
XYZ sized necessary to define the conditions of
illumination and laser scanning. The lay-out and
memory size of the memories AS6C6264 and
CY62256 are shown in Fig.1 and Fig. 2 respectively.
-Build test boards. After decapsulation and sized
integrated circuit, it is necessary to design a printed
circuit board for electronic configuration and
polarization levels. Figure 3 shows the printed circuit
board that has been designed and manufactured for
control of the memory and communication with a
computer for data collection.

Fig.1. Lay-out and size of the memory AS6C6264 (Alliance).
Fig.2. Lay-out and size of the memory AS6C6264 (Alliance).
The microprocessor (right) communicates with a PC
through a USB port (via RS232 serial converter wire to
USB). On the left, in a separate board which will be
subject to the sample holder or platform for the XYZ
movement front of the laser beam is positioned to
radiate memory, which communicates with the
microprocessor through an address bus and data.
Once built the circuit said microprocessor is
programmed to:
Write the memory with the desired pattern. The
program offers the choice between a large number
of patterns, such as: all zeros or all ones, or turn
or random ones and zeros.
Read the contents of memory.
Check the status of all memory cells, ie read all
memory locations, compared with the pattern
written in a file and store the result of the
comparison.
Transmit to PC memory status. The errors, if any,
and the address of the cells in which they were
produced.
Fig.3. Board for control of the memory and communication
with a computer.
-Test set up and data acquisition system. Figure 4
shows a graphic description of laser irradiation system.
After decapsulation and mounted on a printed circuit
board, the circuit with the memory is mounted on a
motorized stage in the three XYZ axes of movement
with an accuracy of 0.1 microns in its movement. The
laser beam is focused by a microscope objective and
long working distance, magnification 50X, suitable for
infrared light, achieving a diameter of "spot" on the
order of 1 to 1.5 microns for the wavelength used. The
laser spot location may be observed with an infrared
CCD camera to allow the correct placement of the
laser. The simultaneity of experiments in the CLUR
imposes a working frequency of the laser at 1kHz. This
means that at each point affect a few thousand pulses
per second. In principle, the working frequency is a
serious problem because to ensure that errors occurring
in the memory and that these are correctly accounted
needed that impact a single pulse on the memory, and
again not influence the next until the memory has been
checked, the result is stored and the memory has been
displaced to be irradiated in the following point. The
time spent in check all the memory and write the result
is greater than the time between pulses operating at a
frequency of 1 kHz, that is 1ms, so it is necessary to
ensure that impact on the memory only one pulse and
not affect the next until the memory is ready to receive
the next shot, that is, until the above actions have been
performed. To achieve it is necessary use of the
shutters, through which, and with proper configuration
of time so that both are open for 1ms, ensures that
actuate them can only pass a single pulse that will
impact on memory and not will actuate again until the
memory is ready again. Moreover, by using such
shutters is achieved that on each point impact exactly
the same radiation flux. Figure 5 shows a real image of
the system described. All elements are controlled by a
timing routine which is programmed by LabView.
This synchronization routine ensures that there is
enough time to check the memory, this is, read it and
write in a text file all information related to errors
(number of errors, direction and error occurred) before
the next pulse of energy arrives at the device. Thus, we
can perform a complete study of the device, both in
two or three dimensions.

