Transcript Slides - Agenda INFN

Characterization of Electronic
Circuits with the SIRAD IEEM:
Developments and First Results
Luca Silvestrin
Physics dept, University of Padova – INFN Padova
Roma - 6 / 2 / 2012
RADIATION:
ubiquitous, problem, hazard, tool
NATURAL human
environment (all of us)
• natural radioactivity of
materials
• sea level cosmics
EXTENDED NATURAL
environment
ARTIFICIAL environment
• satellites (various orbits)
• HEP experiments
(collider halls)
• deep space missions
• radiation therapy facilities
• shuttle
• industrial accelerators and
sources
• high altitude avionics
• nuclear plants
accelerator environments
SCIENCE
MEDICINE
INDUSTRIAL
• High Energy Physics
• diagnostics (X-rays, PET)
• plastics
• structure of matter
(synchrotron facilities)
• artificial isotopes
• composite materials
• oncologic treatment
• semiconductors
• sterilization
• ecology
• materials science
• ...
• ...
Radiation effects in electronics?
Who cares? Why worry?
Use of electronics is pervasive:
 High Energy Physics
 Outer Space
 normal everyday life  all of us
Effects of radiation in
SCIENTIFIC EQUIPMENT
Particle
radiation
Ionizing &
Non-Ionizing
effects
Single
Event
Effects
material
Surface
degradation
Erosion
• Biasing of
instrument
readings
Degradation of:
• -electronics
• Data
corruption
Degradation of:
• Pulsing
• silicon sensors
• Noisy
Images
• thermal,
electrical, optical
properties
• solar cells
• System
shutdowns
• structural
integrity
Charging
• Power drains
• Physical damage
• optical
components
• Circuit
damage
direct effects in electronics
Single Event Effects SEE-ology
Description
Affected devices
SEU upset
Corruption of information
Memories, latches in logic
devices
MBU multiple bit upset
Several memory elements
corrupted by single ion
Memories, latches in logic
devices
SEFI functional interrupt
Loss of normal operation
Complex devices with built in
state/control sections
SET transient
Pulse response of certain
amplitude and duration
Analog, mixed signal devices
SED disturb
Momentary corruption of
info in a bit
Combinatorial logic, latches in
logic devices
SHE hard error
Unalterable change of state of a
memory cell
Memories, latched in logic
devices
SEL latchup
Generation of unexpected high
current
CMOS, BiCMOS
SESB snap back
Generation of unexpected high
current
N-channel power MOSFETs, SOI
SEB burnout
Destructive burn-out
BJT, etc.
SEGR gate rupture
Rupture of gate dielectric
Power MOSFETs
SEDR dielectric rupture
Rupture of dielectric layer
Non-volatile NMOS, FPGA, linear
devices
Single Event Effects
Energy lost by particle due to ionization
per unit length and density
2

1
dE
MeV

cm

 
LET

 

dx
mg


electr
.



Cosmic ray impinging
CMOS transistor
on
a
Ion strike on a DRAM
memory cell
The SIRAD facility of LNL
SIRAD beamline
SIRAD line schematic (line +70°
Tandem experimental hall 1, LNL)
7
SEU studies
The sensitivity respect to ionizing radiation of an electronic device is usually
described giving the measured upset cross-section as a function of the LET value.
2
cm

counts
fluence
SEU cross section for the pipeline
of the APV 25 chip for CMS
SEU sensitivity micro-map
CMOS 256K SRAM unit cell
SEU simulation
Davinci 3D-simulation, P.E.Dodd et al., IEEE
Trans.Nucl.Sci. Vol 48 pp1893-1903, Dec. 2001
Simulations performed for ion
strikes incident every 0.5 m
throughout the unit cell.
New sensitive area
for LET>30 MeV-cm2/mg
Evolution of the S. E. Upset sensitive area as
a function of the ion LET.
Nuclear microprobes vs IEEM
Nuclear Microprobe Analysis
Ion Electron Emission Microscopy
(Xhit,Yhit)
secondary
electrons
electron
optics
Secondary
electrons
emissive
surface
2D electron detector
at focal plane of
electron optics

relatively
broad
ion beam
analysis of
signal
target
Resolution on target determined by beam optics: spot size
and positioning. Difficult to micro-focus different heavy
energetic ion species.
Would have required disruptive and expensive upgrade of
SIRAD beamline
The ion impact position is determined by reconstructing
the position from which secondary electrons are emitted.
Analysis: SEE sensitivity mapping, time resolved IBICC (Ion Beam Induced Charge Collection)
Ion Electron Emission Microscopy
Annular Microchannel
Plate
Annular Phosphor
screen
45° mirror

A broad-heavy ion beam is sent onto the
Device Under Test (DUT).

