Slides - Agenda

PixFEL
Enabling technologies, building blocks
and architectures for advanced Xray pixel cameras at FELs
Presentazioni CSN5
INFN Sezione di Pisa, 1/7/2014
G. Rizzo
PixFEL – Pisa July, 1st 2014
1
Outline
•  The PixFEL project
•  Motivation
•  PixFEL Long Term Purpose & 3–years Activity Program:
•  Building blocks, Technologies, Architectures
•  Pisa contribution
•  Present achievements
•  Activity in 2015
–  Piani
–  Richieste finanziarie e ai Servizi di Sezione
G. Rizzo
PixFEL – Pisa July, 1st 2014
2
The PixFEL project
• 
• 
• 
Use innovative solutions & technology, now explored in HEP comunity (our
background!), to improve performance of pixel device for photon science.
Important synergy with other activities of the groups/Sez. INFN involved: LHC
upgrade (65 nm, active edge sensor), AIDA (3D vertical integration).
In the long term (+6 years) develop a four side buttable module for a large area
X-ray camera for application at FELs. INFN PixFEL project as a first step.
Goal of the 3 years INFN project: investigating the technologies,
designing the fundamental microelectronic building blocks and
exploring the readout architectures for high performance X-ray
imaging instrumentation to be be used in the experiments at the next
generation free electron laser facilities
• 
active edge pixel sensors, low density TSVs
• 
65 nm CMOS technology for front-end and readout electronics
• 
in pixel data storage and readout architectures
Participating INFN groups
• 
INFN Pavia (3.7 FTE; resp. naz.: Lodovico Ratti, resp. loc.: Massimo Manghisoni)
• 
INFN Pisa (2.4 FTE; resp. loc.: Giuliana Rizzo)
• 
INFN Trento (2.9 FTE; gruppo collegato, resp. loc.: Lucio Pancheri)
G. Rizzo
PixFEL – Pisa July, 1st 2014
3
Gruppo PixFEL
Foto al 2014, piccole
variazioni nel 2015.
!"#$"
!"#$%&%'()*)++$
56"#7%97)'!%'("#$)
=)6%#>)'5"?6$9
D"67)'-6"99$
A$%6)'D"&7)."+$
D"99$*)'D"#0B$9)#$
=)H).$7)'I"++$
G"&%6$)'I%
-$"#&@7"'J6".%69$
,
12:
12:
123
128
123
12:
12,
123
,-.
!$%"
-$)."##$'/"+$0#"#$
;+%<"#)'/%++"6$#$
-$@&$"'("9"6)9"
56"#7%97)'5)6+$
D"67%&&)'-$)60$
E@0%#$)'A")&)#$
-$@&$"#"'I$>>)
5"?$)'D)69"#$
123
123
128
123
&'()*+
-$"#456"#7)'!"&&"'/%++"
-$)60$)'5)#+"#"
=@7$)'A"#7B%6$
EF"+%6$#"'A"#$#"
-$)."##$'G%6>%&&%9$
123 Xu Hesong
12:
12:
/-0
&3&456
9
In rosso i responsabili locali e
coordinatori dei 3 WorkPackages
128
12:
12C
12:
1
123
0.5
2.9
1-2
7
• WP1: Enabling technologies: Lucio Pancheri
• WP2: Building blocks: Massimo Manghisoni
• WP3: Architectures and testing: Giuliana Rizzo
G. Rizzo
PixFEL – Pisa July, 1st 2014
4
Motivation
• 
• 
• 
X-rays have been a fundamental probing tool
in many fileds since their discovery
Synchrotron sources extended the application
field, giving relatively “high intensity” X-ray
pulses at high repetition rate.
The advent of free electron laser (FEL)
facilities, with coherent X-rays beams of very
high intensity, ultrafast pulses (100fs),
available in a large energy range, opens up new
possibilities.
– 
• 
Broad science program accessible at FELs:
– 
– 
– 
– 
• 
~9 order of magnitude increase in brilliance w.r.t
previous synchrotron sources
Structural biology
Chemistry
Material science
Atomic and molecular science (AMO)
Many measurements are based on scattering of
coherent X-ray and detection of diffraction
pattern with pixel detectors.
To fully exploit X-ray FELs beam properties, new detectors need to
be developed with extremely challanging requirements!
G. Rizzo
PixFEL – Pisa July, 1st 2014
5
->/,(.0(#B%"*'.#0(#"(+0D%"(&#0,'"+&'.#0
FELs in
operations or under construction
!"#$%&'(
)'*'+,
-.",'(
/*,.01
2%3
ʄ4.0
5*.0(/.0*&(
'%&60#7#18
9:%"*77(
7%01'6
-/;)<
!"##$#%
&''(
)&'''*++,-
./&*012
3(**4
5"6718*9:
./;*0<=
;.(*>
/=/)
!"##$#%
&''?
.3/*;012
./(*4
5"6718*@:
&/A(*0<=
.B''*>
->?5@A>/>22?;
CD#7E!"CE$D#
running
&'.'
./(*012
3'*4
5"6718*@:
;/'*0<=
;B(*>
SACLA@RIKEN
)=))
CD#7E!"CE$D#
running
&'..
A*012
.*4
5"6718*@:
(/B*0<=
B('*>
>+"#B%*0(C->/
CD#7E!"CE$D#
&'.(
.B/(*012
їϭϰ'Ğs
.*4
їϬ͘ϱ
5"6718*9:
./;*0<=
;3''*>
After the success of first FELs, new FELs facilities & upgrades
are in design phase in many countries.
Beam-line and beam-time structure
Beam-line structure @Eu-XFEL
•  Beam lines with different photon
energies available at each facility
Very different beam structure from one
FEL facility to the other
•  Each pulse is always very short
• DBE6G>HDCL>I=:M>HI>C<A><=IHDJG8:H
LCLS: continuous operation @ 120Hz
•  XFEL: 220ns spacing, with time to readout
•  0D96NSHHIDG6<:
Future (i.e. LCLS II): continuous 1MHz spacing
3:6@EJAH:H6I
G>C<MG6NHDJG8:H
PC6CDH:8DC9H
=><=G:EG6I:
-##$$)!
P E>8DH:8DC9H
Most challenging
for detectors are
XFEL & LCLS II, taken as target for PixFEL
Eu-XFEL
Today: LCLS, …
0D96NSHMG6N
A6H:GHDJG8:H
PB>AA>H:8DC9H
-"!!$)!
Z
%CI:CH:EJAH:H6I
ADLG:EG6I:
Z
P ;:BIDH:8DC9H
Tomorrow: NGLS/LCLS II
PB>8GDH:8DC9H
- "&$$)!