Fig. 4. Schematic of the set-up implemented.
-Laser irradiation. The complete scanning of the whole
surface is performed following the state diagram shown
in figure 6. In more detail, by means of the program
developed in LabView memories are radiated in the
following sequence:
1. Memory is written to the selected pattern.
2. Is positioned the laser focus on a reference point for
taking after initial relative positions. For it is used the
lay-out of the chip (Fig. 1 and 2) properly calibrated.
3. From the reference point movement of focus will be
in constant steps in the Y direction and once the sweep
across that row returns at the beginning of it to forward
a step the same as above, but in the direction X. After
that the next row is scanned.
4. Radiation and reading errors are automatically
performed in the LabView program.
a. It acts on the shutters to achieve a single pulse on the
point which is positioned the focus of the beam.
b. It then reads all of the memory cells and compares
their state with the initially stored pattern.
c. Transmitted to PC the comparison result and stored
it in a file.
d. It reverts to original state the full content of the
memory.
e. You move the chip to achieve the next position and
starts a new cycle.
5. The results of radiation are analyzed once finished
the scanning sequence of all memory and are
superimposed on the photograph of the lay-out of the
chip.
4. RESULTS
During radiation laser process were detected several
types of errors such as SEU (Sigle Event Upset), MBU
(Multiple Bit Upset) and MCU (Multiple Cell Upset).
Note that during the process of determining the
radiation energy to CYPRESS memory was observed
that for higher energy values, from 170pJ, occurred
micro latch-up. Is interesting to note that these micro
latch-up does not break the memory, only produced a
malfunction recoverable restarting supply voltage.
Fig.5. Photography of the set-up implemented at the Centre
of Femtosecond Multiphoton Spectroscopy.
Fig. 6. State diagram for laser irradiation.
From text file result of the radiation we can perform
a sensitivity map of the memory, identifying potential
error areas and how many errors occur simultaneously.
The sensitivity map for memories ALLIANCE and
CYPRESS with a wavelength of 800 nm, an energy of
132pJ and 154pJ respectively, and written with the
pattern "01010101" are depicted in the following
figures (Fig. 7 and 8).
The sensitivity map gives an idea with a simple
glance of the sensitivity of the device and we can see
which areas are most likely to generate faults.
In both memories is observed two channels or slits
without errors that correspond to the two horizontal
lines seen in the layout of the chip (Fig. 1 and 2) and
probably are metal lines or areas where the logic is
implemented.

Fig. 7. Sensitivity map for the memory AS6C6264
(Alliance).
Comparing both sensitivity maps is noted that the
memory ALLIANCE presents sensitive zones more
uniform, whereas in the memory CYPRESS we can see
more more variety in sensitivity of the zones, that is,
extremely sensitive areas and high, medium an low
sensitive areas.
5. CONCLUSION
In conclusion, it is demonstrated that the laser can
induce errors in SRAM memories, both SEUs as
MCUs and even micro-latch.
The laser system of emulation space environment
becomes a tool for generating faults on electronic
devices of high interest in space electronics technology.
Fig. 8. Sensitivity map for the memory CY62256 (Cypress).
Moreover, counting all errors can be calculated the
cross section of the memory. Performing a more
exhaustive analysis of the errors caused and its address
is possible to know the topology of memory, that is, to
know the separation of the cells, interleaving ... So the
laser is also a very useful tool in reverse engineering.
References
[1] I. Lopez-Calle, F. J. Franco, J. G. Izquierdo, & J. A.
Agapito, "LASER System for Space Environment
Emulation", IEEE Spanish Conference on Electron
Devices, Palma de Mallorca (Spain), pp. 1-4, Feb 2011.
[2] F. J. Franco, I. Lopez-Calle, J. G. Izquierdo, and J.
A. Agapito, “Modification of the LM124 single event
transients by load resistors,” IEEE Transactions on
Nuclear Science, vol. 57, no. 1, pp. 358–365, Feb.
2010.
[3] F. Miller, "Interest of laser test facility for the
assessment of natural radiation environment effects on
integrated circuits based systems", 7th RADECS, pp.
199– 209, 2003.
References
More filters

Journal ArticleDOI
Abstract: The influence of a load resistor on the shape of the single event transients was investigated in the LM124 operational amplifier by means of laser tests. These experiments indicated that, as a general rule, load resistors modify the size of the transients. SPICE simulations helped to understand the reasons of this behavior and showed that the distortion is related to the necessity of providing or absorbing current from the load resistor, which forces the amplifier to modify its operation point. Finally, load effects were successfully used to explain the distortion of single event transients in typical feed-back networks and the results were used to explain experimental data reported elsewhere.

12 citations


Proceedings Article
01 Jan 2003
Abstract: This paper presents the EADS-CCR’s industrial laser test facility dedicated to the investigation of Single Event Effects. Applications to sensitivity precharacterisation of lntegrated Circuits are presented with a focus on SET tests of analog components.

12 citations


Proceedings ArticleDOI
07 Apr 2011
Abstract: The Sensors and Electronic Instrumentation Group of the University Complutense of Madrid has developed a system to emulate the cosmic radiations effects on electronic devices by LASER irradiations. Of great interest to the nuclear industry and space electronics, this project was carried out partially supported by the company ALTER SPAIN S.A., specialized in engineering, quality management and testing of high reliability electronic components for space applications.

5 citations


"Laser induced single events in SRAM..." refers background in this paper

  • ...A more complete description of the laser can be found in [1]....

    [...]


Network Information