3D Trenched-Electrode Sensors for Charged Particle Tracking and - - PowerPoint PPT Presentation

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

3D Trenched-Electrode Sensors for Charged Particle Tracking and Timing

• R. Mendicino1,2, M. Boscardin3,2, G. T. Forcolin1,2, A. Lai4,
• A. Loi4,5, S. Ronchin3,2, S. Vecchi6, G.-F. Dalla Betta1,2

1 University of Trento, Trento, Italy 2 TIFPA INFN, Trento, Italy 3 Fondazione Bruno Kessler, Trento, Italy 4 INFN Sezione di Cagliari, Cagliari, Italy 5 University of Cagliari, Cagliari, Italy 6 INFN Sezione di Ferrara, Ferrara, Italy

Argonne, 11-14 September 2018

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Outline

• Introduction: 3D Sensors
• Timing with 3D-trench sensors
• Design and technological aspects
• TCAD simulation results
• Process development and sensor layout
• Conclusions

2

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

3D sensors

DISADVANTAGES:

• Non uniform spatial response

(electrodes and low field regions)

• Higher capacitance with respect

to planar (~3-5x for ~ 200 µm thickness)

• Complicated technology (cost, yield)
• S. Parker et. Al. NIMA 395 (1997) 328

Electrode distance (L) and active substrate thickness (D) are decoupled à L<<D by layout

HIGH RADIATION HARDNESS ADVANTAGES:

• Low depletion voltage (low power diss.)
• Short charge collection distance:
• Fast response
• Less trapping probability after irr.
• Lateral drift à cell “shielding” effect:
• Lower charge sharing
• Low sensitivity to magnetic field
• Active edges

3

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

4

1. shorter collection distance 2. higher average fields for any given maximum field (price: larger electrode capacitance) 3. 3D signals are concentrated in time as the track arrives

• 4. Landau fluctuations (delta ray

ionization) arrive nearly simultaneously

• 1. 3D lateral cell size can be smaller than wafer

thickness, so

• 2. in 3D, field lines end on electrodes of larger

area, so

• 3. most of the signal is induced when the

charge is close to the electrode, so planar signals are spread out in time as the charge arrives, whereas

• 4. Landau fluctuations along track arrive

sequentially and may cause secondary peaks

Timing: Planar vs 3D

4." 4." 4."

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Speed with 3D

So far tested with hex-cell 3D’s (L=50µm) & fast current amplifier

• S. Parker et al. IEEE

TNS 58(2) (2011) 404

5

10 100 2 3 4 5 6 7 8 9 10 20

dt - 50% constant fraction dt - fit mean - fit

time (ps) pulse height (mV)

waveform analysis

31 ps 177 ps snoise ~ 0.33 mV

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

6

Schematic diagram of multiple plane arrangement in an active-edge 3D trench-electrode detector. Other offsets (⅓, ⅔, 0, ⅓, ⅔ ..etc.) may also be used. next section offset so signal electrodes do not line up signal electrodes with contact pads to readout

beam in 200 – 300 µm active edge

3D Trenched-electrode sensors

Benefits from trench electrodes

• High average field / peak field
• Uniform Ramo’s weighting field
• Initial pulse time independent
• f the track position

Possible issues

• Fabrication complexity
• Electrode capacitance
• S. Parker et al. IEEE TNS 58(2) (2011) 404

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

3D technology: where we stand

!

• Very impressive progress made in the past few years
• High aspect ratio (depth/width) DRIE feasible:

up to 30:1 and beyond for columnar electrodes, even better for trenches ( > 40:1)

• Flexibility in the choice of active thickness

and inter-electrode distance (small pitch feasible)

• High breakdown voltage can be achieved also

before irradiation, allowing for large Vbias range

• S. Ronchin,

FBK

>150 µm 5 µm

7

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

8

p+ trench

(a) Strip (b) Pixel

n+ trench p+ trench n+ dashed trenches pitch pitch-y pitch-x width gap

Schematic design

• Pixel pitch (55 µm) chosen to be compatible with TIMEPIX ROC
• Trenches are dead volume, they should be as narrow as possible (~4 µm)
• n+ trench width (gap) to be optimized with the aid of TCAD simulations

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

9

Impact of the gap (1): electric field

Plots show the difference with respect to the strip (no gap) situation

strip (no gap) gap=16 µm gap=8 µm gap=12 µm gap=10 µm gap=14 µm

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

10

Impact of the gap (2): electron velocity

Plots show the difference with respect to the strip (no gap) situation

strip (no gap) gap=16 µm gap=8 µm gap=12 µm gap=10 µm gap=14 µm

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

11

Impact of the gap (3): hole velocity

Plots show the difference with respect to the strip (no gap) situation

strip (no gap) gap=16 µm gap=8 µm gap=12 µm gap=10 µm gap=14 µm

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

• As expected, the weighting field distribution is confined within neighbors pixel
• The gaps cause a significant distortion with respect to the uniform case (strip)