0DBDGGDLSHMG6N
A6H:GHDJG8:H
G. Rizzo
%CI:CH:EJAH:H6I
=><=G:EG6I:
P 6IIDH:8DC9H ID ;:BIDH:8DC9H
PixFEL – Pisa July, 1st 2014
7
µ
µ
µ
2. Progress on the detector development program at the European XFEL
The different scientific applications at XFEL.EU will be mainly addressed by the imaging detector
development projects shown in table 1. The first column summarises the requirements of the different
experimental stations, the other columns the specifications of each detector. In more detail the actual
status of the DSSC [6] (hexagonal pixel), the AGIPD and the LPD detector projects are as follows:
XFEL 2D imaging detectors
Table 1. Overview on 2D imaging detector development projects at the European XFEL. All
detectors are based on Si sensor technology and will provide a spatial resolution close to the pixel
size. All detectors are actively cooled and foreseen for vacuum operation, except the LPD.
Requirements AGIPD
Hybrid pixel
10...100’s µm 200x200 µm2
1kx1k
1kx1k
Central,
Multiple tiles,
variable hole variable hole
>80%
>80%
DSSC
Hybrid pixel
Technology
204x236 µm2
Pixel size
1kx1k
Detector size
Multiple tiles,
Tiling, hole
variable hole
>80%
Quantum
0.5-13 keV
efficiency
500 µm
450 µm
Sensor thickn.
0.5-25 keV
Energy range 0.25-25 keV 3-13 keV
3
4
4
Dynamic range 10 … 10 … 10 at 12 keV 104
Single photon 300 el. rms
50 el. rms
Noise
4.5 MHz,
4.5 MHz, 352 4.5 MHz, 640
Frame rate
2700 images, images anaimages digital
10 bursts/s
logue on-chip on-chip
LPD
Hybrid pixel
500x500 µm2
1kx1k
Multiple tiles,
variable hole
>80%
1-13 keV
500 µm
1-25 keV
105 at 12 keV
1000 el. rms
4.5 MHz, 512
images analogue on-chip
pnCCD*
FCCD
CCD
CCD
2
75x75 µm
30x30 µm2
256x256
1kx2k
Monolithic Monolithic
no hole
fixed hole
>80%
>80%
0.3-13 keV 0.3-6 keV
450 µm
200 µm
0.05-20 keV 0.25-6 keV
103 at 12 keV 103 … 104
2 el. rms
25 el. rms
200 Hz,
200 Hz (1kx1k),
continuous continuous
(*prototype)
DSSC: First experimental results have been obtained by coupling the non-linear DEPFET prototype to
an ASIC prototype comprising the complete readout chain from the analog front-end to the ADC and
• Fastgain curve is well controlled
• Slow
the memory. Measurements show that the non-linear
by the applied
G. Rizzo
PixFEL – Pisa July, 1st 2014
voltages, which is an important milestone for further calibration studies. Radiation hardness generally
9
DSSC (DEPFET Sensor with Signal Compression) X-ray camera
• 
Dynamic range 1-104
ph/pix + single photon
resolution
• 
Single shot imaging in
200 ns
21 cm (1024 pixels)
•  1024x 1024 pixels
Pitch 200x240 um2
• 
600/2700 frames in pixel
storage
•  16 ladders/hybrid boards
•  32 monolithic sensors
128x256 6.3x3 cm2
•  DEPFET Sensor bump
bonded to 8 Readout
ASICs (64x64 pixels)
x-y Gap
Try to improve with
new technologies
G. Rizzo
• 
PixFEL – Pisa July, 1st 2014
•  2 DEPFET sensors wire
bonded to a hybrid board
connected to regulator
modules
•  Dead area: ~15%
10
Problem of missing data
Modules of limited size and gaps
between modules è lots of missing data
Reconstruction may become ambiguous
•  Fig. a) b) are 2 single shot diffraction pattern
images of a large virus, 0.75 um diameter,
taken with pnCCD detector at LCLS.
•  Fig. c) is the virus picture (30000 images
averaged) with a transmission electron
microscope as a comparison.
•  All the various possible reconstructed images
of the virus are shown in f) and g) with all the
ambiguities coming from: missing data due to
dead area and saturation of pixel in the
central region.
st
G. Rizzo
PixFEL – Pisa July, 1
2014
11
PixFEL long term goal
•  Develop a four-side buttable module for the assembly of large area detectors
with no or minimum dead area to be used at FEL experiments
•  Multilayer device: active edge thick pixel sensor, two tiers CMOS readout
chip (analog+digital/memory) with low density and high density TSV, with 65
nm to increase memory and functionality, in a smaller pixel pitch of 100 um.
(450 um thick)
Good
efficiency up
to 10 keV
wide dynamic range
(1 to 10000
photons), single
photon sensitivity
burst and
continuous
mode operation
9 bit resolution
(effective), 5 MHz
sampling rate
1 kframe
PixFEL target requirements
à criticità ed innovazioni
1. 
Elevato range dinamico, 1-104 fotoni (1-10 keV) + single photon resolution!
à Preamplificatore a compressione di dinamica
1. 
ADC entro 200 ns
à Successive approx ADC 10 bit: buon compromesso tra frequenza e risoluzione
1.  Tiling senza zone morte
– 
• 
– 
2. 
à Slim-edge sensors sviluppati per Alice, Atlas (FBK)
BUT for FEL need to be thick and operated with high bias to reduce plasma effect. Specific optimization needed!
àlow density TSV to connect I/O chip PAD to hybrid board
Pitch: 100um
– 
3. 
àTecnologia 65nm per aumentare la densità
Readout e memoria: 1k frame depth
– 
4. 
àUso di TSV per espandersi in profondità, 2 layer readout chip
DAQ e banda di uscita
– 
• 
• 
à Critici per macchine continue ad alto frame rates (>15 kHz) ma non abbiamo soluzioni al
momento
Criticita’ dei punti 1-4 studiate nel 2014
Soluzioni proposte vengono provate nelle sottomissioni di Autunno 2014
(chip 8x8, active edge sensor).
– 
G. Rizzo
Dettagli nelle slide successive.
PixFEL – Pisa July, 1st 2014
13
PixFEL Activities
2014
•  define chip specifications.
•  design of test structures with single blocks (analog front-end, ADC, circuits for gain
calibration, single MOS capacitors, I/O circuits), CMOS 65 nm
•  design of a 8x8 matrix, 100 um pitch
•  design of 1st batch slim edge pixel sensors
•  start investigation on readout electronics
2015
•  test the structures from the first run
•  start investigation on 3D integration processes, including low density TSVs
•  design of the 32x32 matrix (accounting for low density TSVs)
•  design 2nd batch slim edge pixel sensors
•  interconnect the front-end chip to the (slim edge) pixel sensor
•  start writing VHDL and design some elementary digital block (memory cells, buffers)
•  start organizing the test beam
2016
•  test the 32x32 front-end chip
•  test the chip after interconnection with the detector
•  test structures including low density TSVs
•  complete VHDL design of the readout electronics
•  test the chip with beam
G. Rizzo
PixFEL – Pisa July, 1st 2014
• 14
14
Attività a Pisa (2014-15-16)
•  Collaborazione alla definizione delle specifiche
•  Contatti con le facilities FEL
•  Progetto logica in-pixel e architettura readout
‘semplificata” per i 2 chip prototipo singolo layer
(8x8, 32x32)
•  Collaborazione progetto ADC SAR con PV
•  Interconnessione chip-sensore
•  Progetto e costruzione schede di test
•  Caratterizzazione in lab dei vari chip prototipo
•  Contributo al progetto del chip digitale per
prototipo 2 layers (3D)
•  Partecipazione al test su fascio (LCLS ?)