The ion impact causes secondary electron
emission from a thin golden membrane which
is placed before the DUT; it may also induce a
Single Event Effect in the DUT

The secondary electrons are collected and
focused by a commercial Photo Electron
Emission Microscope onto a MCP

The secondary electrons are multiplied by
the MCP
SEE?
Device
Under
Test
Secondary
electrons
photons
Ion beam

30 nm
gold membrane
Electrons are converted into photons
by a phosphor screen

Electron
optics
Position sensitive
detector
Photo Electron
Emission
Microscope
Photons are extracted from
the vacuum chamber by a
mirror and focused onto a high
rate high resolution position
sensitive detector.
The axial SIRAD IEEM
On the focal plane of the IEEM,
a 2-stack MCP and a P47
phosphor screen.
The 200 µm diaphragm for
chromatic aberrations
Image Intensifier (2-stack MCP and
P47 phosphor)
STRIDE, a fast sensor and DAQ system designed for the SIRAD IEEM
12
The STRIDE system
STRIDE stands for Space and Time-Resolving Imaging DEvice,
it is the fast Data Acquisition System of the SIRAD IEEM.
It is able to acquire both the spatial and temporal coordinates of
every single light spot with a maximum effective rate of 1000
events/s (temporal resolution 100 µs).
STRIDE implements two linear
NMOS array, each composed of 256
photo receptors. In this way the
readout of only 2N pixels, instead of
N2, must be processed.
A beamsplitter provides 2 copies of the
light spot, each image is then squeezed
by a barrel lens to a blade-shape and
focused on one of the linear NMOS
detectors.
Phototube for fast
coincidence with
DUT (< 50 ns)
To improve speed, the pixels
readout of the two arrays is
processed in parallel by an
FPGA (Xilinx Virtex II).
The acquired data is sent via
a USB cable to the control
PC, that provides the
matching of the X and Y
coordinates of the detected
events
a) Beam splitter
b) PMT
c) Image Intensifier
d) STRIDE beam splitter
e) Squeezing optics
f) NMOS sensors
13
The IEEM chamber
Chamber size
68x41x36 cm3
14
(Exp. 1) The SDRAM system
SDRAM chip in the IEEM chamber
SDRAM metallization lines
15
SDRAM events
for 241 MeV 79Br (LET = 38.6 MeV-cm2/mg)
An ion impact upsets a cluster of memory cells
(in this case 4-6 cells on average).
We assume the best estimator of the ion
impact point to be the cluster centroid.
All SDRAM-detected
events of one acquisition.
The correlation
SDRAM events IEEM events
The red circles (radius 4 μm) are centered at the
positions of the centroid of nine SDRAM clusters.
The blue dots are IEEM events that are temporally
associated to the clusters.
Correlation peak
4 SDRAM centroids temporally
correlated with one IEEM event.
The histogram presents a Gaussian
shaped correlation peak above a
quadratic combinatorial background
 = 2.8  0.1 µm
18
(Exp. 2) The power MOSFET
Copper profile
reference point
Microphotograph of the metallization
layer of the of the power MOSFET
(detail). The Gate distribution line is
visible as the dark vertical line
The power MOSFET chip near the PIN
diode used for the positioning operations
Power MOSFET metallization layout
This experiment is fruit of a
collaboration with the group
led by prof. G. Busatto,
University of Cassino. The
power MOSFET we used was
developed by ST
Microelectronics; transistors
like this are commonly
implemented in DC-DC
converters used in satellites.
Internal structure of the power
MOSFET (detail).
The high operational voltage
makes it prone to radiation
effects.
19
MOSFET simulations
Mixed-mode circuit used for the finiteelements analysis.
The time dependence of the maximum electric
field for the three ion impact positions
The simulations have only been partially successful: they gave important results, showing that SEGR has a
strong dependence on the impact position of the ion, but they also proved to be limited by the great difficulty
to model the oxide layer and the phenomena that take place in it.
Drain current pulses caused by ion strikes
Current gate leakage of
irradiated power MOSFET
an
20
IEEM TRIBICC experiment
The gate polysilicon lines
appear as densely
populated stripes of over
threshold ion strikes.
Online sensitivity map makes it easier
the localization of points of interest in
the DUT
These first results demonstrate the
capability of the IEEM system to
provide in-deep information of the
structure of state-of-the-art electronic
devices and to study ion-induced
charge collection effects
Online sensitivity map
of the power MOSFET
The source metal contacts
are still discernable as areas
of small drain signals.
The separation
between the densely
populated bands reveal
a threefold structure of
the electrical field
below the gate
distribution line.
21
(Exp. 3) SOImager
CMS hybrid pixel sensor
Present CMOS technology cannot provide monolithic detectors
with radiation tolerance of the order of 100 Mrad (vertex tracking
regions of modern high energy physics experiments). The only
viable technology today available is the hybrid one.
CMOS technology can provide excellent detectors for any not so
harsh radiation environment (outermost layers).
Monolithic Active Pixel Sensor (MAPS) advantages
- Lower capacitance at the fully internal detection node => lower
power consumption of detectors => low material budget cooling
systems.
- Detector thinning down to some tens of microns is possible =>
reduced material budget of detector.
- Low production cost => extremely favorable granularity/price
ratio for large areas.
22
SOImager Shift Register
The SOImager layout schematic
The SOImager board
The SOImager is the prototype of a new
generation pixel detector for future high
energy physics experiments.
It is the fruit of a collaboration between Lawrence
Berkeley National Laboratories (LBNL), the Physics
Department of University of Padova, the National
Institute for Nuclear Physics (INFN) at Padova and
the University of California at Santa Cruz. The SIRAD
group in Padova is in charge of the testing and the
characterization of Total Dose tolerance and Single
Event Upset (SEU) sensitivity.
Shift Register schematic,
the pitch P = 13.75 µm
Shift Register cross section: Weibull fits
with Vbias=7V (top) and Vbias=0V (bottom)
23
SOImager SEU IEEM sensitivity map
a) Shift Register sensitivity map.
b) Shift register schematics
The actual resolution of the
IEEM does not allow us to
untangle the most sensitive
nodes inside the cell (we
cannot say which transistor
is responsible for an upset),
but it is sufficient to
distinguish the two FlipFlops and characterize their
relative sensitivity:
the sensitivity of the
Master Flip-Flop is 2.6 ±
0.1 times that of than the
Slave one.
Shift register cell schematics:
the two Flip Flop D are visible
24
Conclusions and prospects
The SIRAD IEEM is now a mature system and we have entered a new phase.
• We acquired experience with the SDRAM diagnostic system
• Original and interesting experiments were performed:
we tested a novel detectors in SOI technology;
we acquired the sensitivity map of a power MOSFET.
• A new IEEM experiment will be performed on Feb. 13th (next Monday) in collaboration with
the RREACT group (University of Padova, dept. of Electronics).
• New IEEM experiments are planned in collaboration with the Cassino group.
The best strategy to drive the future developments of the IEEM system is to make it a more
flexible, user friendly and hence attractive tool to users.
One way to make the IEEM system much more flexible is already under design. The ultra-thin AuSi3N4 membrane will be used to partition space into two volumes: one to always keep the IEEM
under high vacuum and ready to work; the other one to keep the DUT in a low vacuum condition
which could be achieved in minutes.
The IEEM system will now make the already successful SIRAD facility even more attractive:
SIRAD now offers both global and micrometric possibilities for a wide selection of heavy ions.
25
The end
Special thanks to the SIRAD group:
•
Dario Bisello
•
Jeff Wyss
•
Serena Mattiazzo
•
Devis Pantano
•
Mario Tessaro
•
Luca Silvestrin
26
Backup slides
27
Secondary Electrons Yield
79 Br
35
yield
240 MeV
~130 e-
35 Cl
17
yield
170 MeV
~60 e-