Impact of the gap (4): weighting field

gap=16 µm gap=14 µm gap=12 µm gap=10 µm gap=8 µm gap=6 µm 12

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta High dependence with geometry Low dependence with geometry

Weighting field: more in detail

gap=16 µm gap=14 µm gap=12 µm gap=10 µm gap=8 µm gap=6 µm 13

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

14

Impact of the gap (5): MIP signals

x x x

3 Trench widths: 1) 41 µm à d= 14 µm 2) 45 µm à d= 10 µm 3) 49 µm à d= 6 µm 3 Hit positions (X): 1) x =13.75 µm, y = 12.5 µm 2) x = 27.5 µm, y= 12.5 µm 3) x = 27.5 µm, y = 0 µm d Using TCAD HeavyIon model

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

15

Signal comparison

1 2 3 4 5

9

• 10

´ Time (s) 2 4 6 8 10 12 14

6

• 10

´ Current (A)

m µ 41 m µ 45 m µ 49 100V: x=13.75 y=12.5

1 2 3 4 5

9

• 10

´ Time (s) 2 4 6 8 10 12 14

6

• 10

´ Current (A)

m µ 41 m µ 45 m µ 49 100V: x=27.5 y=0

1 2 3 4 5

9

• 10

´ Time (s) 2 4 6 8 10 12 14

6

• 10

´ Current (A)

m µ 41 m µ 45 m µ 49 100V: x=27.5 y=12.5

At 100 V

x

d

x

d

At 100 V At 100 V

x

d

1 2 3

1 2 3 4 5

9

• 10

´ Time (s) 2 4 6 8 10 12 14

6

• 10

´ Current (A)

m x=13.75 y=12.5 µ 41 m x=27.5 y=0 µ 41 m x=27.5 y=12.5 µ 41 m x=13.75 y=12.5 µ 45 m x=27.5 y=0 µ 45 m x=27.5 y=12.5 µ 45 m x=13.75 y=12.5 µ 49 m x=27.5 y=0 µ 49 m x=27.5 y=12.5 µ 49

100V

At 100 V 1 2 3

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

16

Radiation hardness

ü ⃗ " is the drift velocity of charge carriers ü (Electric field à charge trajectory and velocity) ü #$ is the “weighting Field” à charge motion coupling to a specific electrode ü Charge trapping included:

%& = −) ⃗ " * #$ + , = +-./ 0 , 1

• Ramo’s Theorem with from input from TCAD:
• New Perugia model (bulk damage): F. Moscatelli et al., IEEE TNS 64 (2017) 2259
• Very high signal efficiency expected

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

17

Pixel: technological constraints

Device layer Support wafer

junction Ohmic

p-spray Si-Si Direct Wafer Bonding Minimum distance between trenches 4 4 20 4 Minimum distance between metals … Existing FBK 3D-SS process

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

18

Pixel: from schematic to layout

Contact hole

• hmic

junction

• hmic

Contact hole

• Modified process with additional oxide layer between poly and metal
• More design flexibility: no need for full metal overlap to poly
• Ohmic columns are better isolated
• Removal of p-poly-cap would increase distance to bump pad

p-poly-cap p-poly-cap Bump pad Bump pad

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Metal ( N-contact) Oxide N

• T

r e n c h e s + P

• l

y

• S

i P-Trenches + poly-Si Metal (P-contact ) 55 µm 1 5 µ m Bump pad

Full-3D simulation of capacitance

19

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

20

Considered geometries

Thin metal Trench Length: 41 µm 43 µm 45 µm 47 µm 49 µm Thin metal poly Trench Length: 41 µm 43 µm 45 µm 47 µm 49 µm Wide metal Trench Length: 41 µm 43 µm 45 µm 47 µm Bump Trench Length: 45 µm

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

21

Simulated capacitance (1): components

100

• 80
• 60
• 40
• 20
• Voltage (V)

0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

12

• 10

´ Capacitance (F)

m g1-g2 µ 41 m g1-back µ 41 m g1-g2 µ 43 m g1-back µ 43 m g1-g2 µ 45 m g1-back µ 45 m g1-g2 µ 47 m g1-back µ 47 m g1-g2 µ 49 m g1-back µ 49

Thin Metal poly, C-V

Interpixel (1x) Backplane (1x)

• Strong dependance of inter-pixel capacitance on trench dimension
• Minor dependence of back-plane capacitance on trench dimension

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

22

Simulated capacitance (2): interpixel fit

40 42 44 46 48 50 m) µ Trench Length ( 15 20 25 30 35 40 45

15

• 10

´ Capacitance (F)

Simulated Capacitance Calculated Capacitance

Capacitance vs Trench length

Thin metal poly

! = #$#%& ' ( )*++

Interpixel capacitance fitted with simple parallel-plate capacitor model, with gap d

)*++(-./) = 1790567

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

23

Simulated capacitance (3): total cap.