G. Rizzo
PixFEL – Pisa July, 1st 2014
15
Dove siamo arrivati
Sottomissioni chip e sensore autunno 2014
G. Rizzo
PixFEL – Pisa July, 1st 2014
16
Active edge sensor optimization for FELs (1) TN
Field plate
5 µm
•  Active edge sensor to minimize dead area between
tiles. Key steps:
• 
trench etching/polisilicon filling/support wafer removal
•  Already under development by FBK for Alice and
Atlas, but specific optimization for FEL needed.
•  Main difference/issues:
1. 
n+
n/p gap
p+
!"#$%&'()*+"',(-(./&%0"(.1''".$*1#(
Trench edge
104 photons at both 1keV and 12 keV
n-­‐ substrate
Collection time < 30ns for both 300um and 450um thickness
at Vbias = 400V
Plasma effect: high charge concentration from huge signal
(10^4 photons 1 keV à 2.8x10^6 e-/h>100 MIP) reduces the
collection field and deteriorate charge Collection Distance
(CD) and Collection Time (CT)
2.  Thick sensor (450 um) needed for good efficiency @ 10 keV
model available within the Synopsys TCAD package. In
à  High bias voltage needed but breakdown voltage might be
particular, the light wavelength (1015 nm) and intensity have
critical for active edge sensor!
been tuned to emulate 104 X-ray photons of 12 keV energy, an
than 3
that a
very
approach that has already been proposed in [12]. Since the exper
pixels will be bump bonded to the read-out chip, the X-rays which
(i.e., the light pulses) are impinging from the back-side.
(~101
à  High Bias voltage mitigates plasma effect: with Vbias=400 V
increa
PixFEL phone meeting ± June 19, 2014
break8
can reach our target design à coll dist < Pitch=100 um
cm-2
collection Time~30 ns.
that is
In
–  Edge geometry optimized to increase edge breakdown voltage:
region
rings
•  Edge distance & floating guard rings,
As an
•  field plate & oxide thickness
differ
•  Vbreak > 400 V obtained even after 1GRad (Qox=3x10^12)
one f
floati
oxide
G. Rizzo
PixFEL – Pisa July, 1st 2014
17200 V
•  TCAD simulation results:
!"##$%&'
Active edge sensor optimization for FELs (2) TN
!"##$%&'
ƒ !"#$%&'(")*&+,%$+-$.&/0"$1+2#02$%*2(3*(20$4&*56$$
•  Present optimization
of the edge geometry to improve breakdown voltage:
‡ 78('$9).$:(')*+,-.'-/.0'10*400,$'+#("0%;$$
‡ <$9()2#$2&,9%$4$0/*02,)"$-&0"#$.")*0$
ƒ !"#$%&'(")*&+,%$+-$.&/0"$1+2#02$%*2(3*(20$4&*56$$
!"#$%&'$()"*+,%ʹ%*-."$%/0('#$%
‡ <('$=(,3*&+,$#0.*5$
‡ 78('$9).$:(')*+,-.'-/.0'10*400,$'+#("0%;$$
‡ <88,'$+/&#0$*5&3>,0%%$
‡ <$9()2#$2&,9%$4$0/*02,)"$-&0"#$.")*0$
%$122,3$&$'()*$)"+,*-".$/"0$+102($3"#$1.4$56$7$8,9$
‡ <('$=(,3*&+,$#0.*5$
&*$&%$.+%%&1"0$*+$+1*)&,$1
2'3'4551'-+2$)""$*50$#0*03*+2$+.02)*&+,$"&-0*&'0$
1;4"<.$="+*12($7$>".)*1.*$*?0",2?$*?($(.*-0($4(*(>*"0$
:')/$?+/$@$A0BC$3'‡ DC;$<88,'$+/&#0$*5&3>,0%%$
•  First pixel at 90 um from the edge
01*-".$+-/(*-9($
&*$&%$.+%%&1"0$*+$+1*)&,$1
$
2'3'4551'-+2$)""$*50$#0*03*+2$+.02)*&+,$"&-0*&'0
~1% dead area
:')/$?+/$@$A0BC$3'DC;$
~ 1 mm in AGIPD detector for XFEL
$
$
$
$
$
$
$
90 um
PixFEL phone meeting ± June 19, 2014
2
PixFEL phone meeting ± June 19, 2014
ting ± June 19, 2014
G. Rizzo
5
PixFEL – Pisa July, 1st 2014
18
with a suitable design of the MOS channel length L (once
mic range into a reasonable outputMassimo
signal Manghisoni
W has been properly chosen for the low energy gain setnon-linear characteristicon
is required.
Sig- PixFEL
behalf of the
Collaboration
can be achieved at sensor level, as in the ting). The simulated input-output trans-characteristic of
C device [1], or front-end level, like in the the circuit is shown in Fig. 2 (right) where the CSA output
dynamic
range
front-end
channelis W/L
intended asVddthe difference with
respect to the
work
presents
the
designsolution,
of a low-noise
]. In Wide
this work
a novel
based readon voltage
W/L
ited for application
to the next generation DC value. The inset shows the initial, high gain portion of
4 photons
atures
of
a
MOS
capacitor
[3],
is proposed
• 
1-10
ser (FEL) experiments andbetween
based on a 1-10
novel keV
theEnergy
characteristic.
The
dimensions have been choGain (G
setting
1. The
drain and
source
terminals
of theHigh
he )device
sion
solution.
The
readout
architecture,
gnd
•  out
Bilinear
Amplifier:
use the is
non-linear features
Low
Energy
Gain
(G
) setting
lecapacitor
arried
intoa form
65
nmone
CMOS
technology,
ed
together
termi- sen to have a low energy gain of about 0.4 mV/ph and an
Vout amplifier of 500 mV. With
Vout
MOSFET
capacitors
to dynamically
change
low-noise
charge
sensitive
amplifier
(CSA)
output
range of the
Q in
Q in dynamic
ate
formsof
the
other.
The
device
polarity
oo
oo
signal compression, a time-variant shaping respect to other
R compression
solutions underR investigation
the
gain
with
the
input
signal
amplitude
detector
characteristics:
a
PMOS
for
a
w-power analog-to-digital converter (ADC)
for 100
FEL
experiments,
the one proposed here has the adng
electrons
andThe
an
NMOS
for
• pixel.