eV


Y

a
Z LET
0 
 
A

b

2
D. Bisello et at., “Secondary electron yield of Au and Al2O3 surfaces from
swift heavy ion impact in the 2.5–7.9 MeV/amu energy range”
Nucl. Instr. Meth. B, 266 (2008) 173.
28
Experimental: Secondary Electron yields from ion impact
“naive” Sternglass theory:
for a given material the yield is
proportional to the ion LET:
Yield =  LET
Hasselkamp: off Au target
Y  0.1  LET (eV/Å)
Radiation Damage on Semiconductor Devices
Impinging particle
ionization
TID Effects
(surface damage effects)
Single event upset
Single event latchup
Single event whatever
Non-ionizing
DDD Effects
(Bulk damage effects)
Single events effects
The gold membrane
The surface of a silicon integrated circuit is often passivated, hence
an unreliable secondary-electron emitter.
A ultrathin gold membrane (40 nm Au on 100 nm Si3N4) is used to
ensure a uniform and abundant secondary electrons emission.
E
E
The big Si3N4 membrane. (To avoid the bulge
provoked by the electrostatic pressure, the
membrane actually used in the IEEM is smaller:
0.5x0.5 mm)
(a) Former membrane configuration, with the electric field perturbation
provoked by the bulge. (b) Actual configuration: the electric field is uniform.
The IEEM membrane carrier (top view)
MCP dead spot
Shadow of the
biasing wire
Membrane focusing pattern: 16x16 squares (25
µm), separated by 5 µm wide gold strips.
31
Electric field perturbations
Topographic contrast in PEEM created by topographically
induced electrostatic field distortions
Secondary electron trajectories heavily distorted by the
presence of a 1 um wide step in the DUT surface
[C. M. Schneider et al, Rep. Prog. Phys. 65 (2002) R1785–R1839]
32
Membrane: pros & cons
Advantages



The device under test is not exposed to electric field of the IEEM (no
sparking; no E-field perturbations that distort image)
Protects DUT from UV-photons used to focus IEEM on a pattern. UVfocusing is easy and reliable (pattern is on the membrane frame and
hence on the same plane)
Suitable for any device (Ready to go!)
Disadvantages

Resolution degradation (proportional to the distance of the membrane
from the device under test).
33