Massimo Caccia, P.I.

Universita’ dell’Insubria [UINS]

Como (Italy)

&

INFN- Sezione di Milano

GSI, Darmstadt, July 22nd, 2011

The MIMOTERA: a monolithic pixel detector for real-time

beam imaging and profilometry[U.S. patent no. 7,582,875 ]

Featuring: Designers:

G. Deptuch [FERMIlab]W. Dulinski [IRES-Strasbourg]

DAQ: A. Czermak [INP-KraKow] I. Defilippis [SUPSI-Lugano]

Exploitation team:A. Bulgheroni [UINS/JRC Ispra] C. Cappellini [UINS] L. Negrini [UINS] M. Maspero[UINS] F. Toglia [UINS] F. Risigo [UINS] M. Jastrzab[AGH-Krakow] S. Martemyianov [ITEP/UINS]

Staff at HIT & CERN-AD M. Holzscheiter [MPI-Heidelberg & UNM Albuquerque, US] R. Boll [Heidelberg University]

Staff at LARN-Namur:A. C. Heuskin [LARN] A.C. Wera [LARN] S. Lucas [LARN]

Su

bar

ray

0 Su

bar

ray

1 Su

bar

ray

2 Su

bar

ray

3

17.136 × 17.136 mm2

digital

28 columns (30 clocks)

112 rows (114 clocks)

MimoTera

chip size: 17350×19607µm2

array 112×112 square pixels,

four sub-arrays of 28×112 pixels read out in parallel tread/integr<100µs (i.e. 10 000 frames/second)

Backthinned to the epi-layer (~ 15 µm ), back illuminated through an ~80 nm entrance window

AMS CUA 0.6 µm CMOS 15 µm epi,

no dead time

Essentials on the MIMOTERA (1/2):

Essentials on the MIMOTERA (2/2):

Charge To Voltage Factor = ~250nV/e- @ 500fF well capacity of ~ 36 MeV Noise ~1000 e- 280 e- kTC (ENC) @ 500fF

pixel 153×153 µm2 square pixels,

two 9×9 interdigited arrays (A and B) of n-well/p-epi collecting diodes (5×5 µm2) + two independent electronics – avoiding dead area,

In-pixel storage capacitors – choice ~0.5 pF or ~5 pF to cope with signal range (poly1 over tox capacitors),

153 m

Layout

of

one p

ixel

A B

Original push for the MIMOTERA development:minimally invasive real-time profilometry of hadrontherapy beams by secondary electron imaging[IEEE Trans.Nucl.Sci. 51, 133 (2004) and 52, 830 (2005))]

vacuum chamber

HV

secondary emission foil

electron detector

PROFILE/CURRENT MEASUREMENT

beam

Basic principle: collection and imaging of secondary 20 keV electrons emitted by sub- m thin Aluminum foils

The SLIM installed on an extraction line at the Ispra JRC-Cyclotron(p, 2H, 4H at energies 8-38 MeV, 100 nA- 100uA)

Secondary electrons emitted by a proton beam through a multi-pin hole collimator (Ø = 1mm, pitch = 1.5-6.5 mm)

Step 1 Step 2 Step 3

Today: results on beam imaging by DIRECT IMPACT on the sensor

Introductory slides on different imaging modalities

The analysis matrix

Imaging modality

Frame by frameDifferential Imaging

(pair of frames)

Analog pulse

a11 a12

Counting a21 a22meth

od

s

• is any of the matrix elements more robust and reliable?• any irreducible problem preventing to use any of them?

Imaging modalities: Advantages and disadvantages

Single matrix+ single polarity signal+ spot the detail of the

matrix (e.g. hot pixels)- strongly dependent on a

good pedestal calculation

- And even more on a reliable pedestal variation tracking (due to thermal effects, Charge Collection efficiency variations, else)

Differential imaging- both positive and

negative signals- If pile-up occurs, the

signal averages out to ZERO

- Matrix details can be masked off

+ NO NEED of pedestals+ self-tracking of the

pedestal variations

Methods: Advantages and Disadvantages

Analog Info- provides and effective

measurement of the intensity, requiring a [possibly time dependent] calibration

- Possibly affected by Pixel-to-pixel gain variations (requires a normalization, implemented by measuring the signal in sigma units)

+ it works also when pile-up occurs

IDEAL for real-time,

qualitative, imaging More difficult to get a

quantitative measurement

Counting+ provides a direct measurement

of the flux [provided you can spot in an unbiased way the hit pixels…]

+ in principle, no calibration required

+ not expected to depend on local gain variations

- HARD to use it as pile-up approaches [look at the non-fired pixels..]

RELIABLE & ROBUST for a quantitative approach

as the pile-up probability increases, play with the integration time…

Assisting the AD-4 [ACE] collaboration [ACE, http://www.phys.au.dk/~hknudsen/introduction.html]

Michael Holzscheiter, ACE spokesperson (left), retrieves an experimental sample after irradiation with antiprotons, while Niels Bassler (centre) and Helge Knudsen from the University of Aarhus look on [courtesy of ACE]

[courtesy of ACE]

“Cancer therapy is about collateral damage” compared to a proton beam, an antiproton beam causes four times less cell death in the healthy tissue for the same amount of cell deactivation in the cancer.