High
gain
for low
energy
&colLow gain
for high
energy
µm-pitch
blocks
will
beone
operated
- technology PMOS @
PMOS
@ V =1.2on
V; V a
=200
mV
vantage
of
being
based
standard
CMOS
othe
cope
with
the
high
frame
rate
(exceeding
voltage of the gate connection Vin is
in
out
in
out
100
n for future XFEL machines. The work has (not on a customized
L=5 µm process, as in the case of the DSSCL=5 µm
close
to
gnd
for
PMOS
and
close
to
V
DDapIf Vout
<<Vthof
⇒ the
Cgs =PixFEL
Cmin
L=20
µm Cgs = Cmax
ut in the
frame
project
Ifdevice)
Vout >Vand
th ⇒
on a single channel with dynamically changdevice
is
operated
in
inversion,
yielding
stituto Nazionale di Fisica Nucleare (INFN)
1. (not
Charge
sensitive
preamplifier
with of
MOS
nonlinear
gain
on the
parallel
configuration
channels
withfeed2014. The
paper
will 2discuss
detail
teristic
shown
in Fig.
(left). inSince
lowthe ingFig.
C
≈
C
+
C
=
WLC
max
gs
gd
OX
Cmin ≈ Cgsovl + Cgdovl = 2W ∆LCOX
back capacitor: PMOS for detector collecting electrons (left)
ted
for the
the channel
and is
will
W=10 µm
generate
an design
outputofvoltage
Vout which
10 and NMOS for detector collecting holes (right).
ugh analysis of the simulation results.
'"#
&!!
n the device Vg=Vin=+0.8V
threshold
Vth1 , the1 feedback
1
1
'(
G
=
280q
W=20
µm
he
AB-*,C1D,
le = 280q
is set at its Gminimum
andOX')itWis mainly
COX WL
2∆LC
#!!E"
'
%!!
10 FB-*,C1D,
Summary
Vsd=Vout
'*+,
ap gate-to-source
Cgs,ov and gate-to-drain
high Evaluated
energyfor 1photons,
the
output
voltage
is
expected
to
#!!E"
ph @ 1keV
Evaluated for 10 ph
!"&
ces: instrumentation developed for FEL ex- exceed
onic
the threshold thus10 showing
a maximum Cf which
$!!
1
!"
!"
#"
have to satisfy
severe
requirements
in is !"%
mainly F@AB1C:D1
given
by the gate-to-channel
Cgc capacitance 100
Channel
[µm]
#!!E' Length/Width
Gate Area
≈ Camplitude
= 2W ∆LC
(1)
gs,ov + Cgd,ov
ox rate, input
and
resolution,
frame
#!!
?@AB1C:D1
#
!"$
#!!E'
and frame
storage
capability.
Covering
device
channel
width,
∆L
the
extension
C
(2)
In the
low
energy
rangeis(E
<10
photons the
at 1 keV) the gain
f,max ≈ Cgc = W LCox .
$%&
3
In the
high energy range (E >10 photons
at 1 keV) the gain
"!!
00
photons
1iskeV
10 keV) input
egion
and Cis@
the to
gate
capaci-dyox
independent
of theoxide
MOS channel
length Therefore,
L !"#
in
the
high
energy
range
(>103 photons) the
hile preserving
single
photon
resolution
at
area.
Therefore,
in
the
low
energy
range
on the MOS gate area WL
can be adjusted with the MOS channel width
Wdepends
#
gain
! depends on the MOS gate! area W L and can be set
one
of
the
most
challenging
tasks.
In
order
the gain of the preamplifier is almost in!
'!!
#!!
(!! with
$!! the
!
#!!! length
%!!! '!!!
(!!! "!!!!
be adjusted
MOS
channel
L (with
set for
withcan
a suitable
design
of)!!the
MOS
channel
length
LW
(once
amic
range
into alength
reasonable
9: 1=1: 917>:8
:;6<5=,70>,?,",@619
e MOS
channel
L and output
can be signal
adlow energy gain setyllynon-linear
characteristic
is required.
Sig- W has beenst properly chosen for 3the
choosing
the
channel
width
W
.
For
/
11
G. Rizzo
PixFEL –2.Pisa
July,
2014compression
The 1simulated
input-output
trans-characteristic
of
Dynamic
signal
through the
non-linear features 19
n can be achieved at sensor level, as in the Fig.ting).
Front-end channel design (1) PV
gs
)*+,-./0./,123/456,7819
*++,-./012.3./45.6/+173*8 Low Energy Gain [mV/ph]
High Energy Gain [µV/ph]
B
4
&
%
$
#
"
!
!
;<5
46
#
%
'
(
"!
Gated integrator: Flip Capacitor Filter
Front-end channel design (2) PV
2
25'&%$*6$",4(3$*789:4;$&
<4#5*64%"':*2)89&$,,4)"*
01#)1!
2)"3$&,4)"
S3
Fig. 3. Schematic diagram of th
SR
S1
WF
CF
C f0
W/L
V DD
9W/L
gnd
1 keV / 10 keV
Vin
-
25'&%$*6$",4(3$*789:4;$&
<4#5*64%"':*2)89&$,,4)"*
01#)1!
2)"3$&,4)"
Fig. 3. Schematic
different gain, like
in the case of the LPD and the Percival tim
S0
detectors).
Th
Fil
S4 The schematic diagram of the full readout processor deS0
S1-S4
veloped in the frame of the PixFEL project is shown in Sam
S3
V in
S2
Fig. 3. ItVincludes,
beside the charge amplifier with signal gra
out
compression, the
shaping stage, the Sample&Hold capac- zoi
S2-S3
S1
+
itor and a 10-bit successive approximation register (SAR) duc
CF
ADC. Since the readout channel will be bump-bondend to out
0
100
200 t [ns]
a hole collecting pixel sensor, an NMOS has
been used as Th
ADC
01#)1!
Baseline
Signal
Signal
!"#$%&'()"*'"+*,-.#&'/()"*,#'%$
Conversion
Integration
Settiling
Integration
2)"3$&,4)"
the non-linear feedback capacitor of the and
charge
preampli- con
Reset
fier. Starting from the compression idea proposed above, str
a more complex feedback network has been worked out ara
S4
SH
S0
Only one amplifier
in which the MOS is split in two to switch from 1 keV the
to 10
operation
Thecapacitor
additional C
capacitance fou
Trapezoidal S2
weighting function achieved
bykeV
flipping
the mode.
feedback
F
b0
Cf 0 allows for precise low-energy gain settings,
while the filt
transconductor is used for10-bit
the purpose of trimming and and
resetting the MOS device. ADC
The forward stage of the am- seq
+
Vref
plifier is realized with a classical folded b9
cascode architec- gra
Csh
ture
with
a
PMOS
input
device
to
cope
with the bias re- col
5
L. Bombelli, C. Fiorini, S. Facchinetti, M. Porro, G. De Vita, “A fast current readout strategy for the XFEL
quirements
of
the
feedback
MOS.