100

• 80
• 60
• 40
• 20
• Voltage (V)

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

12

• 10

´ Capacitance (F)

m µ 41 m µ 43 m µ 45 m µ 47

Wide Metal, C-V

100

• 80
• 60
• 40
• 20
• Voltage (V)

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

12

• 10

´ Capacitance (F)

m µ 41 m µ 43 m µ 45 m µ 47 m µ 49 Thin Metal, C-V

100

• 80
• 60
• 40
• 20
• Voltage (V)

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

12

• 10

´ Capacitance (F)

m µ 41 m µ 43 m µ 45 m µ 47 m µ 49

Thin Metal poly, C-V

100

• 80
• 60
• 40
• 20
• Voltage (V)

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

12

• 10

´ Capacitance (F)

m µ 45

Bump on trench, C-V

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

24

Simulated capacitance (4): comparison

100

• 80
• 60
• 40
• 20
• Voltage (V)

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

12

• 10

´ Capacitance (F)

Wide Metal Thin Metal Thin Metal poly Bump on trench

m trenches, C-V µ 45

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

25

Technological tests at FBK (1)

• Pixel size 55x55 µm2

à lithography is critical à use stepper (min. size 350 nm, align. accuracy 80 nm)

• Test reticle specially designed at UniTN
• variable shapes, sizes and overlaps
• Technological tests performed to:
• Assess and optimize the key process

steps for trenched electrodes:

• litho and etching of trenches
• litho of poly
• litho of metal
• Define layout rules for sensor design

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

26

Technological tests at FBK (2)

Lithographical problems with standard (3D) resist thickness (10 µm) After exposure After hard bake

• Change of shapes & dimensions
• Trench interruptions

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

27

Technological tests at FBK (3)

Much better results with thinner resist (6 µm) and vacuum bake (VB) 130.04µm 143.42µm 5.18µm 4.35µm 4.18µm 5.52µm After exposure After VB & DRIE

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

28

TIMESPOT1 Layout (1): reticle

Test str. Timepix Using full-size stepper reticle for two blocks:

• Timepix sensor
• Test structures

Wafer level floorplan:

• Pixel array (center)
• Test str. (periphery)

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

29

TIMESPOT1 Layout (2): Timepix

Temporary metal 256x256 pixels, 55 µm pitch

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

30

TIMESPOT1 Layout (3): Test Structures

Single pixels Multi-pixel strips Technological monitors:

• MOS capacitors
• VdP resistors
• Gated Diodes

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Conclusions

• We have started the development of 3D-trenched electrode

sensors for a new tracking system with high space/time resolution

• The latest FBK 3D-SS process has been modified and tuned for

trenched electrodes

• TCAD simulations allowed to investigate the impact of the design

details on the electrical characteristics of the sensors and signal waveforms

• The simulation approach is being extended to account for more

realistic distribution of charge generation profiles (Geant4) and larger statistics with reduced simulation times

• A first layout, including TIMEPIX compatible pixel sensors and

many test structures has been submitted for fabrication to FBK

31

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Acknowledgements

• This work has been partially funded by:
• the 5th Scientific Commission of the Italian National Institute for

Nuclear Physics (INFN) through the Project TIMESPOT (CSN5)

• INFN and FBK through the Framework Project MEMS4
• Special thanks to:
• Sherwood Parker for inspiring this work
• Cinzia Da Via for fruitful discussions

32

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Back Up Slides

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Poly-Si electrode inefficiency

• J. Hasi, PhD thesis, Brunel, 2004

Electrode response using 12 keV X-ray beam (ALS at LBNL), beam size ~ 2μm

• Diffusion, lifetimes (poly-Si grain sizes)
• Oxide barrier effect at the interfaces …

à Replace POCl3 with PH3 à Replace BBr3/O2 with B2H6

34

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

3D Radiation hardness: Signal Efficiency

Compilation by C. Da Via [6.41] ATLAS IBL Collaboration, JINST 7 (2012) P11010 [6.42] G.-F. Dalla Betta, et al., NIMA 765 (2014) 155 [6.44] M. Fernandez et al. NIMA 732 (2013) 137 [6.43] I. Haughton et al., NIMA 806 (2016) 425 [6.52] G.-F. Dalla Betta, et al., IEEE NSS (2015) N3C3-5

Signal Efficiency = Ratio of max. signal after irradiation and before irradiation

F + =

D

v K L SE

t

6 . 1 1

• C. Da Via, S. Watts,

NIMA 603 (2009) 319

35

L~35 µm L~ 56 - 71 µm

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Null field points and delayed signals

• S. Parker et al.