Shot-by-shot beam recording at the CERN anti-proton decelerator tests

beam characteristics:-120 MeV energy- 3x107 particles/spill- 1 spill every 90”- FWHM ~ 8 mm

acquisition modality: - triggered imaging modality: - differential radiation damage: - irrelevant so far

[max no. of spills on a detector: 1436]

data taking runs: - September 2009 - June 2010 - October 2010

Single shot picture

Generally speaking, the beam is stable… unless it moves!

Antiproton.wmv

Quantitative information:• profiling, compared to a GafChromic film • Intensity, compared to a calibrated ionization chamber (UNIDOS, by PTW)

Profiling the beam[PRELIMINARY RESULTS:• FWHM calculation checked, • errors on the GAF still being evaluated]

With a GafChromic Film, integrating the spills over a full

run…and PROJECTING

With the MIMO, overlaying the 120 events in run #37 events to mimic the Gaf

FWHM = 7.64 ± 0.05 mm FWHM = 7.11 ± 0.05 mm

More on the analysis..Step 1: Preparation of the reference beam image.

Find a proper observable, possibly independent from quarter-to-quarter gain variation:

A plot of the mean values of every pixel, averaged over all of the spills in the run

Step 2: check the x & y projections

2 remarks:

projections look pretty smooth, implying the normalized quantity is not so bad

deviations from the gaussian are significant…(qualitative and quantitative analysis)

Step 3: move from projections to slices:

X slicesY slices

FWHMx = f(y) [and the other way round)

i.e.

being F & G the marginal distributions

Step 4: estimate the “width” as the mean FWHM over the central 9 slices

Vs.

Monitoring the intensity fluctuations

MIMO Vs

UNIDOS*

*The PTW UNIDOS is a high performance secondary standard and reference class dosemeter / electrometer

HIT [Heidelberg Ion-Beam Therapy Center]: Quality control of pencil Carbon Ions & proton beamshttp://www.klinikum.uni-heidelberg.de/index.php?id=113005&L=en

The f

aci

lity b

uild

ing

The accelerator complex[patients treated since 2008]

The b

eam

para

mete

rs

Interested in high granularity (in time & space) because they do feature “intensity-controlled raster-scan technique”

beam time characteristics:- duty cycle 50%- spill duration 5 s- FWHM ~f(particle, intensity, energy)

acquisition modality: - free run imaging modality:

radiation damage: - relevant but not dramatic

[Total exposure time so far ~ 3h; about 1’-2’ per run at a specified nrj,

intensity]

data taking runs: - May 2010 - October 2010

Data taking conditions & qualitative information

€

Signal − Pedestal[ ]i

∑i

, i∈ROI

I = 7x107 particles/s, C ionsTime development of the beam

Quantitative information (C ions)

Intensity Scan

Energy Scan

Playing back the data

Carbon5.wmv

Imaging the LARN Tandem beams at Namur (B)

Main interests:

- The MIMO as a real-time, high granularity “digital” alumina screen, to optimize the set-up

- QC of the beam in terms of homogeneity

- quick measurement of the absolute intensity (particle counting!)

beam time characteristics:- continuous beams!- any ion (!) with an energy in MeV/amu range- intensities [103;108] p/cm2/s range

acquisition modality: - free run; MIMOin vacuum

radiation damage: - may really be dramatic![Total exposure time so far ~ 60h; p, , C ion beams]

imaging modality:- standard: signal - pede- differential: (i,j,n) = signal (i,j,n) - signal(i,j,n-1)- based on < 2(i,j,N) >- digital with a pixel dependent threshold

data taking runs: - July 2008, April 2009 + series of short runs since April 2010 performed by the people at LARN - next extensive run planned for end of May 2011

Data taking conditions & qualitative information

Beam profile with different imaging modalities[protons, E = 1.2 MeV, I = 6 x 105 p/cm2/s]

Signal - pede: + easy & classic - critical wrt pedestal variations

Delta: + robust against pedestal variations - failing as pile-up is approaching

Delta2 : + applicable also for high intensities - emphasizes local differences

Counting with pixel dependent thresholds:+ quite robust against detector degradation- failing as pile-up approaches

Real-time profiling

Image of a tilted beam

Flat beam

2 frames

10 frames

20 frames

Preliminary measurements on the intensity

Proton beam, 5 Gy/m = 2 x 106 p/cm2/s

Still working on the absolute particle counting, since it is a tough business…

Intensity by counting, with a pixel dependent threshold based on a common pre-set spurious hit level

DATA SETA page from my logbook

-Scan runs 1,2,4,5• #3 excluded because of a significant beam instability• #6 & 7 excluded here because they require an overlap of several events to achieve the required sensitivity+ ion source flickering

- rely on the unfired pixels to reduce the pile-up effects

Linearity, up to 8.8 x 106 particles/cm2/s

• protons, 1.2 MeV energy; • MIMO clocked at 2.5 MHz• differential mode• pixel dependent threshold at 5% spurious hit level

-p0 ≠ 0, we are still cutting off a few pixels which were actually hit- as usual, a trade off between sensitivity and purity has to be searched for..- clocking at 25 MHz, we can use the MIMO in counting mode till ~ 108 particles/cm2/s

Conclusions

The MIMOTERA appears to be an interesting device for RT beam profilometry at high granularity in space & time

radiation hardness is a certainly a concern, especially at Tandem beams [but so far we did not manage to kill a single detector…]

more tests are on the way, to quantify its lifetime in beam hours

other facilities showed some interests (Krakow, ITEP Moscow) and tests are planned