Moreover,
an improved get
DePFET detector”, Nucl. Instr. Meth., vol. 624, pp. 360-366, 2010.
seq
output stage has been used to provide the high current
17 / 29
values required during the integration and reset phases. (th
The shaping stage is a standard component of an opti- tor
mum channel for charge signal processing. Since FEL fa- the
!"#$%&'()"**'"+*
6'89:$* events with
=>1.4#*
cilities generate
a known repetition rate, a and
,-.#&'/()"*,#'%$
A*B):+
67?*7@2
tim
rm
rat
("!$
diagram of the full readout channel.
Th
(
rea
Good preliminary results from simulation of the
*+*,-./0./,123/456,718
Vref
G
(,0C
!"#'
!"#&
ifferent gain, like in the case of the LPD and the Percival time-variant shaping stage has been adopted in this work.
full front-end channel with 50 ns integration
time:
etectors).
The proposed
architecture is based on the Flip Capacitor
Filter (FCF)
idea [4] which applies a Correlated Double
he schematic
diagram eof the
readout processor
de- photon
•  ENC=51
à full
S/N=5.5
for single
1 keV
a single inteeloped in the frame of the PixFEL project is shown in Sampling (CDS) technique to obtain, with
9:;6,7<=8
feedback capacitor, a trapeig. 3. It includes, beside the charge amplifier with signal grator stage and a flipped
Fig. 4. Simulated output channel voltage swing for 1 and 100 inlinear
transcocomingAn
1 keV additional
photons signals at 5 MHz
timing operation.
ompression, the shaping stage, the Sample&Hold capac- zoidal weighting function.
st
PixFEL
– Pisa July,
1 has
2014
20
ductor
been introduced to convert the voltage at the
or andG.
a Rizzo
10-bit successive approximation register
(SAR)
(!!,0C
!"#%
!"#$
!"#
[1]
>4=63:<6
?</65@4/:2<
!
A:5<43
A6//3:<5
)!
A:5<43
?</65@4/:2<
(!!
[2]
B6=6/
()!
$!!
[3]
[4]
In-pixel SAR - ADC
•  Digital data storage and off chip
data transmission preferable
w.r.t analog approach
•  Specs: 9 bit (guarantees single
photon resolution at small signal,
small quantization noise in
Poisson-limited regime), 5 MHz
sample rate (for XFEL)
ܸ௥௘௙
ܸ௥௘௙
ܸ௥௘௙
൅ ܾଶ
൅ ܾଷ
൅ ‫(ڮ‬
ʹ
Ͷ
ͳ͸
(
•  10 bit SAR (Successive Approximation( Register) ADC architecture chosen:
good compromise between clock frequency (à50 MHz) and resolution.
• 
ܸ௜௡ ൌ ܾଵ
f_clk=N*f_conversion in SAR (2^N in a ramp based ADC)
–  Critical points: area (DAC), power, clock distribution …
–  Present simulation results (PV):
•  ENOB =9.2
•  Power ~ 70 uW
–  Area ~ 5000 um2 from first layout
G. Rizzo
PixFEL – Pisa July, 1st 2014
21
In-pixel logic & readout (1) - Pisa
Selezione set comandi
•  Definito il readout semplificato della
matrice 8x8 con massima flessibilita’ per
studio performance + debug/studio di
cross talk!
due set #0 e #1 di segnali di comando, un bit di
configurazione dentro ogni pixel per scegliere quale set usare
hold,
conv_clk,
conv_start,
wr_buf1_sel,
rd_buf_sel
MASK
– 
SELECT
hold0,
conv_clk0,
conv_start0,
wr_buf1_sel0,
rd_buf_sel0
hold1,
conv_clk1,
conv_start1,
wr_buf1_sel1,
rd_buf_sel1
G. Rizzo - F. Morsani
conv_end1
PixFEL Meeting – 13 Marzo 2014
8
;<=3>-#-#?@'<<=%#A3@#B%03@#
G+#?%4#B.4#
E1#
G+%43#
E*/#
G+%43#
='@A#
<'.F"#<@'<H#
<'.F"#<'.4+'@# <'.F"#,4-+4#
#
@'5%<#
<'.F"#3.A#
J6L+35#
*D;P#
#
RST1EQ#
#
;D$#
#
CDE#
;121EQ#
#
#
#
+'G#,3@3<4#
G. Rizzo
PixFEL – Pisa July,
2014
!"#$%&&'#(#)"#*'+,-.%#
<'@#,3@3<4#
,="'M4#
,=#<@H#
,="%.#
G+L?MN6L3.-#
+AL?MNL,3@#
JK)
+3-A#3.-?@3#
1st
sh.in
bit di
configurazione
del pixel
conv_end0
#
•  Implementati i blocchi della logica nel pixel
ed effettuate prime simulazioni miste del
pixel singolo e di pixel in colonna.
•  In corso le simulazioni delle sequenze di
test “standard” e sequenze di test per
studio di cross talk
sh clk
INJECT
#
2 set di comandi indipendenti/pixel: per studio di
cross talk di un pixel sugli altri pixel
–  2 buffer/pixel: per studio di cross talk di un pixel
su se stesso in diverse fasi (conversione/reaodut)
–  Buf1 configurabile: registro con scrittura e
lettura parallela (nel test chip) oppure come shift
register caricabile parallelamente (con il
contenuto del SAR) e leggibile serialmente per
emulare lettura sequenziale veloce del chip finale
sh.out
B1_CFG
conv_end
#
DCE#A-4-I#
B%03@#O#
/%0)12#*334%.5##²#67#*-+&'#896:#
!"#
22
!"#$%&'(&)'*%(+&
In-pixel
logic & readout (2) Pisa
%&'()*+,-!"".$
."/&0$1+23$".+43.$
!"#$
%&'()!"#$
#5"$630$631$
."/&0$32!%13$
*!.!1131$
"!0!$%&6$
#/2'+7$.37$
F. Morsani
,-&./0#"!'&&1234536415&
PixFEL Meeting
– 19 Giugno 2014
7& 23
Massima flessibilita’ per il debug di possibili
cross talk sul chip di test
TRA PIXEL DIVERSI: sampling (event1) durante il readout di
event0 usando un pixel sniffatore
Hold0
(pixel generatore)
event 0
Hold1
event 1
(pixel sniffatore)
Write buf1 enable
spazzolata
non serve buf1, dati nei rispettivi SAR
Start Conv0
(generatore)
Start Conv1
(sniffatore)
Campion., digitizzazione
T_conv_evt0
storage 10 bit in pixel/evt,
conv_clk up to 50 MHz
Read buf1 sel
Readout sequenziale
della matrice, direct
addressing
G. Rizzo - F. Morsani
G. Rizzo
T_conv_evt1
si leggono i SAR
T_read_evt0
PixFEL Meeting – 13 Marzo 2014
PixFEL – Pisa July, 1st 2014
T_read_evt1
14
24
Studi preliminari sul readout/banda (Pisa)
•  Modularita’ del rivelatore per coprire 20x20 cm2.