NIMA395 (1997) 328 Efield Efield

50V 5V 50V 5V

Current pulses for particle hit at high field point Current pulses for particle hit at null field point

• 3D structure can

potentially yield very fast signals of the

• rder of 1 ns
• But electric field is

not uniform, and null field points are present: signals are delayed due to initial diffusion

• Moreover,

electrodes are (almost) dead regions

• These aspects can

be improved with dedicated designs

10V 10V

36

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Capacitance and noise

• E. Alagoz et al. JINST 7 (2012) P08023
• C. Da Via et al. NIMA 604 (2009) 505

JINST 7 (2012) P11010

CMS PSI46 FE-I3 + SNF 3D FE-I4 (2E)

37 G.-F. Dalla Betta JINST 10 (2015) C07010

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

38

Time resolution …

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

39

Earlier Trenched-electrode simulations

!"# $"# !"# $"#

%&# '&# (&# )&# *&# +&#

,-*(./0(## 10*-)## 23.*340.5# 678(9&#

~30 ps with CFD (INFN TO)

500 ps

Simulated Efield at Vbias = 100 V

IP close to the cell border IP close to the read-out electrode

N-on-p N-on-n Other possible layouts

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

40

Trenched-electrode pixels at BNL-CNM

• Designed and simulated by BNL and Stony Brook University
• 1st batch fabricated at CNM Barcelona in 2013
• It worked, but with high leakage currents
• Charge collection tests performed on large pixels only
• 2nd batch announced but did not follow
• A. Montalbano et. al.

NIMA 765 (2014), 23

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

41

Trenches at FBK

• Full trenches for active edge
• “Staggered” dashed trenches for slim edge

(5 µm width and 40 µm length “small” trenches)

• Trenches ~20 µm deeper than Active Thickness
• Trench partially filled with un-doped poly
• Trench doped via gas sources (POCl3 and BBr3)
• Etching rate depends on area !

Project INFN-FBK "RD_FASE2"

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

42

Effect of signal processing

• Sensor + CSA + Leading edge discriminator
• Preliminary simulation in 65 nm CMOS
• TCAD signals for 5 hit positions (1.2 fC)
• Timing uncertainty at the output of the

discriminator ~34 ps

• Promising result, but other effects should

be modeled more extensively …

Particle tracks

• L. Piccolo, INFN Torino

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

43

Impact of capacitance

• Design ported to 28 nm CMOS
• Jitter estimated based on noise and simulated signal slope
• Strong impact of capacitance, that should be accurately simulated
• L. Piccolo, INFN Torino

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

44

TCAD signal simulations

• TCAD signal simulations account for all technology and design

details, yielding very accurate results, but charge release profile with HeavyIon Model is uniform à no information about real track

• r charge deposit from secondary particles (delta rays)
• A combined Geant4 + TCAD approach was developed
• But there are still problems with re-meshing multiple particle tracks, and
• 3D transient simulations are very time consuming (from ~hours to ~days)

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

45

A self-standing signal simulator

! = −$% & ∗ ()

• A different approach is being developed, based on Ramo’s theorem:

TCAD maps

• Initial results encouraging (and fast), physical model refinement (e.g., diffusion) and

porting to GPU multithreading, using the Nvidia Toolkit, are under way

ULITIMA, 12 Sept. 2018 G.-F. Dalla Betta

Radiation damage model

3-trap level “Perugia” Bulk model: M. Petasecca et al., IEEE TNS 53-5 (2006) 2971 “NEW”: parameters from F. Moscatelli et al., IEEE TNS 64-8 (2017) 2259 Defect E (eV) se (cm2) sh (cm2) h (cm-1) Acceptor Ec-0.42 1.0x10-15 1.0x10-14 1.6 Acceptor (f ≤ 7x1015 cm-2) Ec-0.46 7.0x10-15 7.0x10-14 0.9 Acceptor (7x1015 cm-2 ≤ f ≤ 1.6x1016 cm-2) Ec-0.46 3.0x10-15 3.0x10-14 0.9 Acceptor (1.6x1016 cm-2 ≤ f ≤ 2.2x1016 cm-2) Ec-0.46 1.5x10-15 1.5x10-14 0.9 Donor Ev+0.36 3.23x10-13 3.23x10-14 0.9 “OLD”: parameters as in D. Pennicard et al., NIMA 592 (2008) 16 Defect E (eV) se (cm2) sh (cm2) h (cm-1) Acceptor Ec-0.42 9.5x10-15 9.5x10-14 1.6 Acceptor Ec-0.46 5.0x10-15 5.0x10-14 0.9 Donor Ev+0.36 3.23x10-13 3.23x10-14 0.9