•  Necessita’ di banda (su chip e ladder) per applicazione a diversi FEL &
sviluppo architettura operabile in burst mode/continua.
•  Ipotesi: chip finale da 64x64 pixel (100 um), ladder=sensor=2.56x5.12 cm2,
32 chip/ladder con 10 bit risoluzione . ~ 30 ladder per l’area totale.
• 
• 
• 
• 
• 
LCLS: 120 Hz frame rate continuo à
banda 5 Mb/s/chip e 160 Mb/s/ladder
XFEL: 4.5 MHz frame rate, 1% duty cycle,
1k frame/3k frame storati
à banda: 0.6 Gb/s/chip e 20 Gb/s/
ladder
In queste condizioni XFEL come macchina
continua con frame rate 15 kHz.
Sviluppi fatti al PSI per rivelatore EIGER
(moduli 3.8x7.7 cm2, 8 chip 256x256
pixels) arrivano a moduli con banda 50 Gb/
s/ladder.
Assumiamo che 50 Gb/s/ladder siano
raggiungibili à 37 kHz di frame rate in
continua dovrebbero essere raggiungibili.
G. Rizzo
Eiger, the next generation pixel detector
– Single photon counting pixel detector
– Sensitive area of 38 X 77 mm2
– Pixel size 75 Pm
– 524k pixel module
– Dead time free mode of operation
– Maximum frame rates
• 23 kHz in 4 bit mode
• 12 kHz in 8 bit mode
• 8 kHz in 12 bit mode
– 8 GB of memory on a module
– Two 10 GbE data links per module
!"#$%&'()*&
First full module
diffraction pattern
+,-.&/#01-)&(/,2)&
!"#$%&"$'%
Bernd Schmitt, SRI2013 9/19/2013
PixFEL Meeting – Pavia, 20-21 Feb 2014
Seite 30
25
Attivita’ a Pisa 2015
•  Q1
–  Preparazione schede di test per chip 8x8
–  Integrazione di scheda per DAQ nelle schede di test di PixFEL
•  Compact 1 Gbit Ethernet DAQ developed for PIXIRAD: similar needs for test on beam
lines.
•  M.Minuti interessato/disponibile ad apportare le modifiche necessarie all’integrazione
sulle nostre schede di test.
•  Q2-Q3
–  Caratterizzazione in lab del chip 8x8
•  Debug del nuovo sistema di test e del chip!
–  Contatti con le facilities FEL per definire specifiche architettura
readout e possibilita’ test su fascio.
–  Inizio progetto del prototipo 32x32
•  Logica in-pixel/architettura di readout
• 
Q4
–  Sottomissione prototipo 32x32
–  Inizio implementazione del DAQ per test prototipo 32x32 in lab e su
fascio
–  Inizio del progetto del chip digitale per prototipo 2 layers (3D)
G. Rizzo
PixFEL meeting – June, 24 2014
26
Richieste servizi di Sezione 2015
•  Sviluppi del chip di elettronica e schede
di test/DAQ
–  Morsani: 50%
–  Minuti: ~30%
•  Assemblaggi di chip (un chip nel 2015)
–  Supporto del servizio A.T., per incollaggi,
microsaldatura
G. Rizzo
PixFEL – Pisa July, 1st 2014
27
PixFEL - Personale a Pisa 2015
G. Rizzo
PixFEL meeting – June, 24 2014
28
Richieste Pisa 2015
G. Rizzo
PixFEL meeting – June, 5 2013
29
backup
G. Rizzo - F. Morsani
PixFEL Meeting – 20-21 Feb 2014
30
Storage rings and FELs
• 
• 
• 
• 
• 
Pulse length: 103 shorter (100s fs)
Emittance: 102 hor., 3 ver. lower
Intensity per pulse: 102 higher
Monochromaticity: 10 better
Peak brilliance: 109 higher
m < ! < 1pm
1 eV < E < 10 keV
Various FELs
program accessible at FELs is quite broad:
ral biology
LCLS @ SLAC
e structures of large macromolecular biological systems
XFEL @ DESY
try
erstand catalytic mechanisms (efficient conversion of light into
trical/chemical energy)
al science
y transport and storage of information on increasingly smaller
th- and faster time-scales.
and molecular science (AMO)
y the fundamental interactions among electrons and between
trons and nuclei; explore the frontiers of light-matter
ractions
ities have more than a single beam line
ffer by the photon beam energy
SwissFEL @ PSI
nt lines give access to different types of experiments
Giulia Casarosa
2
!
G."#$%&'!()!*+,'-./#+!01!*2#33"456!7,'!08'&.99!1.+#9#/:!9';$/,!#;+9%<#;$!/,'!'=>'&#-';/.9!,.993!#3!.?0%/!@AA-6!!
Rizzo
PixFEL – Pisa July, 1st 2014
32
M
!"#G#,*5,(F#"(1%=(<.>?3(F*&050'0%,
!"#$%&'(
)'*'+,
-%.
ʄ/01
2*01(301*&(
'%&41#5#67
89%"*55(
5%16'4
)!:;<
!"#$%&'
(")*(+
,-.'/"0
1'2
345#"!'67
,-81'*('1-9'/:;
1<<'=
)=0,,>?3
!"#$%&'
(")*(+
1-8'/"0
>'2
345#"!'67
1-9'/:;
9>1'=
!:3(<>?3
!"#$%&'$&'
)(*%("##
><'/"0
<-?'2
345#"!'67
,-81'*('1-9'/:;
@<<'=
)4*164*0(<>?3
!"#$%&'$&'
)(*%("##
?-.'/"0
>'2
345#"!'67
1-9'/:;
?<<'=
*)+$*&A5'
"B+"&#$*&'*C
DEF'G0'
H-1'/"0'5$&AI
J
J
345#"!'67
,-81''/:;
J
!"#$%&'
(")*(+
,-,1'/"0
>,'2
7-K-''L7
>-H'/:;
9<<'=
!"#$%&'$&'
)(*%("##
,-.'/"0
>,'2
7-K-''L7
>-H'*('>-1'/:;
J
!"#$%&'
(")*(+
,-H'/"0
>,-.'2
7-K-''L7
>-H'/:;
.1<'=
2:<(@A(>?3
B3)
BC3)
D?))E(,#F'(<."*7(>?3
PiXFEL
From LCLS-I to LCLS-II
!
LCLS-I: 120 flashes of light per second
!
addition of a second 1km linac (+ injector) operating independently to LCLS-I
!
‣
independent soft and hard x-ray SASE undulatory
‣
beam rate up to 1MHz in continuos wave
(from LCLS-II Key Parameters (DRAFT) Oct. 4, 2013)
expected to deliver first light in 2019
!
February 21st 2014
Key enhancements of a fully instrumented
LCLS-II over LCLS-I:
‣
extended photon energy range:
250 eV - 18 keV
‣
pulse enhancement of 1000x in brightness,
10x in power, 100x smaller bandwidth
‣
improved x-ray detectors: single photon
sensitivity, dynamic range, number and
size of pixels, 120 Hz for soft x-rays
Giulia Casarosa
34 !4
xFEL beamlines
facility from all the others, be they operational, under construction, or in the
planning stage.
100 ms
600 Ps
100 fs
60 W / 2 -
(global average)
10 kW a
(train-averaged)
20 GW
(peak power)
Figure 1: Pulse operation mode of the European XFEL and corresponding average
and peak power values
Eventually, the present concept emphasizes the use of an extremely versatile
secondary spectrometer in the hard X-ray domain, while maintaining the
simultaneous capability for revealing X-ray diffraction studies down to the few
femtosecond time scale. This versatility can be preserved for a rather simple
sample environment, thus the startup design will focus on dynamic studies in
liquid or physiological environments, but also permit the study of solid
samples, which require vacuum and cryogenic conditions. This concept is
different from the other five scientific instruments planned for startup at the
European XFEL, and thus provides specific capabilities that cannot be
(easily) followed at any other scientific instrument at the European XFEL
facility.
3keV < E(𝛾) < 25keV
G.th Rizzo
June 13 2013
SASE = Self-Amplified
Stimulated
PixFEL
– Pisa July,Emission
1st 2014
Giulia Casarosa
0.26keV < E(𝛾) < 3keV
35
3
xFEL instrumentation
2D pixel
HPAD
2D pixel
LPD
2D pixel
HPAD
2D pixel
CCD
2D pixels
detetor
2D pixel
DEPFET
G. Rizzo
June 13th 2013
PixFEL – Pisa July, 1st 2014
Giulia Casarosa
36
5
Protein imaging
Using extremely short and intense X-ray pulses to
capture images of objects such as proteins before the Xrays destroy the sample.
Single-molecule diffractive imaging with an X-ray freeelectron laser.
Individual biological molecules will be made to fall
through the X-ray beam, one at a time, and their
structural information recorded in the form of a
diffraction pattern.
The pulse will ultimately destroy each molecule, but not
before the pulse has diffracted from the undamaged
structure.
The patterns are combined to form an atomic-resolution
image of the molecule.
The speed record of 25 femtoseconds for flash imaging
was achieved.
Lawrence Livermore National Laboratory (LLNL)
Models indicate that atomic-resolution imaging can be
achieved with pulses shorter than 20 femtoseconds. M.Ferrario make a movie of chemical reactions
Chemical reactions often take place
incredibly quickly: orders of magnitude of
femtosecond are not rare. The atomic
changes that occur when molecules react
with one another take place in moments that
brief.
The XFEL X-ray laser flashes make it
possible to film these rapid processes with an
unprecedented level of quality.
Since the flash duration is less than 100
femtoseconds, images can be made in which
the movements of detail are not blurred.
xfel.desy.de
And thanks to the short wavelength, atomic
details become visible in the films.
To film a chemical reaction, one needs a series of pairs of X-ray laser flashes.
The first flash in each pair triggers the chemical reaction. With the second flash, a snapshot is then
made.
The delay between the two flashes can be precisely modified to within femtosecond and a series of
snapshots can be made at various times following the start of the reaction.
In each case, the images are of different molecules, but these images can be combined into a film.
M.Ferrario Science program
The science base accessible at FELs is quite broad
•  structural biology: study and solve the structures of large
macromolecular biological systems
•  chemistry: understand the mechanisms of catalytic processes
responsible for efficient conversion of light into electrical/
chemical energy
•  material science: study the mechanisms of transport and storage
of on increasingly smaller lengths and at faster time scales;
•  atomic and molecular
science: is concerned with
the study of fundamental
interactions among
electrons, between
electrons and nuclei and
between light and matter
,';#(6%&1<='>(6%
,>1==6<';#
%%?;=6;@6%A6B=C@6>C;D%
%%%%%%%%%%%%%%%%%%%%)"(@6@4
!
%%E89%!F%A<1B6%<1=64
!
%%,';#(6%)$C=C;%%
%%%%%%%%%%%%%%@6;@'='G'=H4%
!
!"#$%&$'('))
34%5674%899:
*+*,%-%&./%01(2
10
XFEL Instrumentation
Although each experiment at FELs may require a specific detection
system, two main scientific cases may be identified
•  energy sensitive detectors with Fano limited energy resolution for
spectroscopic experiments, possible position sensitivity for angular
dispersive experiments (0D or 1D)
•  silicon drift detectors
•  high-Z detectors
•  cryogenic detectors
•  area detectors for imaging
experiments, based on X-ray
diffraction (2D)
•  charge coupled devices
•  hybrid pixel detectors
•  monolithic active pixel
sensors
Progetto PixFEL
•  Nasce dall’idea di applicare i piu’ recenti sviluppi di rivelatori a
pixel per HEP per migliorare le performance dei sistemi di
rivelazione per i FEL.
•  Scopo finale di costruire un sistema utilizzabile dai gruppi di
ricerca nel campo della photon physics.
•  Nella fase iniziale (3 years) ci concentremo su un dimostratore
che verifichi gli elementi innovativi.
•  Fasi successive da realizzare anche attraverso Horizon2020
•  Tentativo di coinvolgere da subito le industrie per
l’ingegnerizzazione del sistema (FBK, CAEN, …)
•  Sviluppo per: XFEL, LCLS, SwissFEL,LCLS II(NGLS) … à
necessario sviluppare contatti con le comunità degli utenti dei FEL
• Long term activity plan
• In the long term (6+ years) the project aims at developing a 2D X-ray, pixellated
camera for applications at FELs, complying with the following characteristics
•  100 um pitch
•  1 kFrame (?) storage capability
•  reconfigurability (?), for operation in burst and continuous mode
•  burst operation at a maximum frequency no less than 5 MHz
•  continuous operation at no less than 10 kHz (?)
•  9 bits effective resolution
•  single photon detection capability
•  104 photon @1 keV dynamic range
To enable the integration of all the needed functionalities in a relatively small area,
a 65 nm CMOS technology, in conjunction with a vertical integration process
(including high density TSVs), will be adopted
The final instrument will be based on the tiling of elementary blocks with minimum
dead area to be achieved also by using low density TSV and active edge sensor
technologies
• 43
G. Rizzo
PixFEL – Pisa July, 1st 2014
44
European X-FEL experiments [1] and considered, among
others: a very wide dynamic range, from 1 to 104 photons at
fixed energy (in the range from ~1 keV to ~10 keV), single
photon resolution capability at low energies (up to about 100
photons), a pixel pitch of 100 µm, a frame rate of 5 MHz, and
tolerance to very high ionizing radiation doses (>10 MGy).
In orderedge
to allow pixel
for a multilayer
four-sideto
buttable module,
Active
sensors
we plan to use planar active-edge pixel sensors. First
minimize
gap Nanofabrication
between the
introduced at the
the Stanford
Facility (SNF) as
an
extension
of
3D
sensor
technology
[2],
active
edges have
active area and the edge of the
been later applied also to planar sensors [3]. Active edges
detector;
steps
consist of deepkey
trenches,
etched at the sensor periphery by
Deep Reactive Ion Etching (DRIE), and doped to act as wall
•  electrodes
trenchto etching
terminate the active area, with arbitrary shapes
Moreover,
full signal
sensitivity up to a few
•  possible.
trench
polisilicon
filling
micrometers from the physical edge can be obtained [4], [5].
•  These
support
wafer
advantages
come atremoval
the expense of an increased process
• n-­‐ substrate complexity, due to the DRIE step and the need for a support
wafer to hold different dice together once the trenches have
been etched [2]. Besides the original proponents at SNF,
active-edge technologies have also been developed at other
processing facilities, like SINTEF [7], VTT [8], and FBK [811], the latter being a partner in the PixFEL project.
Compared to pixel sensors for High Energy Physics
applications, where the current trend is towards thinner
substrates, for FEL applications relatively thick sensors are
optimization of the design by means of TCAD simulations. To
this purpose, we have simulated both the charge collection
properties and the breakdown voltage characteristics.
Enabling technologies for a 4-side buttable module
45"%#&
*+#,-.&
6#)($&
!"#$%&'$()#&
,3&
'3&
'3&
!"#$%&#$
,/&0120)+()#&
,3&
!"#!$%&'()*$&'+&$ȡ*$,-../$012.1%10$3345
G.-F. Dalla Betta is with TIFPA INFN Trento, and with Dipartimento di
Ingegneria Industriale, Università di Trento, Via Sommarive, 9, I-38123 Povo
di
Trento
(TN),
Italy
(telephone:
+39-0461-283904,
e-mail:
[email protected]).
Figure 1 Schematic cross-section of a planar active-edge sensor.
A schematic cross-section of the devices is shown in Fig. 1.
The sensors are p-on-n. The high-resistivity (n-) substrate is
450 µm thick and has <100> crystal orientation. N+ ohmic
contact regions are at the back-side and along the trench at the
edge. Metal field-plates are used (5 µm wide) on all pixels.
Depending
on the type of≥ simulations,
• Thickness
450 umdifferent numbers of
pixels and different edge terminations have been used. In
needed
good
efficiency
order to
investigate for
the plasma
effect,
simulations have been
carried@10
out considering
charge generated in the
keV the maximum
applications of interest (~3x107 e-h pairs), corresponding to
• Already
104 X-ray
photons of under
12 keV energy. To account for the
photondevelopment
absorption probability,by
this FBK
charge for
has been released
with an exponential distribution, using the optical generation
Alice and Atlas
Enabling technologies for a 4-side buttable module
Low density TSVs for chip to PCB bump-bonding
No gaps, no need for
complicate tiling
(provided that the
detector has minimum
dead area)
CEA-LETI
T-Micro
Aim of the PixFEL project in the first phase (3-years)
Investigating the enabling technologies for the design of chips with
minimum dead area and high functional densities
•  slim edge sensors
•  vertical integration for double tier design of the front-end
•  low density TSVs for chip interconnection to the hybrid board
•  interposers for sensor to front-end pitch adaptation
Studying, designing and testing the building blocks (CMOS 65 nm) for the
front-end electronics (100 um pitch), complying with the application
requirements
•  low noise, (reconfigurable) wide input range front-end channel (1 to 104)
with dynamic compression, single photon detection
•  9 bit (effective), 5 MS/s ADC (successive approximation register)
•  circuits for gain calibration
Looking into architectures for fast chip operation and readout
•  frame storage mode (memory cell, maximum memory size, readout)
•  continuous readout mode (maximum speed, accounting for DAQ
limitations)
•  reconfigurability (impact on the performance)
Work Packages
WP1: Enabling technologies (Lucio Pancheri, UNITN and INFN TN) – will
investigate the technologies with potential to enable the fabrication of
advanced 2D X-ray imagers to be used at FELs; the activity will mainly
focus on active edge pixel sensors and low density through silicon vias, as
the most important processes for the fabrication of a four-side buttable
chip.
WP2: Building blocks (Massimo Manghisoni, UNIBG and INFN PV) – will
address the design of the fundamental building blocks for the readout of
a pixel detector in a 2D X-ray imager to be operated at FELs, and
concentrate on the development of individual stages and with their
integration in a single tier 8×8 matrix at first, and with the design of a
32×32, single tier matrix to be interconnected to a fully depleted pixel
sensor in a later phase of the WP
WP3: Architectures and testing (Giuliana Rizzo, UNIPI and INFN PI) –
will deal with a two-phase task: study of the readout architecture for
the X-ray imager front-end chip; development of the hardware and
software test systems and the placement of the final characterization
of the detector in a beam line
Readout architectures
• No sparsification technique can be applied to imaging detectors à
a large amount of data needs to be read out in a relatively short
amount of time, also depending on the structure of the X-ray beam
• Burst mode operation: data need to be stored locally and read out
in the interval between two bursts; 65 nm CMOS technology is
supposed to guarantee enough density to exceed the state-of-theart of FEL instrumentation in terms of storage capacity
• Continuous operation: data are read out as soon as they are
collected, frame by frame; the relatively low repetition rate (~100
Hz) foreseen in FELs operated in continuous mode makes direct
readout of a megapixel detector a task within reach of present
technology
• The capability of switching from one mode of operation to the
other may be an important asset for a 2D imager
Pavia 2015
Activities and personnel
Activities
Characterization of the test structures submitted in the first year runs
Design of a 32×32 chip with 100 µm pitch (4.5 µm×4.5 µm)
Personnel
Name
Position
Commitment
Lodovico Ratti
Assistant Professor
50%
Massimo Manghisoni
Assistant Professor
30%
Full Professor
10%
Gianluca Traversi
Assistant Professor
20%
Daniele Comotti
PhD Student
70%
Research Fellow
20%
Piero Malcovati
Associate Professor
20%
Lorenzo Fabris
PhD Student and
Senior R&D Engineer, ORNL
40%
PhD Student
70%
Valerio Re
Marco Grassi
Luca Lodola
Total FTE number for Pavia
G. Rizzo
PixFEL – Pisa July, 1st 2014
3.3
2/3
50
Progetti TN gr.V 2015
Riassunto FTE 2015
Nome e Cognome
Ruolo
Percentuale
PA
30%
Roberto Mendicino
Dott.
50%
Lucio Pancheri
(responsabile locale)
RTD
50%
Hesong Xu
Dott.
50%
PO
30%
Gian-Franco Dalla Betta
Giovanni Verzellesi
Totale FTE
G. Rizzo
2,10
PixFEL – Pisa July, 1st 2014
51